User Manual

Contents

User Manual#

Introduction#

rocSPARSE is a library that contains basic linear algebra subroutines for sparse matrices and vectors written in HiP for GPU devices. It is designed to be used from C and C++ code. The functionality of rocSPARSE is organized in the following categories:

The code is open and hosted here: ROCmSoftwarePlatform/rocSPARSE

Building and Installing#

Prerequisites#

rocSPARSE requires a ROCm enabled platform, more information here.

Installing pre-built packages#

rocSPARSE can be installed from AMD ROCm repository. For detailed instructions on how to set up ROCm on different platforms, see the AMD ROCm Platform Installation Guide for Linux.

rocSPARSE can be installed on e.g. Ubuntu using

$ sudo apt-get update
$ sudo apt-get install rocsparse

Once installed, rocSPARSE can be used just like any other library with a C API. The header file will need to be included in the user code in order to make calls into rocSPARSE, and the rocSPARSE shared library will become link-time and run-time dependent for the user application.

Building rocSPARSE from source#

Building from source is not necessary, as rocSPARSE can be used after installing the pre-built packages as described above. If desired, the following instructions can be used to build rocSPARSE from source. Furthermore, the following compile-time dependencies must be met

Download rocSPARSE#

The rocSPARSE source code is available at the rocSPARSE GitHub page. Download the master branch using:

$ git clone -b master https://github.com/ROCmSoftwarePlatform/rocSPARSE.git
$ cd rocSPARSE

Below are steps to build different packages of the library, including dependencies and clients. It is recommended to install rocSPARSE using the install.sh script.

Using install.sh to build rocSPARSE with dependencies#

The following table lists common uses of install.sh to build dependencies + library.

Command

Description

./install.sh -h

Print help information.

./install.sh -d

Build dependencies and library in your local directory. The -d flag only needs to be used once. For subsequent invocations of install.sh it is not necessary to rebuild the dependencies.

./install.sh

Build library in your local directory. It is assumed dependencies are available.

./install.sh -i

Build library, then build and install rocSPARSE package in /opt/rocm/rocsparse. You will be prompted for sudo access. This will install for all users.

Using install.sh to build rocSPARSE with dependencies and clients#

The client contains example code, unit tests and benchmarks. Common uses of install.sh to build them are listed in the table below.

Command

Description

./install.sh -h

Print help information.

./install.sh -dc

Build dependencies, library and client in your local directory. The -d flag only needs to be used once. For subsequent invocations of install.sh it is not necessary to rebuild the dependencies.

./install.sh -c

Build library and client in your local directory. It is assumed dependencies are available.

./install.sh -idc

Build library, dependencies and client, then build and install rocSPARSE package in /opt/rocm/rocsparse. You will be prompted for sudo access. This will install for all users.

./install.sh -ic

Build library and client, then build and install rocSPARSE package in opt/rocm/rocsparse. You will be prompted for sudo access. This will install for all users.

Using individual commands to build rocSPARSE#

CMake 3.5 or later is required in order to build rocSPARSE. The rocSPARSE library contains both, host and device code, therefore the HIP compiler must be specified during cmake configuration process.

rocSPARSE can be built using the following commands:

# Create and change to build directory
$ mkdir -p build/release ; cd build/release

# Default install path is /opt/rocm, use -DCMAKE_INSTALL_PREFIX=<path> to adjust it
$ CXX=/opt/rocm/bin/hipcc cmake ../..

# Compile rocSPARSE library
$ make -j$(nproc)

# Install rocSPARSE to /opt/rocm
$ make install

GoogleTest is required in order to build rocSPARSE clients.

rocSPARSE with dependencies and clients can be built using the following commands:

# Install googletest
$ mkdir -p build/release/deps ; cd build/release/deps
$ cmake ../../../deps
$ make -j$(nproc) install

# Change to build directory
$ cd ..

# Default install path is /opt/rocm, use -DCMAKE_INSTALL_PREFIX=<path> to adjust it
$ CXX=/opt/rocm/bin/hipcc cmake ../.. -DBUILD_CLIENTS_TESTS=ON \
                                      -DBUILD_CLIENTS_BENCHMARKS=ON \
                                      -DBUILD_CLIENTS_SAMPLES=ON

# Compile rocSPARSE library
$ make -j$(nproc)

# Install rocSPARSE to /opt/rocm
$ make install

Common build problems#

  1. Issue: Could not find a package configuration file provided by “ROCM” with any of the following names: ROCMConfig.cmake, rocm-config.cmake

    Solution: Install ROCm cmake modules

Simple Test#

You can test the installation by running one of the rocSPARSE examples, after successfully compiling the library with clients.

# Navigate to clients binary directory
$ cd rocSPARSE/build/release/clients/staging

# Execute rocSPARSE example
$ ./example_csrmv 1000

Supported Targets#

Currently, rocSPARSE is supported under the following operating systems

To compile and run rocSPARSE, AMD ROCm Platform is required.

The following HIP capable devices are currently supported

  • gfx803 (e.g. Fiji)

  • gfx900 (e.g. Vega10, MI25)

  • gfx906 (e.g. Vega20, MI50, MI60)

  • gfx908

Device and Stream Management#

hipSetDevice() and hipGetDevice() are HIP device management APIs. They are NOT part of the rocSPARSE API.

Asynchronous Execution#

All rocSPARSE library functions, unless otherwise stated, are non blocking and executed asynchronously with respect to the host. They may return before the actual computation has finished. To force synchronization, hipDeviceSynchronize() or hipStreamSynchronize() can be used. This will ensure that all previously executed rocSPARSE functions on the device / this particular stream have completed.

HIP Device Management#

Before a HIP kernel invocation, users need to call hipSetDevice() to set a device, e.g. device 1. If users do not explicitly call it, the system by default sets it as device 0. Unless users explicitly call hipSetDevice() to set to another device, their HIP kernels are always launched on device 0.

The above is a HIP (and CUDA) device management approach and has nothing to do with rocSPARSE. rocSPARSE honors the approach above and assumes users have already set the device before a rocSPARSE routine call.

Once users set the device, they create a handle with rocsparse_create_handle().

Subsequent rocSPARSE routines take this handle as an input parameter. rocSPARSE ONLY queries (by hipGetDevice()) the user’s device; rocSPARSE does NOT set the device for users. If rocSPARSE does not see a valid device, it returns an error message. It is the users’ responsibility to provide a valid device to rocSPARSE and ensure the device safety.

Users CANNOT switch devices between rocsparse_create_handle() and rocsparse_destroy_handle(). If users want to change device, they must destroy the current handle and create another rocSPARSE handle.

HIP Stream Management#

HIP kernels are always launched in a queue (also known as stream).

If users do not explicitly specify a stream, the system provides a default stream, maintained by the system. Users cannot create or destroy the default stream. However, users can freely create new streams (with hipStreamCreate()) and bind it to the rocSPARSE handle using rocsparse_set_stream(). HIP kernels are invoked in rocSPARSE routines. The rocSPARSE handle is always associated with a stream, and rocSPARSE passes its stream to the kernels inside the routine. One rocSPARSE routine only takes one stream in a single invocation. If users create a stream, they are responsible for destroying it.

Multiple Streams and Multiple Devices#

If the system under test has multiple HIP devices, users can run multiple rocSPARSE handles concurrently, but can NOT run a single rocSPARSE handle on different discrete devices. Each handle is associated with a particular singular device, and a new handle should be created for each additional device.

Storage Formats#

COO storage format#

The Coordinate (COO) storage format represents a \(m \times n\) matrix by

m

number of rows (integer).

n

number of columns (integer).

nnz

number of non-zero elements (integer).

coo_val

array of nnz elements containing the data (floating point).

coo_row_ind

array of nnz elements containing the row indices (integer).

coo_col_ind

array of nnz elements containing the column indices (integer).

The COO matrix is expected to be sorted by row indices and column indices per row. Furthermore, each pair of indices should appear only once. Consider the following \(3 \times 5\) matrix and the corresponding COO structures, with \(m = 3, n = 5\) and \(\text{nnz} = 8\) using zero based indexing:

\[\begin{split}A = \begin{pmatrix} 1.0 & 2.0 & 0.0 & 3.0 & 0.0 \\ 0.0 & 4.0 & 5.0 & 0.0 & 0.0 \\ 6.0 & 0.0 & 0.0 & 7.0 & 8.0 \\ \end{pmatrix}\end{split}\]

where

\[\begin{split}\begin{array}{ll} \text{coo_val}[8] & = \{1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0\} \\ \text{coo_row_ind}[8] & = \{0, 0, 0, 1, 1, 2, 2, 2\} \\ \text{coo_col_ind}[8] & = \{0, 1, 3, 1, 2, 0, 3, 4\} \end{array}\end{split}\]

COO (AoS) storage format#

The Coordinate (COO) Array of Structure (AoS) storage format represents a \(m \times n\) matrix by

m

number of rows (integer).

n

number of columns (integer).

nnz

number of non-zero elements (integer).

coo_val

array of nnz elements containing the data (floating point).

coo_ind

array of 2 * nnz elements containing alternating row and column indices (integer).

The COO (AoS) matrix is expected to be sorted by row indices and column indices per row. Furthermore, each pair of indices should appear only once. Consider the following \(3 \times 5\) matrix and the corresponding COO (AoS) structures, with \(m = 3, n = 5\) and \(\text{nnz} = 8\) using zero based indexing:

\[\begin{split}A = \begin{pmatrix} 1.0 & 2.0 & 0.0 & 3.0 & 0.0 \\ 0.0 & 4.0 & 5.0 & 0.0 & 0.0 \\ 6.0 & 0.0 & 0.0 & 7.0 & 8.0 \\ \end{pmatrix}\end{split}\]

where

\[\begin{split}\begin{array}{ll} \text{coo_val}[8] & = \{1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0\} \\ \text{coo_ind}[16] & = \{0, 0, 0, 1, 0, 3, 1, 1, 1, 2, 2, 0, 2, 3, 2, 4\} \\ \end{array}\end{split}\]

CSR storage format#

The Compressed Sparse Row (CSR) storage format represents a \(m \times n\) matrix by

m

number of rows (integer).

n

number of columns (integer).

nnz

number of non-zero elements (integer).

csr_val

array of nnz elements containing the data (floating point).

csr_row_ptr

array of m+1 elements that point to the start of every row (integer).

csr_col_ind

array of nnz elements containing the column indices (integer).

The CSR matrix is expected to be sorted by column indices within each row. Furthermore, each pair of indices should appear only once. Consider the following \(3 \times 5\) matrix and the corresponding CSR structures, with \(m = 3, n = 5\) and \(\text{nnz} = 8\) using one based indexing:

\[\begin{split}A = \begin{pmatrix} 1.0 & 2.0 & 0.0 & 3.0 & 0.0 \\ 0.0 & 4.0 & 5.0 & 0.0 & 0.0 \\ 6.0 & 0.0 & 0.0 & 7.0 & 8.0 \\ \end{pmatrix}\end{split}\]

where

\[\begin{split}\begin{array}{ll} \text{csr_val}[8] & = \{1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0\} \\ \text{csr_row_ptr}[4] & = \{1, 4, 6, 9\} \\ \text{csr_col_ind}[8] & = \{1, 2, 4, 2, 3, 1, 4, 5\} \end{array}\end{split}\]

CSC storage format#

The Compressed Sparse Column (CSC) storage format represents a \(m \times n\) matrix by

m

number of rows (integer).

n

number of columns (integer).

nnz

number of non-zero elements (integer).

csc_val

array of nnz elements containing the data (floating point).

csc_col_ptr

array of n+1 elements that point to the start of every column (integer).

csc_row_ind

array of nnz elements containing the row indices (integer).

The CSC matrix is expected to be sorted by row indices within each column. Furthermore, each pair of indices should appear only once. Consider the following \(3 \times 5\) matrix and the corresponding CSC structures, with \(m = 3, n = 5\) and \(\text{nnz} = 8\) using one based indexing:

\[\begin{split}A = \begin{pmatrix} 1.0 & 2.0 & 0.0 & 3.0 & 0.0 \\ 0.0 & 4.0 & 5.0 & 0.0 & 0.0 \\ 6.0 & 0.0 & 0.0 & 7.0 & 8.0 \\ \end{pmatrix}\end{split}\]

where

\[\begin{split}\begin{array}{ll} \text{csc_val}[8] & = \{1.0, 6.0, 2.0, 4.0, 5.0, 3.0, 7.0, 8.0\} \\ \text{csc_col_ptr}[6] & = \{1, 3, 5, 6, 8, 9\} \\ \text{csc_row_ind}[8] & = \{1, 3, 1, 2, 2, 1, 3, 3\} \end{array}\end{split}\]

BSR storage format#

The Block Compressed Sparse Row (BSR) storage format represents a \((mb \cdot \text{bsr_dim}) \times (nb \cdot \text{bsr_dim})\) matrix by

mb

number of block rows (integer)

nb

number of block columns (integer)

nnzb

number of non-zero blocks (integer)

bsr_val

array of nnzb * bsr_dim * bsr_dim elements containing the data (floating point). Blocks can be stored column-major or row-major.

bsr_row_ptr

array of mb+1 elements that point to the start of every block row (integer).

bsr_col_ind

array of nnzb elements containing the block column indices (integer).

bsr_dim

dimension of each block (integer).

The BSR matrix is expected to be sorted by column indices within each row. If \(m\) or \(n\) are not evenly divisible by the block dimension, then zeros are padded to the matrix, such that \(mb = (m + \text{bsr_dim} - 1) / \text{bsr_dim}\) and \(nb = (n + \text{bsr_dim} - 1) / \text{bsr_dim}\). Consider the following \(4 \times 3\) matrix and the corresponding BSR structures, with \(\text{bsr_dim} = 2, mb = 2, nb = 2\) and \(\text{nnzb} = 4\) using zero based indexing and column-major storage:

\[\begin{split}A = \begin{pmatrix} 1.0 & 0.0 & 2.0 \\ 3.0 & 0.0 & 4.0 \\ 5.0 & 6.0 & 0.0 \\ 7.0 & 0.0 & 8.0 \\ \end{pmatrix}\end{split}\]

with the blocks \(A_{ij}\)

\[\begin{split}A_{00} = \begin{pmatrix} 1.0 & 0.0 \\ 3.0 & 0.0 \\ \end{pmatrix}, A_{01} = \begin{pmatrix} 2.0 & 0.0 \\ 4.0 & 0.0 \\ \end{pmatrix}, A_{10} = \begin{pmatrix} 5.0 & 6.0 \\ 7.0 & 0.0 \\ \end{pmatrix}, A_{11} = \begin{pmatrix} 0.0 & 0.0 \\ 8.0 & 0.0 \\ \end{pmatrix}\end{split}\]

such that

\[\begin{split}A = \begin{pmatrix} A_{00} & A_{01} \\ A_{10} & A_{11} \\ \end{pmatrix}\end{split}\]

with arrays representation

\[\begin{split}\begin{array}{ll} \text{bsr_val}[16] & = \{1.0, 3.0, 0.0, 0.0, 2.0, 4.0, 0.0, 0.0, 5.0, 7.0, 6.0, 0.0, 0.0, 8.0, 0.0, 0.0\} \\ \text{bsr_row_ptr}[3] & = \{0, 2, 4\} \\ \text{bsr_col_ind}[4] & = \{0, 1, 0, 1\} \end{array}\end{split}\]

GEBSR storage format#

The General Block Compressed Sparse Row (GEBSR) storage format represents a \((mb \cdot \text{bsr_row_dim}) \times (nb \cdot \text{bsr_col_dim})\) matrix by

mb

number of block rows (integer)

nb

number of block columns (integer)

nnzb

number of non-zero blocks (integer)

bsr_val

array of nnzb * bsr_row_dim * bsr_col_dim elements containing the data (floating point). Blocks can be stored column-major or row-major.

bsr_row_ptr

array of mb+1 elements that point to the start of every block row (integer).

bsr_col_ind

array of nnzb elements containing the block column indices (integer).

bsr_row_dim

row dimension of each block (integer).

bsr_col_dim

column dimension of each block (integer).

The GEBSR matrix is expected to be sorted by column indices within each row. If \(m\) is not evenly divisible by the row block dimension or \(n\) is not evenly divisible by the column block dimension, then zeros are padded to the matrix, such that \(mb = (m + \text{bsr_row_dim} - 1) / \text{bsr_row_dim}\) and \(nb = (n + \text{bsr_col_dim} - 1) / \text{bsr_col_dim}\). Consider the following \(4 \times 5\) matrix and the corresponding GEBSR structures, with \(\text{bsr_row_dim} = 2\), \(\text{bsr_col_dim} = 3\), mb = 2, nb = 2` and \(\text{nnzb} = 4\) using zero based indexing and column-major storage:

\[\begin{split}A = \begin{pmatrix} 1.0 & 0.0 & 0.0 & 2.0 & 0.0 \\ 3.0 & 0.0 & 4.0 & 0.0 & 0.0 \\ 5.0 & 6.0 & 0.0 & 7.0 & 0.0 \\ 0.0 & 0.0 & 8.0 & 0.0 & 9.0 \\ \end{pmatrix}\end{split}\]

with the blocks \(A_{ij}\)

\[\begin{split}A_{00} = \begin{pmatrix} 1.0 & 0.0 & 0.0 \\ 3.0 & 0.0 & 4.0 \\ \end{pmatrix}, A_{01} = \begin{pmatrix} 2.0 & 0.0 & 0.0 \\ 0.0 & 0.0 & 0.0 \\ \end{pmatrix}, A_{10} = \begin{pmatrix} 5.0 & 6.0 & 0.0 \\ 0.0 & 0.0 & 8.0 \\ \end{pmatrix}, A_{11} = \begin{pmatrix} 7.0 & 0.0 & 0.0 \\ 0.0 & 9.0 & 0.0 \\ \end{pmatrix}\end{split}\]

such that

\[\begin{split}A = \begin{pmatrix} A_{00} & A_{01} \\ A_{10} & A_{11} \\ \end{pmatrix}\end{split}\]

with arrays representation

\[\begin{split}\begin{array}{ll} \text{bsr_val}[24] & = \{1.0, 3.0, 0.0, 0.0, 0.0, 4.0, 2.0, 0.0, 0.0, 0.0, 0.0, 0.0, 5.0, 0.0, 6.0, 0.0, 0.0, 8.0, 7.0, 0.0, 0.0, 9.0, 0.0, 0.0\} \\ \text{bsr_row_ptr}[3] & = \{0, 2, 4\} \\ \text{bsr_col_ind}[4] & = \{0, 1, 0, 1\} \end{array}\end{split}\]

ELL storage format#

The Ellpack-Itpack (ELL) storage format represents a \(m \times n\) matrix by

m

number of rows (integer).

n

number of columns (integer).

ell_width

maximum number of non-zero elements per row (integer)

ell_val

array of m times ell_width elements containing the data (floating point).

ell_col_ind

array of m times ell_width elements containing the column indices (integer).

The ELL matrix is assumed to be stored in column-major format. Rows with less than ell_width non-zero elements are padded with zeros (ell_val) and \(-1\) (ell_col_ind). Consider the following \(3 \times 5\) matrix and the corresponding ELL structures, with \(m = 3, n = 5\) and \(\text{ell_width} = 3\) using zero based indexing:

\[\begin{split}A = \begin{pmatrix} 1.0 & 2.0 & 0.0 & 3.0 & 0.0 \\ 0.0 & 4.0 & 5.0 & 0.0 & 0.0 \\ 6.0 & 0.0 & 0.0 & 7.0 & 8.0 \\ \end{pmatrix}\end{split}\]

where

\[\begin{split}\begin{array}{ll} \text{ell_val}[9] & = \{1.0, 4.0, 6.0, 2.0, 5.0, 7.0, 3.0, 0.0, 8.0\} \\ \text{ell_col_ind}[9] & = \{0, 1, 0, 1, 2, 3, 3, -1, 4\} \end{array}\end{split}\]

HYB storage format#

The Hybrid (HYB) storage format represents a \(m \times n\) matrix by

m

number of rows (integer).

n

number of columns (integer).

nnz

number of non-zero elements of the COO part (integer)

ell_width

maximum number of non-zero elements per row of the ELL part (integer)

ell_val

array of m times ell_width elements containing the ELL part data (floating point).

ell_col_ind

array of m times ell_width elements containing the ELL part column indices (integer).

coo_val

array of nnz elements containing the COO part data (floating point).

coo_row_ind

array of nnz elements containing the COO part row indices (integer).

coo_col_ind

array of nnz elements containing the COO part column indices (integer).

The HYB format is a combination of the ELL and COO sparse matrix formats. Typically, the regular part of the matrix is stored in ELL storage format, and the irregular part of the matrix is stored in COO storage format. Three different partitioning schemes can be applied when converting a CSR matrix to a matrix in HYB storage format. For further details on the partitioning schemes, see rocsparse_hyb_partition.

Types#

rocsparse_handle#

typedef struct _rocsparse_handle *rocsparse_handle#

Handle to the rocSPARSE library context queue.

The rocSPARSE handle is a structure holding the rocSPARSE library context. It must be initialized using rocsparse_create_handle() and the returned handle must be passed to all subsequent library function calls. It should be destroyed at the end using rocsparse_destroy_handle().

rocsparse_mat_descr#

typedef struct _rocsparse_mat_descr *rocsparse_mat_descr#

Descriptor of the matrix.

The rocSPARSE matrix descriptor is a structure holding all properties of a matrix. It must be initialized using rocsparse_create_mat_descr() and the returned descriptor must be passed to all subsequent library calls that involve the matrix. It should be destroyed at the end using rocsparse_destroy_mat_descr().

rocsparse_mat_info#

typedef struct _rocsparse_mat_info *rocsparse_mat_info#

Info structure to hold all matrix meta data.

The rocSPARSE matrix info is a structure holding all matrix information that is gathered during analysis routines. It must be initialized using rocsparse_create_mat_info() and the returned info structure must be passed to all subsequent library calls that require additional matrix information. It should be destroyed at the end using rocsparse_destroy_mat_info().

rocsparse_hyb_mat#

typedef struct _rocsparse_hyb_mat *rocsparse_hyb_mat#

HYB matrix storage format.

The rocSPARSE HYB matrix structure holds the HYB matrix. It must be initialized using rocsparse_create_hyb_mat() and the returned HYB matrix must be passed to all subsequent library calls that involve the matrix. It should be destroyed at the end using rocsparse_destroy_hyb_mat().

For more details on the HYB format, see HYB storage format.

rocsparse_action#

enum rocsparse_action#

Specify where the operation is performed on.

The rocsparse_action indicates whether the operation is performed on the full matrix, or only on the sparsity pattern of the matrix.

Values:

enumerator rocsparse_action_symbolic#

Operate only on indices.

enumerator rocsparse_action_numeric#

Operate on data and indices.

rocsparse_direction#

enum rocsparse_direction#

Specify the matrix direction.

The rocsparse_direction indicates whether a dense matrix should be parsed by rows or by columns, assuming column-major storage.

Values:

enumerator rocsparse_direction_row#

Parse the matrix by rows.

enumerator rocsparse_direction_column#

Parse the matrix by columns.

rocsparse_hyb_partition#

enum rocsparse_hyb_partition#

HYB matrix partitioning type.

The rocsparse_hyb_partition type indicates how the hybrid format partitioning between COO and ELL storage formats is performed.

Values:

enumerator rocsparse_hyb_partition_auto#

automatically decide on ELL nnz per row.

enumerator rocsparse_hyb_partition_user#

user given ELL nnz per row.

enumerator rocsparse_hyb_partition_max#

max ELL nnz per row, no COO part.

rocsparse_index_base#

enum rocsparse_index_base#

Specify the matrix index base.

The rocsparse_index_base indicates the index base of the indices. For a given rocsparse_mat_descr, the rocsparse_index_base can be set using rocsparse_set_mat_index_base(). The current rocsparse_index_base of a matrix can be obtained by rocsparse_get_mat_index_base().

Values:

enumerator rocsparse_index_base_zero#

zero based indexing.

enumerator rocsparse_index_base_one#

one based indexing.

rocsparse_matrix_type#

enum rocsparse_matrix_type#

Specify the matrix type.

The rocsparse_matrix_type indices the type of a matrix. For a given rocsparse_mat_descr, the rocsparse_matrix_type can be set using rocsparse_set_mat_type(). The current rocsparse_matrix_type of a matrix can be obtained by rocsparse_get_mat_type().

Values:

enumerator rocsparse_matrix_type_general#

general matrix type.

enumerator rocsparse_matrix_type_symmetric#

symmetric matrix type.

enumerator rocsparse_matrix_type_hermitian#

hermitian matrix type.

enumerator rocsparse_matrix_type_triangular#

triangular matrix type.

rocsparse_fill_mode#

enum rocsparse_fill_mode#

Specify the matrix fill mode.

The rocsparse_fill_mode indicates whether the lower or the upper part is stored in a sparse triangular matrix. For a given rocsparse_mat_descr, the rocsparse_fill_mode can be set using rocsparse_set_mat_fill_mode(). The current rocsparse_fill_mode of a matrix can be obtained by rocsparse_get_mat_fill_mode().

Values:

enumerator rocsparse_fill_mode_lower#

lower triangular part is stored.

enumerator rocsparse_fill_mode_upper#

upper triangular part is stored.

rocsparse_storage_mode#

enum rocsparse_storage_mode#

Specify whether the matrix is stored sorted or not.

The rocsparse_storage_mode indicates whether the matrix is stored sorted or not. For a given rocsparse_mat_descr, the rocsparse_storage_mode can be set using rocsparse_set_storage_mode(). The current rocsparse_storage_mode of a matrix can be obtained by rocsparse_get_mat_storage_mode().

Values:

enumerator rocsparse_storage_mode_sorted#

matrix is sorted.

enumerator rocsparse_storage_mode_unsorted#

matrix is unsorted.

rocsparse_diag_type#

enum rocsparse_diag_type#

Indicates if the diagonal entries are unity.

The rocsparse_diag_type indicates whether the diagonal entries of a matrix are unity or not. If rocsparse_diag_type_unit is specified, all present diagonal values will be ignored. For a given rocsparse_mat_descr, the rocsparse_diag_type can be set using rocsparse_set_mat_diag_type(). The current rocsparse_diag_type of a matrix can be obtained by rocsparse_get_mat_diag_type().

Values:

enumerator rocsparse_diag_type_non_unit#

diagonal entries are non-unity.

enumerator rocsparse_diag_type_unit#

diagonal entries are unity

rocsparse_operation#

enum rocsparse_operation#

Specify whether the matrix is to be transposed or not.

The rocsparse_operation indicates the operation performed with the given matrix.

Values:

enumerator rocsparse_operation_none#

Operate with matrix.

enumerator rocsparse_operation_transpose#

Operate with transpose.

enumerator rocsparse_operation_conjugate_transpose#

Operate with conj. transpose.

rocsparse_pointer_mode#

enum rocsparse_pointer_mode#

Indicates if the pointer is device pointer or host pointer.

The rocsparse_pointer_mode indicates whether scalar values are passed by reference on the host or device. The rocsparse_pointer_mode can be changed by rocsparse_set_pointer_mode(). The currently used pointer mode can be obtained by rocsparse_get_pointer_mode().

Values:

enumerator rocsparse_pointer_mode_host#

scalar pointers are in host memory.

enumerator rocsparse_pointer_mode_device#

scalar pointers are in device memory.

rocsparse_analysis_policy#

enum rocsparse_analysis_policy#

Specify policy in analysis functions.

The rocsparse_analysis_policy specifies whether gathered analysis data should be re-used or not. If meta data from a previous e.g. rocsparse_csrilu0_analysis() call is available, it can be re-used for subsequent calls to e.g. rocsparse_csrsv_analysis() and greatly improve performance of the analysis function.

Values:

enumerator rocsparse_analysis_policy_reuse#

try to re-use meta data.

enumerator rocsparse_analysis_policy_force#

force to re-build meta data.

rocsparse_solve_policy#

enum rocsparse_solve_policy#

Specify policy in triangular solvers and factorizations.

This is a placeholder.

Values:

enumerator rocsparse_solve_policy_auto#

automatically decide on level information.

rocsparse_layer_mode#

enum rocsparse_layer_mode#

Indicates if layer is active with bitmask.

The rocsparse_layer_mode bit mask indicates the logging characteristics.

Values:

enumerator rocsparse_layer_mode_none#

layer is not active.

enumerator rocsparse_layer_mode_log_trace#

layer is in logging mode.

enumerator rocsparse_layer_mode_log_bench#

layer is in benchmarking mode.

enumerator rocsparse_layer_mode_log_debug#

layer is in debug mode.

For more details on logging, see Logging.

rocsparse_status#

enum rocsparse_status#

List of rocsparse status codes definition.

This is a list of the rocsparse_status types that are used by the rocSPARSE library.

Values:

enumerator rocsparse_status_success#

success.

enumerator rocsparse_status_invalid_handle#

handle not initialized, invalid or null.

enumerator rocsparse_status_not_implemented#

function is not implemented.

enumerator rocsparse_status_invalid_pointer#

invalid pointer parameter.

enumerator rocsparse_status_invalid_size#

invalid size parameter.

enumerator rocsparse_status_memory_error#

failed memory allocation, copy, dealloc.

enumerator rocsparse_status_internal_error#

other internal library failure.

enumerator rocsparse_status_invalid_value#

invalid value parameter.

enumerator rocsparse_status_arch_mismatch#

device arch is not supported.

enumerator rocsparse_status_zero_pivot#

encountered zero pivot.

enumerator rocsparse_status_not_initialized#

descriptor has not been initialized.

enumerator rocsparse_status_type_mismatch#

index types do not match.

enumerator rocsparse_status_requires_sorted_storage#

sorted storage required.

rocsparse_indextype#

enum rocsparse_indextype#

List of rocsparse index types.

Indicates the index width of a rocsparse index type.

Values:

enumerator rocsparse_indextype_u16#

16 bit unsigned integer.

enumerator rocsparse_indextype_i32#

32 bit signed integer.

enumerator rocsparse_indextype_i64#

64 bit signed integer.

rocsparse_datatype#

enum rocsparse_datatype#

List of rocsparse data types.

Indicates the precision width of data stored in a rocsparse type.

Values:

enumerator rocsparse_datatype_f32_r#

32 bit floating point, real.

enumerator rocsparse_datatype_f64_r#

64 bit floating point, real.

enumerator rocsparse_datatype_f32_c#

32 bit floating point, complex.

enumerator rocsparse_datatype_f64_c#

64 bit floating point, complex.

rocsparse_format#

enum rocsparse_format#

List of sparse matrix formats.

This is a list of supported rocsparse_format types that are used to describe a sparse matrix.

Values:

enumerator rocsparse_format_coo#

COO sparse matrix format.

enumerator rocsparse_format_coo_aos#

COO AoS sparse matrix format.

enumerator rocsparse_format_csr#

CSR sparse matrix format.

enumerator rocsparse_format_csc#

CSC sparse matrix format.

enumerator rocsparse_format_ell#

ELL sparse matrix format.

enumerator rocsparse_format_bell#

BLOCKED ELL sparse matrix format.

enumerator rocsparse_format_bsr#

BSR sparse matrix format.

rocsparse_order#

enum rocsparse_order#

List of dense matrix ordering.

This is a list of supported rocsparse_order types that are used to describe the memory layout of a dense matrix

Values:

enumerator rocsparse_order_row#

Row major.

enumerator rocsparse_order_column#

Column major.

rocsparse_spmv_alg#

enum rocsparse_spmv_alg#

List of SpMV algorithms.

This is a list of supported rocsparse_spmv_alg types that are used to perform matrix vector product.

Values:

enumerator rocsparse_spmv_alg_default#

Default SpMV algorithm for the given format.

enumerator rocsparse_spmv_alg_coo#

COO SpMV algorithm 1 (segmented) for COO matrices.

enumerator rocsparse_spmv_alg_csr_adaptive#

CSR SpMV algorithm 1 (adaptive) for CSR matrices.

enumerator rocsparse_spmv_alg_csr_stream#

CSR SpMV algorithm 2 (stream) for CSR matrices.

enumerator rocsparse_spmv_alg_ell#

ELL SpMV algorithm for ELL matrices.

enumerator rocsparse_spmv_alg_coo_atomic#

COO SpMV algorithm 2 (atomic) for COO matrices.

enumerator rocsparse_spmv_alg_bsr#

BSR SpMV algorithm 1 for BSR matrices.

rocsparse_spmv_stage#

enum rocsparse_spmv_stage#

List of SpMV stages.

This is a list of possible stages during SpMV computation. Typical order is rocsparse_spmv_buffer_size, rocsparse_spmv_preprocess, rocsparse_spmv_compute.

Values:

enumerator rocsparse_spmv_stage_auto#

Automatic stage detection.

enumerator rocsparse_spmv_stage_buffer_size#

Returns the required buffer size.

enumerator rocsparse_spmv_stage_preprocess#

Preprocess data.

enumerator rocsparse_spmv_stage_compute#

Performs the actual SpMV computation.

rocsparse_spsv_alg#

enum rocsparse_spsv_alg#

List of SpSV algorithms.

This is a list of supported rocsparse_spsv_alg types that are used to perform triangular solve.

Values:

enumerator rocsparse_spsv_alg_default#

Default SpSV algorithm for the given format.

rocsparse_spsv_stage#

enum rocsparse_spsv_stage#

List of SpSV stages.

This is a list of possible stages during SpSV computation. Typical order is rocsparse_spsv_buffer_size, rocsparse_spsv_preprocess, rocsparse_spsv_compute.

Values:

enumerator rocsparse_spsv_stage_auto#

Automatic stage detection.

enumerator rocsparse_spsv_stage_buffer_size#

Returns the required buffer size.

enumerator rocsparse_spsv_stage_preprocess#

Preprocess data.

enumerator rocsparse_spsv_stage_compute#

Performs the actual SpSV computation.

rocsparse_spsm_alg#

enum rocsparse_spsm_alg#

List of SpSM algorithms.

This is a list of supported rocsparse_spsm_alg types that are used to perform triangular solve.

Values:

enumerator rocsparse_spsm_alg_default#

Default SpSM algorithm for the given format.

rocsparse_spsm_stage#

enum rocsparse_spsm_stage#

List of SpSM stages.

This is a list of possible stages during SpSM computation. Typical order is rocsparse_spsm_buffer_size, rocsparse_spsm_preprocess, rocsparse_spsm_compute.

Values:

enumerator rocsparse_spsm_stage_auto#

Automatic stage detection.

enumerator rocsparse_spsm_stage_buffer_size#

Returns the required buffer size.

enumerator rocsparse_spsm_stage_preprocess#

Preprocess data.

enumerator rocsparse_spsm_stage_compute#

Performs the actual SpSM computation.

rocsparse_spmm_alg#

enum rocsparse_spmm_alg#

List of SpMM algorithms.

This is a list of supported rocsparse_spmm_alg types that are used to perform matrix vector product.

Values:

enumerator rocsparse_spmm_alg_default#

Default SpMM algorithm for the given format.

enumerator rocsparse_spmm_alg_csr#

SpMM algorithm for CSR format using row split and shared memory.

enumerator rocsparse_spmm_alg_coo_segmented#

SpMM algorithm for COO format using segmented scan.

enumerator rocsparse_spmm_alg_coo_atomic#

SpMM algorithm for COO format using atomics.

enumerator rocsparse_spmm_alg_csr_row_split#

SpMM algorithm for CSR format using row split and shfl.

enumerator rocsparse_spmm_alg_csr_merge#

SpMM algorithm for CSR format using conversion to COO.

enumerator rocsparse_spmm_alg_coo_segmented_atomic#

SpMM algorithm for COO format using segmented scan and atomics.

enumerator rocsparse_spmm_alg_bell#

SpMM algorithm for Blocked ELL format.

enumerator rocsparse_spmm_alg_bsr#

SpMM algorithm for BSR format.

rocsparse_spmm_stage#

enum rocsparse_spmm_stage#

List of SpMM stages.

This is a list of possible stages during SpMM computation. Typical order is rocsparse_spmm_buffer_size, rocsparse_spmm_preprocess, rocsparse_spmm_compute.

Values:

enumerator rocsparse_spmm_stage_auto#

Automatic stage detection.

enumerator rocsparse_spmm_stage_buffer_size#

Returns the required buffer size.

enumerator rocsparse_spmm_stage_preprocess#

Preprocess data.

enumerator rocsparse_spmm_stage_compute#

Performs the actual SpMM computation.

rocsparse_sddmm_alg#

enum rocsparse_sddmm_alg#

List of sddmm algorithms.

This is a list of supported rocsparse_sddmm_alg types that are used to perform matrix vector product.

Values:

enumerator rocsparse_sddmm_alg_default#

Default sddmm algorithm for the given format.

rocsparse_spgemm_stage#

enum rocsparse_spgemm_stage#

List of SpGEMM stages.

This is a list of possible stages during SpGEMM computation. Typical order is rocsparse_spgemm_buffer_size, rocsparse_spgemm_nnz, rocsparse_spgemm_compute.

Values:

enumerator rocsparse_spgemm_stage_auto#

Automatic stage detection.

enumerator rocsparse_spgemm_stage_buffer_size#

Returns the required buffer size.

enumerator rocsparse_spgemm_stage_nnz#

Computes number of non-zero entries.

enumerator rocsparse_spgemm_stage_compute#

Performs the actual SpGEMM computation.

enumerator rocsparse_spgemm_stage_symbolic#

Performs the actual SpGEMM symbolic computation.

enumerator rocsparse_spgemm_stage_numeric#

Performs the actual SpGEMM numeric computation.

rocsparse_spgemm_alg#

enum rocsparse_spgemm_alg#

List of SpGEMM algorithms.

This is a list of supported rocsparse_spgemm_alg types that are used to perform sparse matrix sparse matrix product.

Values:

enumerator rocsparse_spgemm_alg_default#

Default SpGEMM algorithm for the given format.

rocsparse_sparse_to_dense_alg#

enum rocsparse_sparse_to_dense_alg#

List of sparse to dense algorithms.

This is a list of supported rocsparse_sparse_to_dense_alg types that are used to perform sparse to dense conversion.

Values:

enumerator rocsparse_sparse_to_dense_alg_default#

Default sparse to dense algorithm for the given format.

rocsparse_dense_to_sparse_alg#

enum rocsparse_dense_to_sparse_alg#

List of dense to sparse algorithms.

This is a list of supported rocsparse_dense_to_sparse_alg types that are used to perform dense to sparse conversion.

Values:

enumerator rocsparse_dense_to_sparse_alg_default#

Default dense to sparse algorithm for the given format.

rocsparse_gtsv_interleaved_alg#

enum rocsparse_gtsv_interleaved_alg#

List of interleaved gtsv algorithms.

This is a list of supported rocsparse_gtsv_interleaved_alg types that are used to perform interleaved tridiagonal solve.

Values:

enumerator rocsparse_gtsv_interleaved_alg_default#

Solve interleaved gtsv using QR algorithm (stable).

enumerator rocsparse_gtsv_interleaved_alg_thomas#

Solve interleaved gtsv using thomas algorithm (unstable).

enumerator rocsparse_gtsv_interleaved_alg_lu#

Solve interleaved gtsv using LU algorithm (stable).

enumerator rocsparse_gtsv_interleaved_alg_qr#

Solve interleaved gtsv using QR algorithm (stable).

Logging#

Three different environment variables can be set to enable logging in rocSPARSE: ROCSPARSE_LAYER, ROCSPARSE_LOG_TRACE_PATH, ROCSPARSE_LOG_BENCH_PATH and ROCSPARSE_LOG_DEBUG_PATH.

ROCSPARSE_LAYER is a bit mask, where several logging modes (rocsparse_layer_mode) can be combined as follows:

ROCSPARSE_LAYER unset

logging is disabled.

ROCSPARSE_LAYER set to 1

trace logging is enabled.

ROCSPARSE_LAYER set to 2

bench logging is enabled.

ROCSPARSE_LAYER set to 3

trace logging and bench logging is enabled.

ROCSPARSE_LAYER set to 4

debug logging is enabled.

ROCSPARSE_LAYER set to 5

trace logging and debug logging is enabled.

ROCSPARSE_LAYER set to 6

bench logging and debug logging is enabled.

ROCSPARSE_LAYER set to 7

trace logging and bench logging and debug logging is enabled.

When logging is enabled, each rocSPARSE function call will write the function name as well as function arguments to the logging stream. The default logging stream is stderr.

If the user sets the environment variable ROCSPARSE_LOG_TRACE_PATH to the full path name for a file, the file is opened and trace logging is streamed to that file. If the user sets the environment variable ROCSPARSE_LOG_BENCH_PATH to the full path name for a file, the file is opened and bench logging is streamed to that file. If the file cannot be opened, logging output is stream to stderr.

Note that performance will degrade when logging is enabled. By default, the environment variable ROCSPARSE_LAYER is unset and logging is disabled.

Exported Sparse Functions#

Auxiliary Functions#

Function name

rocsparse_create_handle()

rocsparse_destroy_handle()

rocsparse_set_stream()

rocsparse_get_stream()

rocsparse_set_pointer_mode()

rocsparse_get_pointer_mode()

rocsparse_get_version()

rocsparse_get_git_rev()

rocsparse_create_mat_descr()

rocsparse_destroy_mat_descr()

rocsparse_copy_mat_descr()

rocsparse_set_mat_index_base()

rocsparse_get_mat_index_base()

rocsparse_set_mat_type()

rocsparse_get_mat_type()

rocsparse_set_mat_fill_mode()

rocsparse_get_mat_fill_mode()

rocsparse_set_mat_diag_type()

rocsparse_get_mat_diag_type()

rocsparse_set_mat_storage_mode()

rocsparse_get_mat_storage_mode()

rocsparse_create_hyb_mat()

rocsparse_destroy_hyb_mat()

rocsparse_copy_hyb_mat()

rocsparse_create_mat_info()

rocsparse_copy_mat_info()

rocsparse_destroy_mat_info()

rocsparse_create_color_info()

rocsparse_destroy_color_info()

rocsparse_copy_color_info()

rocsparse_create_spvec_descr()

rocsparse_destroy_spvec_descr()

rocsparse_spvec_get()

rocsparse_spvec_get_index_base()

rocsparse_spvec_get_values()

rocsparse_spvec_set_values()

rocsparse_create_coo_descr()

rocsparse_create_coo_aos_descr()

rocsparse_create_csr_descr()

rocsparse_create_csc_descr()

rocsparse_create_ell_descr()

rocsparse_create_bell_descr()

rocsparse_destroy_spmat_descr()

rocsparse_coo_get()

rocsparse_coo_aos_get()

rocsparse_csr_get()

rocsparse_ell_get()

rocsparse_bell_get()

rocsparse_coo_set_pointers()

rocsparse_coo_aos_set_pointers()

rocsparse_csr_set_pointers()

rocsparse_csc_set_pointers()

rocsparse_ell_set_pointers()

rocsparse_spmat_get_size()

rocsparse_spmat_get_format()

rocsparse_spmat_get_index_base()

rocsparse_spmat_get_values()

rocsparse_spmat_set_values()

rocsparse_spmat_get_strided_batch()

rocsparse_spmat_set_strided_batch()

rocsparse_coo_set_strided_batch()

rocsparse_csr_set_strided_batch()

rocsparse_csc_set_strided_batch()

rocsparse_spmat_get_attribute()

rocsparse_spmat_set_attribute()

rocsparse_create_dnvec_descr()

rocsparse_destroy_dnvec_descr()

rocsparse_dnvec_get()

rocsparse_dnvec_get_values()

rocsparse_dnvec_set_values()

rocsparse_create_dnmat_descr()

rocsparse_destroy_dnmat_descr()

rocsparse_dnmat_get()

rocsparse_dnmat_get_values()

rocsparse_dnmat_set_values()

rocsparse_dnmat_get_strided_batch()

rocsparse_dnmat_set_strided_batch()

Sparse Level 1 Functions#

Function name

single

double

single complex

double complex

rocsparse_Xaxpyi()

x

x

x

x

rocsparse_Xdoti()

x

x

x

x

rocsparse_Xdotci()

x

x

rocsparse_Xgthr()

x

x

x

x

rocsparse_Xgthrz()

x

x

x

x

rocsparse_Xroti()

x

x

rocsparse_Xsctr()

x

x

x

x

Sparse Level 2 Functions#

Sparse Level 3 Functions#

Sparse Extra Functions#

Preconditioner Functions#

Conversion Functions#

Function name

single

double

single complex

double complex

rocsparse_csr2coo()

rocsparse_csr2csc_buffer_size()

rocsparse_Xcsr2csc()

x

x

x

x

rocsparse_Xgebsr2gebsc_buffer_size()

x

x

x

x

rocsparse_Xgebsr2gebsc()

x

x

x

x

rocsparse_csr2ell_width()

rocsparse_Xcsr2ell()

x

x

x

x

rocsparse_Xcsr2hyb()

x

x

x

x

rocsparse_csr2bsr_nnz()

rocsparse_Xcsr2bsr()

x

x

x

x

rocsparse_csr2gebsr_nnz()

rocsparse_Xcsr2gebsr_buffer_size()

x

x

x

x

rocsparse_Xcsr2gebsr()

x

x

x

x

rocsparse_coo2csr()

rocsparse_ell2csr_nnz()

rocsparse_Xell2csr()

x

x

x

x

rocsparse_hyb2csr_buffer_size()

rocsparse_Xhyb2csr()

x

x

x

x

rocsparse_Xbsr2csr()

x

x

x

x

rocsparse_Xgebsr2csr()

x

x

x

x

rocsparse_Xgebsr2gebsr_buffer_size()

x

x

x

x

rocsparse_gebsr2gebsr_nnz()

rocsparse_Xgebsr2gebsr()

x

x

x

x

rocsparse_Xcsr2csr_compress()

x

x

x

x

rocsparse_create_identity_permutation()

rocsparse_cscsort_buffer_size()

rocsparse_cscsort()

rocsparse_csrsort_buffer_size()

rocsparse_csrsort()

rocsparse_coosort_buffer_size()

rocsparse_coosort_by_row()

rocsparse_coosort_by_column()

rocsparse_Xdense2csr()

x

x

x

x

rocsparse_Xdense2csc()

x

x

x

x

rocsparse_Xdense2coo()

x

x

x

x

rocsparse_Xcsr2dense()

x

x

x

x

rocsparse_Xcsc2dense()

x

x

x

x

rocsparse_Xcoo2dense()

x

x

x

x

rocsparse_Xnnz_compress()

x

x

x

x

rocsparse_Xnnz()

x

x

x

x

rocsparse_Xprune_dense2csr_buffer_size()

x

x

rocsparse_Xprune_dense2csr_nnz()

x

x

rocsparse_Xprune_dense2csr()

x

x

rocsparse_Xprune_csr2csr_buffer_size()

x

x

rocsparse_Xprune_csr2csr_nnz()

x

x

rocsparse_Xprune_csr2csr()

x

x

rocsparse_Xprune_dense2csr_by_percentage_buffer_size()

x

x

rocsparse_Xprune_dense2csr_nnz_by_percentage()

x

x

rocsparse_Xprune_dense2csr_by_percentage()

x

x

rocsparse_Xprune_csr2csr_by_percentage_buffer_size()

x

x

rocsparse_Xprune_csr2csr_nnz_by_percentage()

x

x

rocsparse_Xprune_csr2csr_by_percentage()

x

x

rocsparse_Xbsrpad_value()

x

x

x

x

Reordering Functions#

Function name

single

double

single complex

double complex

rocsparse_Xcsrcolor()

x

x

x

x

Utility Functions#

Sparse Generic Functions#

Storage schemes and indexing base#

rocSPARSE supports 0 and 1 based indexing. The index base is selected by the rocsparse_index_base type which is either passed as standalone parameter or as part of the rocsparse_mat_descr type.

Furthermore, dense vectors are represented with a 1D array, stored linearly in memory. Sparse vectors are represented by a 1D data array stored linearly in memory that hold all non-zero elements and a 1D indexing array stored linearly in memory that hold the positions of the corresponding non-zero elements.

Pointer mode#

The auxiliary functions rocsparse_set_pointer_mode() and rocsparse_get_pointer_mode() are used to set and get the value of the state variable rocsparse_pointer_mode. If rocsparse_pointer_mode is equal to rocsparse_pointer_mode_host, then scalar parameters must be allocated on the host. If rocsparse_pointer_mode is equal to rocsparse_pointer_mode_device, then scalar parameters must be allocated on the device.

There are two types of scalar parameter:

  1. Scaling parameters, such as alpha and beta used in e.g. rocsparse_scsrmv(), rocsparse_scoomv(), …

  2. Scalar results from functions such as rocsparse_sdoti(), rocsparse_cdotci(), …

For scalar parameters such as alpha and beta, memory can be allocated on the host heap or stack, when rocsparse_pointer_mode is equal to rocsparse_pointer_mode_host. The kernel launch is asynchronous, and if the scalar parameter is on the heap, it can be freed after the return from the kernel launch. When rocsparse_pointer_mode is equal to rocsparse_pointer_mode_device, the scalar parameter must not be changed till the kernel completes.

For scalar results, when rocsparse_pointer_mode is equal to rocsparse_pointer_mode_host, the function blocks the CPU till the GPU has copied the result back to the host. Using rocsparse_pointer_mode equal to rocsparse_pointer_mode_device, the function will return after the asynchronous launch. Similarly to vector and matrix results, the scalar result is only available when the kernel has completed execution.

Asynchronous API#

Except a functions having memory allocation inside preventing asynchronicity, all rocSPARSE functions are configured to operate in non-blocking fashion with respect to CPU, meaning these library functions return immediately.

hipSPARSE#

hipSPARSE is a SPARSE marshalling library, with multiple supported backends. It sits between the application and a worker SPARSE library, marshalling inputs into the backend library and marshalling results back to the application. hipSPARSE exports an interface that does not require the client to change, regardless of the chosen backend. Currently, hipSPARSE supports rocSPARSE and cuSPARSE as backends. hipSPARSE focuses on convenience and portability. If performance outweighs these factors, then using rocSPARSE itself is highly recommended. hipSPARSE can be found on GitHub.

Sparse Auxiliary Functions#

This module holds all sparse auxiliary functions.

The functions that are contained in the auxiliary module describe all available helper functions that are required for subsequent library calls.

rocsparse_create_handle()#

rocsparse_status rocsparse_create_handle(rocsparse_handle *handle)#

Create a rocsparse handle.

rocsparse_create_handle creates the rocSPARSE library context. It must be initialized before any other rocSPARSE API function is invoked and must be passed to all subsequent library function calls. The handle should be destroyed at the end using rocsparse_destroy_handle().

Parameters:

handle[out] the pointer to the handle to the rocSPARSE library context.

Return values:
  • rocsparse_status_success – the initialization succeeded.

  • rocsparse_status_invalid_handlehandle pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_destroy_handle()#

rocsparse_status rocsparse_destroy_handle(rocsparse_handle handle)#

Destroy a rocsparse handle.

rocsparse_destroy_handle destroys the rocSPARSE library context and releases all resources used by the rocSPARSE library.

Parameters:

handle[in] the handle to the rocSPARSE library context.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handlehandle is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_set_stream()#

rocsparse_status rocsparse_set_stream(rocsparse_handle handle, hipStream_t stream)#

Specify user defined HIP stream.

rocsparse_set_stream specifies the stream to be used by the rocSPARSE library context and all subsequent function calls.

Example

This example illustrates, how a user defined stream can be used in rocSPARSE.

// Create rocSPARSE handle
rocsparse_handle handle;
rocsparse_create_handle(&handle);

// Create stream
hipStream_t stream;
hipStreamCreate(&stream);

// Set stream to rocSPARSE handle
rocsparse_set_stream(handle, stream);

// Do some work
// ...

// Clean up
rocsparse_destroy_handle(handle);
hipStreamDestroy(stream);

Parameters:
  • handle[inout] the handle to the rocSPARSE library context.

  • stream[in] the stream to be used by the rocSPARSE library context.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handlehandle is invalid.

rocsparse_get_stream()#

rocsparse_status rocsparse_get_stream(rocsparse_handle handle, hipStream_t *stream)#

Get current stream from library context.

rocsparse_get_stream gets the rocSPARSE library context stream which is currently used for all subsequent function calls.

Parameters:
  • handle[in] the handle to the rocSPARSE library context.

  • stream[out] the stream currently used by the rocSPARSE library context.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handlehandle is invalid.

rocsparse_set_pointer_mode()#

rocsparse_status rocsparse_set_pointer_mode(rocsparse_handle handle, rocsparse_pointer_mode pointer_mode)#

Specify pointer mode.

rocsparse_set_pointer_mode specifies the pointer mode to be used by the rocSPARSE library context and all subsequent function calls. By default, all values are passed by reference on the host. Valid pointer modes are rocsparse_pointer_mode_host or rocsparse_pointer_mode_device.

Parameters:
  • handle[in] the handle to the rocSPARSE library context.

  • pointer_mode[in] the pointer mode to be used by the rocSPARSE library context.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handlehandle is invalid.

rocsparse_get_pointer_mode()#

rocsparse_status rocsparse_get_pointer_mode(rocsparse_handle handle, rocsparse_pointer_mode *pointer_mode)#

Get current pointer mode from library context.

rocsparse_get_pointer_mode gets the rocSPARSE library context pointer mode which is currently used for all subsequent function calls.

Parameters:
  • handle[in] the handle to the rocSPARSE library context.

  • pointer_mode[out] the pointer mode that is currently used by the rocSPARSE library context.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handlehandle is invalid.

rocsparse_get_version()#

rocsparse_status rocsparse_get_version(rocsparse_handle handle, int *version)#

Get rocSPARSE version.

rocsparse_get_version gets the rocSPARSE library version number.

  • patch = version % 100

  • minor = version / 100 % 1000

  • major = version / 100000

Parameters:
  • handle[in] the handle to the rocSPARSE library context.

  • version[out] the version number of the rocSPARSE library.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handlehandle is invalid.

rocsparse_get_git_rev()#

rocsparse_status rocsparse_get_git_rev(rocsparse_handle handle, char *rev)#

Get rocSPARSE git revision.

rocsparse_get_git_rev gets the rocSPARSE library git commit revision (SHA-1).

Parameters:
  • handle[in] the handle to the rocSPARSE library context.

  • rev[out] the git commit revision (SHA-1).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handlehandle is invalid.

rocsparse_create_mat_descr()#

rocsparse_status rocsparse_create_mat_descr(rocsparse_mat_descr *descr)#

Create a matrix descriptor.

rocsparse_create_mat_descr creates a matrix descriptor. It initializes rocsparse_matrix_type to rocsparse_matrix_type_general and rocsparse_index_base to rocsparse_index_base_zero. It should be destroyed at the end using rocsparse_destroy_mat_descr().

Parameters:

descr[out] the pointer to the matrix descriptor.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerdescr pointer is invalid.

rocsparse_destroy_mat_descr()#

rocsparse_status rocsparse_destroy_mat_descr(rocsparse_mat_descr descr)#

Destroy a matrix descriptor.

rocsparse_destroy_mat_descr destroys a matrix descriptor and releases all resources used by the descriptor.

Parameters:

descr[in] the matrix descriptor.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerdescr is invalid.

rocsparse_copy_mat_descr()#

rocsparse_status rocsparse_copy_mat_descr(rocsparse_mat_descr dest, const rocsparse_mat_descr src)#

Copy a matrix descriptor.

rocsparse_copy_mat_descr copies a matrix descriptor. Both, source and destination matrix descriptors must be initialized prior to calling rocsparse_copy_mat_descr.

Parameters:
  • dest[out] the pointer to the destination matrix descriptor.

  • src[in] the pointer to the source matrix descriptor.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointersrc or dest pointer is invalid.

rocsparse_set_mat_index_base()#

rocsparse_status rocsparse_set_mat_index_base(rocsparse_mat_descr descr, rocsparse_index_base base)#

Specify the index base of a matrix descriptor.

rocsparse_set_mat_index_base sets the index base of a matrix descriptor. Valid options are rocsparse_index_base_zero or rocsparse_index_base_one.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerdescr pointer is invalid.

  • rocsparse_status_invalid_valuebase is invalid.

rocsparse_get_mat_index_base()#

rocsparse_index_base rocsparse_get_mat_index_base(const rocsparse_mat_descr descr)#

Get the index base of a matrix descriptor.

rocsparse_get_mat_index_base returns the index base of a matrix descriptor.

Parameters:

descr[in] the matrix descriptor.

Returns:

rocsparse_index_base_zero or rocsparse_index_base_one.

rocsparse_set_mat_type()#

rocsparse_status rocsparse_set_mat_type(rocsparse_mat_descr descr, rocsparse_matrix_type type)#

Specify the matrix type of a matrix descriptor.

rocsparse_set_mat_type sets the matrix type of a matrix descriptor. Valid matrix types are rocsparse_matrix_type_general, rocsparse_matrix_type_symmetric, rocsparse_matrix_type_hermitian or rocsparse_matrix_type_triangular.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerdescr pointer is invalid.

  • rocsparse_status_invalid_valuetype is invalid.

rocsparse_get_mat_type()#

rocsparse_matrix_type rocsparse_get_mat_type(const rocsparse_mat_descr descr)#

Get the matrix type of a matrix descriptor.

rocsparse_get_mat_type returns the matrix type of a matrix descriptor.

Parameters:

descr[in] the matrix descriptor.

Returns:

rocsparse_matrix_type_general, rocsparse_matrix_type_symmetric, rocsparse_matrix_type_hermitian or rocsparse_matrix_type_triangular.

rocsparse_set_mat_fill_mode()#

rocsparse_status rocsparse_set_mat_fill_mode(rocsparse_mat_descr descr, rocsparse_fill_mode fill_mode)#

Specify the matrix fill mode of a matrix descriptor.

rocsparse_set_mat_fill_mode sets the matrix fill mode of a matrix descriptor. Valid fill modes are rocsparse_fill_mode_lower or rocsparse_fill_mode_upper.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerdescr pointer is invalid.

  • rocsparse_status_invalid_valuefill_mode is invalid.

rocsparse_get_mat_fill_mode()#

rocsparse_fill_mode rocsparse_get_mat_fill_mode(const rocsparse_mat_descr descr)#

Get the matrix fill mode of a matrix descriptor.

rocsparse_get_mat_fill_mode returns the matrix fill mode of a matrix descriptor.

Parameters:

descr[in] the matrix descriptor.

Returns:

rocsparse_fill_mode_lower or rocsparse_fill_mode_upper.

rocsparse_set_mat_diag_type()#

rocsparse_status rocsparse_set_mat_diag_type(rocsparse_mat_descr descr, rocsparse_diag_type diag_type)#

Specify the matrix diagonal type of a matrix descriptor.

rocsparse_set_mat_diag_type sets the matrix diagonal type of a matrix descriptor. Valid diagonal types are rocsparse_diag_type_unit or rocsparse_diag_type_non_unit.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerdescr pointer is invalid.

  • rocsparse_status_invalid_valuediag_type is invalid.

rocsparse_get_mat_diag_type()#

rocsparse_diag_type rocsparse_get_mat_diag_type(const rocsparse_mat_descr descr)#

Get the matrix diagonal type of a matrix descriptor.

rocsparse_get_mat_diag_type returns the matrix diagonal type of a matrix descriptor.

Parameters:

descr[in] the matrix descriptor.

Returns:

rocsparse_diag_type_unit or rocsparse_diag_type_non_unit.

rocsparse_set_mat_storage_mode()#

rocsparse_status rocsparse_set_mat_storage_mode(rocsparse_mat_descr descr, rocsparse_storage_mode storage_mode)#

Specify the matrix storage mode of a matrix descriptor.

rocsparse_set_mat_storage_mode sets the matrix storage mode of a matrix descriptor. Valid fill modes are rocsparse_storage_mode_sorted or rocsparse_storage_mode_unsorted.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerdescr pointer is invalid.

  • rocsparse_status_invalid_valuestorage_mode is invalid.

rocsparse_get_mat_storage_mode()#

rocsparse_storage_mode rocsparse_get_mat_storage_mode(const rocsparse_mat_descr descr)#

Get the matrix storage mode of a matrix descriptor.

rocsparse_get_mat_storage_mode returns the matrix storage mode of a matrix descriptor.

Parameters:

descr[in] the matrix descriptor.

Returns:

rocsparse_storage_mode_sorted or rocsparse_storage_mode_unsorted.

rocsparse_create_hyb_mat()#

rocsparse_status rocsparse_create_hyb_mat(rocsparse_hyb_mat *hyb)#

Create a HYB matrix structure.

rocsparse_create_hyb_mat creates a structure that holds the matrix in HYB storage format. It should be destroyed at the end using rocsparse_destroy_hyb_mat().

Parameters:

hyb[inout] the pointer to the hybrid matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerhyb pointer is invalid.

rocsparse_destroy_hyb_mat()#

rocsparse_status rocsparse_destroy_hyb_mat(rocsparse_hyb_mat hyb)#

Destroy a HYB matrix structure.

rocsparse_destroy_hyb_mat destroys a HYB structure.

Parameters:

hyb[in] the hybrid matrix structure.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerhyb pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_copy_hyb_mat()#

rocsparse_status rocsparse_copy_hyb_mat(rocsparse_hyb_mat dest, const rocsparse_hyb_mat src)#

Copy a HYB matrix structure.

rocsparse_copy_hyb_mat copies a matrix info structure. Both, source and destination matrix info structure must be initialized prior to calling rocsparse_copy_hyb_mat.

Parameters:
  • dest[out] the pointer to the destination matrix info structure.

  • src[in] the pointer to the source matrix info structure.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerhyb pointer is invalid.

rocsparse_create_mat_info()#

rocsparse_status rocsparse_create_mat_info(rocsparse_mat_info *info)#

Create a matrix info structure.

rocsparse_create_mat_info creates a structure that holds the matrix info data that is gathered during the analysis routines available. It should be destroyed at the end using rocsparse_destroy_mat_info().

Parameters:

info[inout] the pointer to the info structure.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerinfo pointer is invalid.

rocsparse_copy_mat_info()#

rocsparse_status rocsparse_copy_mat_info(rocsparse_mat_info dest, const rocsparse_mat_info src)#

Copy a matrix info structure.

rocsparse_copy_mat_info copies a matrix info structure. Both, source and destination matrix info structure must be initialized prior to calling rocsparse_copy_mat_info.

Parameters:
  • dest[out] the pointer to the destination matrix info structure.

  • src[in] the pointer to the source matrix info structure.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointersrc or dest pointer is invalid.

rocsparse_destroy_mat_info()#

rocsparse_status rocsparse_destroy_mat_info(rocsparse_mat_info info)#

Destroy a matrix info structure.

rocsparse_destroy_mat_info destroys a matrix info structure.

Parameters:

info[in] the info structure.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerinfo pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_create_color_info()#

rocsparse_status rocsparse_create_color_info(rocsparse_color_info *info)#

Create a color info structure.

rocsparse_create_color_info creates a structure that holds the color info data that is gathered during the analysis routines available. It should be destroyed at the end using rocsparse_destroy_color_info().

Parameters:

info[inout] the pointer to the info structure.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerinfo pointer is invalid.

rocsparse_destroy_color_info()#

rocsparse_status rocsparse_destroy_color_info(rocsparse_color_info info)#

Destroy a color info structure.

rocsparse_destroy_color_info destroys a color info structure.

Parameters:

info[in] the info structure.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerinfo pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_copy_color_info()#

rocsparse_status rocsparse_copy_color_info(rocsparse_color_info dest, const rocsparse_color_info src)#

Copy a color info structure.

rocsparse_copy_color_info copies a color info structure. Both, source and destination color info structure must be initialized prior to calling rocsparse_copy_color_info.

Parameters:
  • dest[out] the pointer to the destination color info structure.

  • src[in] the pointer to the source color info structure.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointersrc or dest pointer is invalid.

rocsparse_create_spvec_descr()#

rocsparse_status rocsparse_create_spvec_descr(rocsparse_spvec_descr *descr, int64_t size, int64_t nnz, void *indices, void *values, rocsparse_indextype idx_type, rocsparse_index_base idx_base, rocsparse_datatype data_type)#

Create a sparse vector descriptor.

rocsparse_create_spvec_descr creates a sparse vector descriptor. It should be destroyed at the end using rocsparse_destroy_mat_descr().

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or indices or values is invalid.

  • rocsparse_status_invalid_size – if size or nnz is invalid.

  • rocsparse_status_invalid_value – if idx_type or idx_base or data_type is invalid.

rocsparse_destroy_spvec_descr()#

rocsparse_status rocsparse_destroy_spvec_descr(rocsparse_spvec_descr descr)#

Destroy a sparse vector descriptor.

rocsparse_destroy_spvec_descr destroys a sparse vector descriptor and releases all resources used by the descriptor.

Parameters:

descr[in] the matrix descriptor.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerdescr is invalid.

rocsparse_spvec_get()#

rocsparse_status rocsparse_spvec_get(const rocsparse_spvec_descr descr, int64_t *size, int64_t *nnz, void **indices, void **values, rocsparse_indextype *idx_type, rocsparse_index_base *idx_base, rocsparse_datatype *data_type)#

Get the fields of the sparse vector descriptor.

rocsparse_spvec_get gets the fields of the sparse vector descriptor

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or indices or values is invalid.

  • rocsparse_status_invalid_size – if size or nnz is invalid.

  • rocsparse_status_invalid_value – if idx_type or idx_base or data_type is invalid.

rocsparse_spvec_get_index_base()#

rocsparse_status rocsparse_spvec_get_index_base(const rocsparse_spvec_descr descr, rocsparse_index_base *idx_base)#

Get the index base stored in the sparse vector descriptor.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr is invalid.

  • rocsparse_status_invalid_value – if idx_base is invalid.

rocsparse_spvec_get_values()#

rocsparse_status rocsparse_spvec_get_values(const rocsparse_spvec_descr descr, void **values)#

Get the values array stored in the sparse vector descriptor.

Parameters:
  • descr[in] the pointer to the sparse vector descriptor.

  • values[out] non-zero values in the sparse vector (must be array of length nnz ).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or values is invalid.

rocsparse_spvec_set_values()#

rocsparse_status rocsparse_spvec_set_values(rocsparse_spvec_descr descr, void *values)#

Set the values array in the sparse vector descriptor.

Parameters:
  • descr[inout] the pointer to the sparse vector descriptor.

  • values[in] non-zero values in the sparse vector (must be array of length nnz ).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or values is invalid.

rocsparse_create_coo_descr#

rocsparse_status rocsparse_create_coo_descr(rocsparse_spmat_descr *descr, int64_t rows, int64_t cols, int64_t nnz, void *coo_row_ind, void *coo_col_ind, void *coo_val, rocsparse_indextype idx_type, rocsparse_index_base idx_base, rocsparse_datatype data_type)#

Create a sparse COO matrix descriptor.

rocsparse_create_coo_descr creates a sparse COO matrix descriptor. It should be destroyed at the end using rocsparse_destroy_spmat_descr.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or coo_row_ind or coo_col_ind or coo_val is invalid.

  • rocsparse_status_invalid_size – if rows or cols or nnz is invalid.

  • rocsparse_status_invalid_value – if idx_type or idx_base or data_type is invalid.

rocsparse_create_coo_aos_descr#

rocsparse_status rocsparse_create_coo_aos_descr(rocsparse_spmat_descr *descr, int64_t rows, int64_t cols, int64_t nnz, void *coo_ind, void *coo_val, rocsparse_indextype idx_type, rocsparse_index_base idx_base, rocsparse_datatype data_type)#

Create a sparse COO AoS matrix descriptor.

rocsparse_create_coo_aos_descr creates a sparse COO AoS matrix descriptor. It should be destroyed at the end using rocsparse_destroy_spmat_descr.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or coo_ind or coo_val is invalid.

  • rocsparse_status_invalid_size – if rows or cols or nnz is invalid.

  • rocsparse_status_invalid_value – if idx_type or idx_base or data_type is invalid.

rocsparse_create_csr_descr#

rocsparse_status rocsparse_create_csr_descr(rocsparse_spmat_descr *descr, int64_t rows, int64_t cols, int64_t nnz, void *csr_row_ptr, void *csr_col_ind, void *csr_val, rocsparse_indextype row_ptr_type, rocsparse_indextype col_ind_type, rocsparse_index_base idx_base, rocsparse_datatype data_type)#

Create a sparse CSR matrix descriptor.

rocsparse_create_csr_descr creates a sparse CSR matrix descriptor. It should be destroyed at the end using rocsparse_destroy_spmat_descr.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or csr_row_ptr or csr_col_ind or csr_val is invalid.

  • rocsparse_status_invalid_size – if rows or cols or nnz is invalid.

  • rocsparse_status_invalid_value – if row_ptr_type or col_ind_type or idx_base or data_type is invalid.

rocsparse_create_csc_descr#

rocsparse_status rocsparse_create_csc_descr(rocsparse_spmat_descr *descr, int64_t rows, int64_t cols, int64_t nnz, void *csc_col_ptr, void *csc_row_ind, void *csc_val, rocsparse_indextype col_ptr_type, rocsparse_indextype row_ind_type, rocsparse_index_base idx_base, rocsparse_datatype data_type)#

Create a sparse CSC matrix descriptor.

rocsparse_create_csc_descr creates a sparse CSC matrix descriptor. It should be destroyed at the end using rocsparse_destroy_spmat_descr.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or csc_col_ptr or csc_row_ind or csc_val is invalid.

  • rocsparse_status_invalid_size – if rows or cols or nnz is invalid.

  • rocsparse_status_invalid_value – if col_ptr_type or row_ind_type or idx_base or data_type is invalid.

rocsparse_create_ell_descr#

rocsparse_status rocsparse_create_ell_descr(rocsparse_spmat_descr *descr, int64_t rows, int64_t cols, void *ell_col_ind, void *ell_val, int64_t ell_width, rocsparse_indextype idx_type, rocsparse_index_base idx_base, rocsparse_datatype data_type)#

Create a sparse ELL matrix descriptor.

rocsparse_create_ell_descr creates a sparse ELL matrix descriptor. It should be destroyed at the end using rocsparse_destroy_spmat_descr.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or ell_col_ind or ell_val is invalid.

  • rocsparse_status_invalid_size – if rows or cols or ell_width is invalid.

  • rocsparse_status_invalid_value – if idx_type or idx_base or data_type is invalid.

rocsparse_create_bell_descr#

rocsparse_status rocsparse_create_bell_descr(rocsparse_spmat_descr *descr, int64_t rows, int64_t cols, rocsparse_direction ell_block_dir, int64_t ell_block_dim, int64_t ell_cols, void *ell_col_ind, void *ell_val, rocsparse_indextype idx_type, rocsparse_index_base idx_base, rocsparse_datatype data_type)#

Create a sparse blocked ELL matrix descriptor.

rocsparse_create_bell_descr creates a sparse blocked ELL matrix descriptor. It should be destroyed at the end using rocsparse_destroy_spmat_descr.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or ell_cols or ell_col_ind or ell_val is invalid.

  • rocsparse_status_invalid_size – if rows or cols is invalid.

  • rocsparse_status_invalid_value – if idx_type or idx_base or data_type is invalid.

rocsparse_destroy_spmat_descr#

rocsparse_status rocsparse_destroy_spmat_descr(rocsparse_spmat_descr descr)#

Destroy a sparse matrix descriptor.

rocsparse_destroy_spmat_descr destroys a sparse matrix descriptor and releases all resources used by the descriptor.

Parameters:

descr[in] the matrix descriptor.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerdescr is invalid.

rocsparse_coo_get#

rocsparse_status rocsparse_coo_get(const rocsparse_spmat_descr descr, int64_t *rows, int64_t *cols, int64_t *nnz, void **coo_row_ind, void **coo_col_ind, void **coo_val, rocsparse_indextype *idx_type, rocsparse_index_base *idx_base, rocsparse_datatype *data_type)#

Get the fields of the sparse COO matrix descriptor.

rocsparse_coo_get gets the fields of the sparse COO matrix descriptor

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or coo_row_ind or coo_col_ind or coo_val is invalid.

  • rocsparse_status_invalid_size – if rows or cols or nnz is invalid.

  • rocsparse_status_invalid_value – if idx_type or idx_base or data_type is invalid.

rocsparse_coo_aos_get#

rocsparse_status rocsparse_coo_aos_get(const rocsparse_spmat_descr descr, int64_t *rows, int64_t *cols, int64_t *nnz, void **coo_ind, void **coo_val, rocsparse_indextype *idx_type, rocsparse_index_base *idx_base, rocsparse_datatype *data_type)#

Get the fields of the sparse COO AoS matrix descriptor.

rocsparse_coo_aos_get gets the fields of the sparse COO AoS matrix descriptor

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or coo_ind or coo_val is invalid.

  • rocsparse_status_invalid_size – if rows or cols or nnz is invalid.

  • rocsparse_status_invalid_value – if idx_type or idx_base or data_type is invalid.

rocsparse_csr_get#

rocsparse_status rocsparse_csr_get(const rocsparse_spmat_descr descr, int64_t *rows, int64_t *cols, int64_t *nnz, void **csr_row_ptr, void **csr_col_ind, void **csr_val, rocsparse_indextype *row_ptr_type, rocsparse_indextype *col_ind_type, rocsparse_index_base *idx_base, rocsparse_datatype *data_type)#

Get the fields of the sparse CSR matrix descriptor.

rocsparse_csr_get gets the fields of the sparse CSR matrix descriptor

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or csr_row_ptr or csr_col_ind or csr_val is invalid.

  • rocsparse_status_invalid_size – if rows or cols or nnz is invalid.

  • rocsparse_status_invalid_value – if row_ptr_type or col_ind_type or idx_base or data_type is invalid.

rocsparse_ell_get#

rocsparse_status rocsparse_ell_get(const rocsparse_spmat_descr descr, int64_t *rows, int64_t *cols, void **ell_col_ind, void **ell_val, int64_t *ell_width, rocsparse_indextype *idx_type, rocsparse_index_base *idx_base, rocsparse_datatype *data_type)#

Get the fields of the sparse ELL matrix descriptor.

rocsparse_ell_get gets the fields of the sparse ELL matrix descriptor

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or ell_col_ind or ell_val is invalid.

  • rocsparse_status_invalid_size – if rows or cols or ell_width is invalid.

  • rocsparse_status_invalid_value – if idx_type or idx_base or data_type is invalid.

rocsparse_bell_get#

rocsparse_status rocsparse_bell_get(const rocsparse_spmat_descr descr, int64_t *rows, int64_t *cols, rocsparse_direction *ell_block_dir, int64_t *ell_block_dim, int64_t *ell_cols, void **ell_col_ind, void **ell_val, rocsparse_indextype *idx_type, rocsparse_index_base *idx_base, rocsparse_datatype *data_type)#

Get the fields of the sparse blocked ELL matrix descriptor.

rocsparse_bell_get gets the fields of the sparse blocked ELL matrix descriptor

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or ell_cols or ell_col_ind or ell_val is invalid.

  • rocsparse_status_invalid_size – if rows or cols or ell_block_dim is invalid.

  • rocsparse_status_invalid_value – if ell_block_dir or idx_type or idx_base or data_type is invalid.

rocsparse_coo_set_pointers#

rocsparse_status rocsparse_coo_set_pointers(rocsparse_spmat_descr descr, void *coo_row_ind, void *coo_col_ind, void *coo_val)#

Set the row indices, column indices and values array in the sparse COO matrix descriptor.

Parameters:
  • descr[inout] the pointer to the sparse vector descriptor.

  • coo_row_ind[in] row indices of the COO matrix (must be array of length nnz ).

  • coo_col_ind[in] column indices of the COO matrix (must be array of length nnz ).

  • coo_val[in] values of the COO matrix (must be array of length nnz ).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or coo_row_ind or coo_col_ind or coo_val is invalid.

rocsparse_coo_aos_set_pointers#

rocsparse_status rocsparse_coo_aos_set_pointers(rocsparse_spmat_descr descr, void *coo_ind, void *coo_val)#

Set the <row, column> indices and values array in the sparse COO AoS matrix descriptor.

Parameters:
  • descr[inout] the pointer to the sparse vector descriptor.

  • coo_ind[in] <row, column> indices of the COO matrix (must be array of length nnz ).

  • coo_val[in] values of the COO matrix (must be array of length nnz ).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or coo_ind or coo_val is invalid.

rocsparse_csr_set_pointers#

rocsparse_status rocsparse_csr_set_pointers(rocsparse_spmat_descr descr, void *csr_row_ptr, void *csr_col_ind, void *csr_val)#

Set the row offsets, column indices and values array in the sparse CSR matrix descriptor.

Parameters:
  • descr[inout] the pointer to the sparse vector descriptor.

  • csr_row_ptr[in] row offsets of the CSR matrix (must be array of length rows+1 ).

  • csr_col_ind[in] column indices of the CSR matrix (must be array of length nnz ).

  • csr_val[in] values of the CSR matrix (must be array of length nnz ).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or coo_ind or coo_val is invalid.

rocsparse_csc_set_pointers#

rocsparse_status rocsparse_csc_set_pointers(rocsparse_spmat_descr descr, void *csc_col_ptr, void *csc_row_ind, void *csc_val)#

Set the column offsets, row indices and values array in the sparse CSC matrix descriptor.

Parameters:
  • descr[inout] the pointer to the sparse vector descriptor.

  • csc_col_ptr[in] column offsets of the CSC matrix (must be array of length cols+1 ).

  • csc_row_ind[in] row indices of the CSC matrix (must be array of length nnz ).

  • csc_val[in] values of the CSC matrix (must be array of length nnz ).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or csc_col_ptr or csc_row_ind or csc_val is invalid.

rocsparse_ell_set_pointers#

rocsparse_status rocsparse_ell_set_pointers(rocsparse_spmat_descr descr, void *ell_col_ind, void *ell_val)#

Set the column indices and values array in the sparse ELL matrix descriptor.

Parameters:
  • descr[inout] the pointer to the sparse vector descriptor.

  • ell_col_ind[in] column indices of the ELL matrix (must be array of length rows*ell_width ).

  • ell_val[in] values of the ELL matrix (must be array of length rows*ell_width ).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or ell_col_ind or ell_val is invalid.

rocsparse_spmat_get_size#

rocsparse_status rocsparse_spmat_get_size(rocsparse_spmat_descr descr, int64_t *rows, int64_t *cols, int64_t *nnz)#

Get the number of rows, columns and non-zeros from the sparse matrix descriptor.

Parameters:
  • descr[in] the pointer to the sparse matrix descriptor.

  • rows[out] number of rows in the sparse matrix.

  • cols[out] number of columns in the sparse matrix.

  • nnz[out] number of non-zeros in sparse matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr is invalid.

  • rocsparse_status_invalid_size – if rows or cols or nnz is invalid.

rocsparse_spmat_get_format#

rocsparse_status rocsparse_spmat_get_format(const rocsparse_spmat_descr descr, rocsparse_format *format)#

Get the sparse matrix format from the sparse matrix descriptor.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr is invalid.

  • rocsparse_status_invalid_value – if format is invalid.

rocsparse_spmat_get_index_base#

rocsparse_status rocsparse_spmat_get_index_base(const rocsparse_spmat_descr descr, rocsparse_index_base *idx_base)#

Get the sparse matrix index base from the sparse matrix descriptor.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr is invalid.

  • rocsparse_status_invalid_value – if idx_base is invalid.

rocsparse_spmat_get_values#

rocsparse_status rocsparse_spmat_get_values(rocsparse_spmat_descr descr, void **values)#

Get the values array from the sparse matrix descriptor.

Parameters:
  • descr[in] the pointer to the sparse matrix descriptor.

  • values[out] values array of the sparse matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or values is invalid.

rocsparse_spmat_set_values#

rocsparse_status rocsparse_spmat_set_values(rocsparse_spmat_descr descr, void *values)#

Set the values array in the sparse matrix descriptor.

Parameters:
  • descr[inout] the pointer to the sparse matrix descriptor.

  • values[in] values array of the sparse matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or values is invalid.

rocsparse_spmat_get_strided_batch#

rocsparse_status rocsparse_spmat_get_strided_batch(rocsparse_spmat_descr descr, int *batch_count)#

Get the strided batch count from the sparse matrix descriptor.

Parameters:
  • descr[in] the pointer to the sparse matrix descriptor.

  • batch_count[out] batch_count of the sparse matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr is invalid.

  • rocsparse_status_invalid_size – if batch_count is invalid.

rocsparse_spmat_set_strided_batch#

rocsparse_status rocsparse_spmat_set_strided_batch(rocsparse_spmat_descr descr, int batch_count)#

Set the strided batch count in the sparse matrix descriptor.

Parameters:
  • descr[in] the pointer to the sparse matrix descriptor.

  • batch_count[in] batch_count of the sparse matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr is invalid.

  • rocsparse_status_invalid_size – if batch_count is invalid.

rocsparse_coo_set_strided_batch#

rocsparse_status rocsparse_coo_set_strided_batch(rocsparse_spmat_descr descr, int batch_count, int64_t batch_stride)#

Set the batch count and batch stride in the sparse COO matrix descriptor.

Parameters:
  • descr[inout] the pointer to the sparse COO matrix descriptor.

  • batch_count[in] batch_count of the sparse COO matrix.

  • batch_stride[in] batch stride of the sparse COO matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr is invalid.

  • rocsparse_status_invalid_size – if batch_count or batch_stride is invalid.

rocsparse_csr_set_strided_batch#

rocsparse_status rocsparse_csr_set_strided_batch(rocsparse_spmat_descr descr, int batch_count, int64_t offsets_batch_stride, int64_t columns_values_batch_stride)#

Set the batch count, row offset batch stride and the column indices batch stride in the sparse CSR matrix descriptor.

Parameters:
  • descr[inout] the pointer to the sparse CSR matrix descriptor.

  • batch_count[in] batch_count of the sparse CSR matrix.

  • offsets_batch_stride[in] row offset batch stride of the sparse CSR matrix.

  • columns_values_batch_stride[in] column indices batch stride of the sparse CSR matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr is invalid.

  • rocsparse_status_invalid_size – if batch_count or offsets_batch_stride or columns_values_batch_stride is invalid.

rocsparse_csc_set_strided_batch#

rocsparse_status rocsparse_csc_set_strided_batch(rocsparse_spmat_descr descr, int batch_count, int64_t offsets_batch_stride, int64_t rows_values_batch_stride)#

Set the batch count, column offset batch stride and the row indices batch stride in the sparse CSC matrix descriptor.

Parameters:
  • descr[inout] the pointer to the sparse CSC matrix descriptor.

  • batch_count[in] batch_count of the sparse CSC matrix.

  • offsets_batch_stride[in] column offset batch stride of the sparse CSC matrix.

  • rows_values_batch_stride[in] row indices batch stride of the sparse CSC matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr is invalid.

  • rocsparse_status_invalid_size – if batch_count or offsets_batch_stride or rows_values_batch_stride is invalid.

rocsparse_spmat_get_attribute#

rocsparse_status rocsparse_spmat_get_attribute(rocsparse_spmat_descr descr, rocsparse_spmat_attribute attribute, void *data, size_t data_size)#

Get the requested attribute data from the sparse matrix descriptor.

Parameters:
  • descr[in] the pointer to the sparse matrix descriptor.

  • attribute[in] rocsparse_spmat_fill_mode or rocsparse_spmat_diag_type or rocsparse_spmat_matrix_type or rocsparse_spmat_storage_mode

  • data[out] attribute data

  • data_size[in] attribute data size.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or data is invalid.

  • rocsparse_status_invalid_value – if attribute is invalid.

  • rocsparse_status_invalid_size – if data_size is invalid.

rocsparse_spmat_set_attribute#

rocsparse_status rocsparse_spmat_set_attribute(rocsparse_spmat_descr descr, rocsparse_spmat_attribute attribute, const void *data, size_t data_size)#

Set the requested attribute data in the sparse matrix descriptor.

Parameters:
  • descr[inout] the pointer to the sparse matrix descriptor.

  • attribute[in] rocsparse_spmat_fill_mode or rocsparse_spmat_diag_type or rocsparse_spmat_matrix_type or rocsparse_spmat_storage_mode

  • data[in] attribute data

  • data_size[in] attribute data size.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or data is invalid.

  • rocsparse_status_invalid_value – if attribute is invalid.

  • rocsparse_status_invalid_size – if data_size is invalid.

rocsparse_create_dnvec_descr#

rocsparse_status rocsparse_create_dnvec_descr(rocsparse_dnvec_descr *descr, int64_t size, void *values, rocsparse_datatype data_type)#

Create a dense vector descriptor.

rocsparse_create_dnvec_descr creates a dense vector descriptor. It should be destroyed at the end using rocsparse_create_dnvec_descr().

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or values is invalid.

  • rocsparse_status_invalid_size – if size is invalid.

  • rocsparse_status_invalid_value – if data_type is invalid.

rocsparse_destroy_dnvec_descr#

rocsparse_status rocsparse_destroy_dnvec_descr(rocsparse_dnvec_descr descr)#

Destroy a dense vector descriptor.

rocsparse_destroy_dnvec_descr destroys a dense vector descriptor and releases all resources used by the descriptor.

Parameters:

descr[in] the matrix descriptor.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerdescr is invalid.

rocsparse_dnvec_get#

rocsparse_status rocsparse_dnvec_get(const rocsparse_dnvec_descr descr, int64_t *size, void **values, rocsparse_datatype *data_type)#

Get the fields of the dense vector descriptor.

rocsparse_dnvec_get gets the fields of the dense vector descriptor

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or values is invalid.

  • rocsparse_status_invalid_size – if size is invalid.

  • rocsparse_status_invalid_value – if data_type is invalid.

rocsparse_dnvec_get_values#

rocsparse_status rocsparse_dnvec_get_values(const rocsparse_dnvec_descr descr, void **values)#

Get the values array from a dense vector descriptor.

Parameters:
  • descr[in] the matrix descriptor.

  • values[out] non-zero values in the dense vector (must be array of length size ).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerdescr or values is invalid.

rocsparse_dnvec_set_values#

rocsparse_status rocsparse_dnvec_set_values(rocsparse_dnvec_descr descr, void *values)#

Set the values array in a dense vector descriptor.

Parameters:
  • descr[inout] the matrix descriptor.

  • values[in] non-zero values in the dense vector (must be array of length size ).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerdescr or values is invalid.

rocsparse_create_dnmat_descr#

rocsparse_status rocsparse_create_dnmat_descr(rocsparse_dnmat_descr *descr, int64_t rows, int64_t cols, int64_t ld, void *values, rocsparse_datatype data_type, rocsparse_order order)#

Create a dense matrix descriptor.

rocsparse_create_dnmat_descr creates a dense matrix descriptor. It should be destroyed at the end using rocsparse_create_dnmat_descr().

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or values is invalid.

  • rocsparse_status_invalid_size – if rows or cols or ld is invalid.

  • rocsparse_status_invalid_value – if data_type or order is invalid.

rocsparse_destroy_dnmat_descr#

rocsparse_status rocsparse_destroy_dnmat_descr(rocsparse_dnmat_descr descr)#

Destroy a dense matrix descriptor.

rocsparse_destroy_dnmat_descr destroys a dense matrix descriptor and releases all resources used by the descriptor.

Parameters:

descr[in] the matrix descriptor.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerdescr is invalid.

rocsparse_dnmat_get#

rocsparse_status rocsparse_dnmat_get(const rocsparse_dnmat_descr descr, int64_t *rows, int64_t *cols, int64_t *ld, void **values, rocsparse_datatype *data_type, rocsparse_order *order)#

Get the fields of the dense matrix descriptor.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or values is invalid.

  • rocsparse_status_invalid_size – if rows or cols or ld is invalid.

  • rocsparse_status_invalid_value – if data_type or order is invalid.

rocsparse_dnmat_get_values#

rocsparse_status rocsparse_dnmat_get_values(const rocsparse_dnmat_descr descr, void **values)#

Get the values array from the dense matrix descriptor.

Parameters:
  • descr[in] the pointer to the dense matrix descriptor.

  • values[out] non-zero values in the dense matrix (must be array of length ld*rows if order=rocsparse_order_column or ld*cols if order=rocsparse_order_row ).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr or values is invalid.

rocsparse_dnmat_set_values#

rocsparse_status rocsparse_dnmat_set_values(rocsparse_dnmat_descr descr, void *values)#

Set the values array in a dense matrix descriptor.

Parameters:
  • descr[inout] the matrix descriptor.

  • values[in] non-zero values in the dense matrix (must be array of length ld*rows if order=rocsparse_order_column or ld*cols if order=rocsparse_order_row ).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointerdescr or values is invalid.

rocsparse_dnmat_get_strided_batch#

rocsparse_status rocsparse_dnmat_get_strided_batch(rocsparse_dnmat_descr descr, int *batch_count, int64_t *batch_stride)#

Get the batch count and batch stride from the dense matrix descriptor.

Parameters:
  • descr[in] the pointer to the dense matrix descriptor.

  • batch_count[out] the batch count in the dense matrix.

  • batch_stride[out] the batch stride in the dense matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr is invalid.

  • rocsparse_status_invalid_size – if batch_count or batch_stride is invalid.

rocsparse_dnmat_set_strided_batch#

rocsparse_status rocsparse_dnmat_set_strided_batch(rocsparse_dnmat_descr descr, int batch_count, int64_t batch_stride)#

Set the batch count and batch stride in the dense matrix descriptor.

Parameters:
  • descr[inout] the pointer to the dense matrix descriptor.

  • batch_count[in] the batch count in the dense matrix.

  • batch_stride[in] the batch stride in the dense matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_pointer – if descr is invalid.

  • rocsparse_status_invalid_size – if batch_count or batch_stride is invalid.

Sparse Level 1 Functions#

The sparse level 1 routines describe operations between a vector in sparse format and a vector in dense format. This section describes all rocSPARSE level 1 sparse linear algebra functions.

rocsparse_axpyi()#

rocsparse_status rocsparse_saxpyi(rocsparse_handle handle, rocsparse_int nnz, const float *alpha, const float *x_val, const rocsparse_int *x_ind, float *y, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_daxpyi(rocsparse_handle handle, rocsparse_int nnz, const double *alpha, const double *x_val, const rocsparse_int *x_ind, double *y, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_caxpyi(rocsparse_handle handle, rocsparse_int nnz, const rocsparse_float_complex *alpha, const rocsparse_float_complex *x_val, const rocsparse_int *x_ind, rocsparse_float_complex *y, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_zaxpyi(rocsparse_handle handle, rocsparse_int nnz, const rocsparse_double_complex *alpha, const rocsparse_double_complex *x_val, const rocsparse_int *x_ind, rocsparse_double_complex *y, rocsparse_index_base idx_base)#

Scale a sparse vector and add it to a dense vector.

rocsparse_axpyi multiplies the sparse vector \(x\) with scalar \(\alpha\) and adds the result to the dense vector \(y\), such that

\[ y := y + \alpha \cdot x \]

for(i = 0; i < nnz; ++i)
{
    y[x_ind[i]] = y[x_ind[i]] + alpha * x_val[i];
}

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • nnz[in] number of non-zero entries of vector \(x\).

  • alpha[in] scalar \(\alpha\).

  • x_val[in] array of nnz elements containing the values of \(x\).

  • x_ind[in] array of nnz elements containing the indices of the non-zero values of \(x\).

  • y[inout] array of values in dense format.

  • idx_base[in] rocsparse_index_base_zero or rocsparse_index_base_one.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base is invalid.

  • rocsparse_status_invalid_sizennz is invalid.

  • rocsparse_status_invalid_pointeralpha, x_val, x_ind or y pointer is invalid.

rocsparse_doti()#

rocsparse_status rocsparse_sdoti(rocsparse_handle handle, rocsparse_int nnz, const float *x_val, const rocsparse_int *x_ind, const float *y, float *result, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_ddoti(rocsparse_handle handle, rocsparse_int nnz, const double *x_val, const rocsparse_int *x_ind, const double *y, double *result, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_cdoti(rocsparse_handle handle, rocsparse_int nnz, const rocsparse_float_complex *x_val, const rocsparse_int *x_ind, const rocsparse_float_complex *y, rocsparse_float_complex *result, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_zdoti(rocsparse_handle handle, rocsparse_int nnz, const rocsparse_double_complex *x_val, const rocsparse_int *x_ind, const rocsparse_double_complex *y, rocsparse_double_complex *result, rocsparse_index_base idx_base)#

Compute the dot product of a sparse vector with a dense vector.

rocsparse_doti computes the dot product of the sparse vector \(x\) with the dense vector \(y\), such that

\[ \text{result} := y^T x \]

for(i = 0; i < nnz; ++i)
{
    result += x_val[i] * y[x_ind[i]];
}

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • nnz[in] number of non-zero entries of vector \(x\).

  • x_val[in] array of nnz values.

  • x_ind[in] array of nnz elements containing the indices of the non-zero values of \(x\).

  • y[in] array of values in dense format.

  • result[out] pointer to the result, can be host or device memory

  • idx_base[in] rocsparse_index_base_zero or rocsparse_index_base_one.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base is invalid.

  • rocsparse_status_invalid_sizennz is invalid.

  • rocsparse_status_invalid_pointerx_val, x_ind, y or result pointer is invalid.

  • rocsparse_status_memory_error – the buffer for the dot product reduction could not be allocated.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_dotci()#

rocsparse_status rocsparse_cdotci(rocsparse_handle handle, rocsparse_int nnz, const rocsparse_float_complex *x_val, const rocsparse_int *x_ind, const rocsparse_float_complex *y, rocsparse_float_complex *result, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_zdotci(rocsparse_handle handle, rocsparse_int nnz, const rocsparse_double_complex *x_val, const rocsparse_int *x_ind, const rocsparse_double_complex *y, rocsparse_double_complex *result, rocsparse_index_base idx_base)#

Compute the dot product of a complex conjugate sparse vector with a dense vector.

rocsparse_dotci computes the dot product of the complex conjugate sparse vector \(x\) with the dense vector \(y\), such that

\[ \text{result} := \bar{x}^H y \]

for(i = 0; i < nnz; ++i)
{
    result += conj(x_val[i]) * y[x_ind[i]];
}

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • nnz[in] number of non-zero entries of vector \(x\).

  • x_val[in] array of nnz values.

  • x_ind[in] array of nnz elements containing the indices of the non-zero values of \(x\).

  • y[in] array of values in dense format.

  • result[out] pointer to the result, can be host or device memory

  • idx_base[in] rocsparse_index_base_zero or rocsparse_index_base_one.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base is invalid.

  • rocsparse_status_invalid_sizennz is invalid.

  • rocsparse_status_invalid_pointerx_val, x_ind, y or result pointer is invalid.

  • rocsparse_status_memory_error – the buffer for the dot product reduction could not be allocated.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_gthr()#

rocsparse_status rocsparse_sgthr(rocsparse_handle handle, rocsparse_int nnz, const float *y, float *x_val, const rocsparse_int *x_ind, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_dgthr(rocsparse_handle handle, rocsparse_int nnz, const double *y, double *x_val, const rocsparse_int *x_ind, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_cgthr(rocsparse_handle handle, rocsparse_int nnz, const rocsparse_float_complex *y, rocsparse_float_complex *x_val, const rocsparse_int *x_ind, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_zgthr(rocsparse_handle handle, rocsparse_int nnz, const rocsparse_double_complex *y, rocsparse_double_complex *x_val, const rocsparse_int *x_ind, rocsparse_index_base idx_base)#

Gather elements from a dense vector and store them into a sparse vector.

rocsparse_gthr gathers the elements that are listed in x_ind from the dense vector \(y\) and stores them in the sparse vector \(x\).

for(i = 0; i < nnz; ++i)
{
    x_val[i] = y[x_ind[i]];
}

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • nnz[in] number of non-zero entries of \(x\).

  • y[in] array of values in dense format.

  • x_val[out] array of nnz elements containing the values of \(x\).

  • x_ind[in] array of nnz elements containing the indices of the non-zero values of \(x\).

  • idx_base[in] rocsparse_index_base_zero or rocsparse_index_base_one.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base is invalid.

  • rocsparse_status_invalid_sizennz is invalid.

  • rocsparse_status_invalid_pointery, x_val or x_ind pointer is invalid.

rocsparse_gthrz()#

rocsparse_status rocsparse_sgthrz(rocsparse_handle handle, rocsparse_int nnz, float *y, float *x_val, const rocsparse_int *x_ind, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_dgthrz(rocsparse_handle handle, rocsparse_int nnz, double *y, double *x_val, const rocsparse_int *x_ind, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_cgthrz(rocsparse_handle handle, rocsparse_int nnz, rocsparse_float_complex *y, rocsparse_float_complex *x_val, const rocsparse_int *x_ind, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_zgthrz(rocsparse_handle handle, rocsparse_int nnz, rocsparse_double_complex *y, rocsparse_double_complex *x_val, const rocsparse_int *x_ind, rocsparse_index_base idx_base)#

Gather and zero out elements from a dense vector and store them into a sparse vector.

rocsparse_gthrz gathers the elements that are listed in x_ind from the dense vector \(y\) and stores them in the sparse vector \(x\). The gathered elements in \(y\) are replaced by zero.

for(i = 0; i < nnz; ++i)
{
    x_val[i]    = y[x_ind[i]];
    y[x_ind[i]] = 0;
}

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • nnz[in] number of non-zero entries of \(x\).

  • y[inout] array of values in dense format.

  • x_val[out] array of nnz elements containing the non-zero values of \(x\).

  • x_ind[in] array of nnz elements containing the indices of the non-zero values of \(x\).

  • idx_base[in] rocsparse_index_base_zero or rocsparse_index_base_one.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base is invalid.

  • rocsparse_status_invalid_sizennz is invalid.

  • rocsparse_status_invalid_pointery, x_val or x_ind pointer is invalid.

rocsparse_roti()#

rocsparse_status rocsparse_sroti(rocsparse_handle handle, rocsparse_int nnz, float *x_val, const rocsparse_int *x_ind, float *y, const float *c, const float *s, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_droti(rocsparse_handle handle, rocsparse_int nnz, double *x_val, const rocsparse_int *x_ind, double *y, const double *c, const double *s, rocsparse_index_base idx_base)#

Apply Givens rotation to a dense and a sparse vector.

rocsparse_roti applies the Givens rotation matrix \(G\) to the sparse vector \(x\) and the dense vector \(y\), where

\[\begin{split} G = \begin{pmatrix} c & s \\ -s & c \end{pmatrix} \end{split}\]

for(i = 0; i < nnz; ++i)
{
    x_tmp = x_val[i];
    y_tmp = y[x_ind[i]];

    x_val[i]    = c * x_tmp + s * y_tmp;
    y[x_ind[i]] = c * y_tmp - s * x_tmp;
}

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • nnz[in] number of non-zero entries of \(x\).

  • x_val[inout] array of nnz elements containing the non-zero values of \(x\).

  • x_ind[in] array of nnz elements containing the indices of the non-zero values of \(x\).

  • y[inout] array of values in dense format.

  • c[in] pointer to the cosine element of \(G\), can be on host or device.

  • s[in] pointer to the sine element of \(G\), can be on host or device.

  • idx_base[in] rocsparse_index_base_zero or rocsparse_index_base_one.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base is invalid.

  • rocsparse_status_invalid_sizennz is invalid.

  • rocsparse_status_invalid_pointerc, s, x_val, x_ind or y pointer is invalid.

rocsparse_sctr()#

rocsparse_status rocsparse_ssctr(rocsparse_handle handle, rocsparse_int nnz, const float *x_val, const rocsparse_int *x_ind, float *y, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_dsctr(rocsparse_handle handle, rocsparse_int nnz, const double *x_val, const rocsparse_int *x_ind, double *y, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_csctr(rocsparse_handle handle, rocsparse_int nnz, const rocsparse_float_complex *x_val, const rocsparse_int *x_ind, rocsparse_float_complex *y, rocsparse_index_base idx_base)#
rocsparse_status rocsparse_zsctr(rocsparse_handle handle, rocsparse_int nnz, const rocsparse_double_complex *x_val, const rocsparse_int *x_ind, rocsparse_double_complex *y, rocsparse_index_base idx_base)#

Scatter elements from a dense vector across a sparse vector.

rocsparse_sctr scatters the elements that are listed in x_ind from the sparse vector \(x\) into the dense vector \(y\). Indices of \(y\) that are not listed in x_ind remain unchanged.

for(i = 0; i < nnz; ++i)
{
    y[x_ind[i]] = x_val[i];
}

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • nnz[in] number of non-zero entries of \(x\).

  • x_val[in] array of nnz elements containing the non-zero values of \(x\).

  • x_ind[in] array of nnz elements containing the indices of the non-zero values of x.

  • y[inout] array of values in dense format.

  • idx_base[in] rocsparse_index_base_zero or rocsparse_index_base_one.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base is invalid.

  • rocsparse_status_invalid_sizennz is invalid.

  • rocsparse_status_invalid_pointerx_val, x_ind or y pointer is invalid.

Sparse Level 2 Functions#

This module holds all sparse level 2 routines.

The sparse level 2 routines describe operations between a matrix in sparse format and a vector in dense format.

rocsparse_bsrmv_ex_analysis()#

rocsparse_status rocsparse_sbsrmv_ex_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info)#
rocsparse_status rocsparse_dbsrmv_ex_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info)#
rocsparse_status rocsparse_cbsrmv_ex_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info)#
rocsparse_status rocsparse_zbsrmv_ex_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info)#

Sparse matrix vector multiplication using BSR storage format.

rocsparse_bsrmv_ex_analysis performs the analysis step for rocsparse_sbsrmv(), rocsparse_dbsrmv(), rocsparse_cbsrmv() and rocsparse_zbsrmv(). It is expected that this function will be executed only once for a given matrix and particular operation type. The gathered analysis meta data can be cleared by rocsparse_bsrmv_ex_clear().

Note

If the matrix sparsity pattern changes, the gathered information will become invalid.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] matrix storage of BSR blocks.

  • trans[in] matrix operation type.

  • mb[in] number of block rows of the sparse BSR matrix.

  • nb[in] number of block columns of the sparse BSR matrix.

  • nnzb[in] number of non-zero blocks of the sparse BSR matrix.

  • descr[in] descriptor of the sparse BSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • bsr_val[in] array of nnzb blocks of the sparse BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnzb elements containing the block column indices of the sparse BSR matrix.

  • block_dim[in] block dimension of the sparse BSR matrix.

  • info[out] structure that holds the information collected during the analysis step.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, nb or nnzb is invalid.

  • rocsparse_status_invalid_pointerdescr, bsr_val, bsr_row_ptr, bsr_col_ind or info pointer is invalid.

  • rocsparse_status_memory_error – the buffer for the gathered information could not be allocated.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedtrans != rocsparse_operation_none or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_bsrmv_ex()#

rocsparse_status rocsparse_sbsrmv_ex(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const float *alpha, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, const float *x, const float *beta, float *y)#
rocsparse_status rocsparse_dbsrmv_ex(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const double *alpha, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, const double *x, const double *beta, double *y)#
rocsparse_status rocsparse_cbsrmv_ex(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, const rocsparse_float_complex *x, const rocsparse_float_complex *beta, rocsparse_float_complex *y)#
rocsparse_status rocsparse_zbsrmv_ex(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, const rocsparse_double_complex *x, const rocsparse_double_complex *beta, rocsparse_double_complex *y)#

Sparse matrix vector multiplication using BSR storage format.

rocsparse_bsrmv_ex multiplies the scalar \(\alpha\) with a sparse \((mb \cdot \text{block_dim}) \times (nb \cdot \text{block_dim})\) matrix, defined in BSR storage format, and the dense vector \(x\) and adds the result to the dense vector \(y\) that is multiplied by the scalar \(\beta\), such that

\[ y := \alpha \cdot op(A) \cdot x + \beta \cdot y, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans == rocsparse_operation_none} \\ A^T, & \text{if trans == rocsparse_operation_transpose} \\ A^H, & \text{if trans == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Currently, only trans == rocsparse_operation_none is supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] matrix storage of BSR blocks.

  • trans[in] matrix operation type.

  • mb[in] number of block rows of the sparse BSR matrix.

  • nb[in] number of block columns of the sparse BSR matrix.

  • nnzb[in] number of non-zero blocks of the sparse BSR matrix.

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse BSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • bsr_val[in] array of nnzb blocks of the sparse BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnzb elements containing the block column indices of the sparse BSR matrix.

  • block_dim[in] block dimension of the sparse BSR matrix.

  • x[in] array of nb*block_dim elements ( \(op(A) = A\)) or mb*block_dim elements ( \(op(A) = A^T\) or \(op(A) = A^H\)).

  • beta[in] scalar \(\beta\).

  • y[inout] array of mb*block_dim elements ( \(op(A) = A\)) or nb*block_dim elements ( \(op(A) = A^T\) or \(op(A) = A^H\)).

  • info[out] structure that holds the information collected during the analysis step.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, nb, nnzb or block_dim is invalid.

  • rocsparse_status_invalid_pointerdescr, alpha, bsr_val, bsr_row_ind, bsr_col_ind, x, beta or y pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_not_implementedtrans != rocsparse_operation_none or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_bsrmv()#

rocsparse_status rocsparse_sbsrmv(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const float *alpha, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, const float *x, const float *beta, float *y)#
rocsparse_status rocsparse_dbsrmv(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const double *alpha, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, const double *x, const double *beta, double *y)#
rocsparse_status rocsparse_cbsrmv(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, const rocsparse_float_complex *x, const rocsparse_float_complex *beta, rocsparse_float_complex *y)#
rocsparse_status rocsparse_zbsrmv(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, const rocsparse_double_complex *x, const rocsparse_double_complex *beta, rocsparse_double_complex *y)#

Sparse matrix vector multiplication using BSR storage format.

rocsparse_bsrmv multiplies the scalar \(\alpha\) with a sparse \((mb \cdot \text{block_dim}) \times (nb \cdot \text{block_dim})\) matrix, defined in BSR storage format, and the dense vector \(x\) and adds the result to the dense vector \(y\) that is multiplied by the scalar \(\beta\), such that

\[ y := \alpha \cdot op(A) \cdot x + \beta \cdot y, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans == rocsparse_operation_none} \\ A^T, & \text{if trans == rocsparse_operation_transpose} \\ A^H, & \text{if trans == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Currently, only trans == rocsparse_operation_none is supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] matrix storage of BSR blocks.

  • trans[in] matrix operation type.

  • mb[in] number of block rows of the sparse BSR matrix.

  • nb[in] number of block columns of the sparse BSR matrix.

  • nnzb[in] number of non-zero blocks of the sparse BSR matrix.

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse BSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • bsr_val[in] array of nnzb blocks of the sparse BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnzb elements containing the block column indices of the sparse BSR matrix.

  • block_dim[in] block dimension of the sparse BSR matrix.

  • x[in] array of nb*block_dim elements ( \(op(A) = A\)) or mb*block_dim elements ( \(op(A) = A^T\) or \(op(A) = A^H\)).

  • beta[in] scalar \(\beta\).

  • y[inout] array of mb*block_dim elements ( \(op(A) = A\)) or nb*block_dim elements ( \(op(A) = A^T\) or \(op(A) = A^H\)).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, nb, nnzb or block_dim is invalid.

  • rocsparse_status_invalid_pointerdescr, alpha, bsr_val, bsr_row_ind, bsr_col_ind, x, beta or y pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_not_implementedtrans != rocsparse_operation_none or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_bsrxmv()#

rocsparse_status rocsparse_sbsrxmv(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int size_of_mask, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const float *alpha, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_mask_ptr, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_end_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, const float *x, const float *beta, float *y)#
rocsparse_status rocsparse_dbsrxmv(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int size_of_mask, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const double *alpha, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_mask_ptr, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_end_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, const double *x, const double *beta, double *y)#
rocsparse_status rocsparse_cbsrxmv(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int size_of_mask, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_mask_ptr, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_end_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, const rocsparse_float_complex *x, const rocsparse_float_complex *beta, rocsparse_float_complex *y)#
rocsparse_status rocsparse_zbsrxmv(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int size_of_mask, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_mask_ptr, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_end_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, const rocsparse_double_complex *x, const rocsparse_double_complex *beta, rocsparse_double_complex *y)#

Sparse matrix vector multiplication with mask operation using BSR storage format.

rocsparse_bsrxmv multiplies the scalar \(\alpha\) with a sparse \((mb \cdot \text{block_dim}) \times (nb \cdot \text{block_dim})\) modified matrix, defined in BSR storage format, and the dense vector \(x\) and adds the result to the dense vector \(y\) that is multiplied by the scalar \(\beta\), such that

\[ y := \left( \alpha \cdot op(A) \cdot x + \beta \cdot y \right)\left( \text{mask} \right), \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans == rocsparse_operation_none} \\ A^T, & \text{if trans == rocsparse_operation_transpose} \\ A^H, & \text{if trans == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

The \(\text{mask}\) is defined as an array of block row indices. The input sparse matrix is defined with a modified BSR storage format where the beginning and the end of each row is defined with two arrays, bsr_row_ptr and bsr_end_ptr (both of size mb), rather the usual bsr_row_ptr of size mb+1.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Currently, only trans == rocsparse_operation_none is supported. Currently, block_dim==1 is not supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] matrix storage of BSR blocks.

  • trans[in] matrix operation type.

  • size_of_mask[in] number of updated block rows of the array y.

  • mb[in] number of block rows of the sparse BSR matrix.

  • nb[in] number of block columns of the sparse BSR matrix.

  • nnzb[in] number of non-zero blocks of the sparse BSR matrix.

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse BSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • bsr_val[in] array of nnzb blocks of the sparse BSR matrix.

  • bsr_mask_ptr[in] array of size_of_mask elements that give the indices of the updated block rows.

  • bsr_row_ptr[in] array of mb elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_end_ptr[in] array of mb elements that point to the end of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnzb elements containing the block column indices of the sparse BSR matrix.

  • block_dim[in] block dimension of the sparse BSR matrix.

  • x[in] array of nb*block_dim elements ( \(op(A) = A\)) or mb*block_dim elements ( \(op(A) = A^T\) or \(op(A) = A^H\)).

  • beta[in] scalar \(\beta\).

  • y[inout] array of mb*block_dim elements ( \(op(A) = A\)) or nb*block_dim elements ( \(op(A) = A^T\) or \(op(A) = A^H\)).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, nb, nnzb, block_dim or size_of_mask is invalid.

  • rocsparse_status_invalid_valuesize_of_mask is greater than mb.

  • rocsparse_status_invalid_pointerdescr, alpha, bsr_val, bsr_row_ind, bsr_col_ind, x, beta or y pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_not_implementedblock_dim==1, trans != rocsparse_operation_none or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_bsrsv_zero_pivot()#

rocsparse_status rocsparse_bsrsv_zero_pivot(rocsparse_handle handle, rocsparse_mat_info info, rocsparse_int *position)#

Sparse triangular solve using BSR storage format.

rocsparse_bsrsv_zero_pivot returns rocsparse_status_zero_pivot, if either a structural or numerical zero has been found during rocsparse_sbsrsv_solve(), rocsparse_dbsrsv_solve(), rocsparse_cbsrsv_solve() or rocsparse_zbsrsv_solve() computation. The first zero pivot \(j\) at \(A_{j,j}\) is stored in position, using same index base as the BSR matrix.

position can be in host or device memory. If no zero pivot has been found, position is set to -1 and rocsparse_status_success is returned instead.

Note

rocsparse_bsrsv_zero_pivot is a blocking function. It might influence performance negatively.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[in] structure that holds the information collected during the analysis step.

  • position[inout] pointer to zero pivot \(j\), can be in host or device memory.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo or position pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_zero_pivot – zero pivot has been found.

rocsparse_bsrsv_buffer_size()#

rocsparse_status rocsparse_sbsrsv_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_dbsrsv_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_cbsrsv_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_zbsrsv_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, size_t *buffer_size)#

Sparse triangular solve using BSR storage format.

rocsparse_bsrsv_buffer_size returns the size of the temporary storage buffer that is required by rocsparse_sbsrsv_analysis(), rocsparse_dbsrsv_analysis(), rocsparse_cbsrsv_analysis(), rocsparse_zbsrsv_analysis(), rocsparse_sbsrsv_solve(), rocsparse_dbsrsv_solve(), rocsparse_cbsrsv_solve() and rocsparse_zbsrsv_solve(). The temporary storage buffer must be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] matrix storage of BSR blocks.

  • trans[in] matrix operation type.

  • mb[in] number of block rows of the sparse BSR matrix.

  • nnzb[in] number of non-zero blocks of the sparse BSR matrix.

  • descr[in] descriptor of the sparse BSR matrix.

  • bsr_val[in] array of nnzb blocks of the sparse BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnz containing the block column indices of the sparse BSR matrix.

  • block_dim[in] block dimension of the sparse BSR matrix.

  • info[out] structure that holds the information collected during the analysis step.

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_sbsrsv_analysis(), rocsparse_dbsrsv_analysis(), rocsparse_cbsrsv_analysis(), rocsparse_zbsrsv_analysis(), rocsparse_sbsrsv_solve(), rocsparse_dbsrsv_solve(), rocsparse_cbsrsv_solve() and rocsparse_zbsrsv_solve().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, nnzb or block_dim is invalid.

  • rocsparse_status_invalid_pointerdescr, bsr_val, bsr_row_ptr, bsr_col_ind, info or buffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedtrans == rocsparse_operation_conjugate_transpose or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_bsrsv_analysis()#

rocsparse_status rocsparse_sbsrsv_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_dbsrsv_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_cbsrsv_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_zbsrsv_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#

Sparse triangular solve using BSR storage format.

rocsparse_bsrsv_analysis performs the analysis step for rocsparse_sbsrsv_solve(), rocsparse_dbsrsv_solve(), rocsparse_cbsrsv_solve() and rocsparse_zbsrsv_solve(). It is expected that this function will be executed only once for a given matrix and particular operation type. The analysis meta data can be cleared by rocsparse_bsrsv_clear().

rocsparse_bsrsv_analysis can share its meta data with rocsparse_sbsrsm_analysis(), rocsparse_dbsrsm_analysis(), rocsparse_cbsrsm_analysis(), rocsparse_zbsrsm_analysis(), rocsparse_sbsrilu0_analysis(), rocsparse_dbsrilu0_analysis(), rocsparse_cbsrilu0_analysis(), rocsparse_zbsrilu0_analysis(), rocsparse_sbsric0_analysis(), rocsparse_dbsric0_analysis(), rocsparse_cbsric0_analysis() and rocsparse_zbsric0_analysis(). Selecting rocsparse_analysis_policy_reuse policy can greatly improve computation performance of meta data. However, the user need to make sure that the sparsity pattern remains unchanged. If this cannot be assured, rocsparse_analysis_policy_force has to be used.

Note

If the matrix sparsity pattern changes, the gathered information will become invalid.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] matrix storage of BSR blocks.

  • trans[in] matrix operation type.

  • mb[in] number of block rows of the sparse BSR matrix.

  • nnzb[in] number of non-zero blocks of the sparse BSR matrix.

  • descr[in] descriptor of the sparse BSR matrix.

  • bsr_val[in] array of nnzb blocks of the sparse BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnz containing the block column indices of the sparse BSR matrix.

  • block_dim[in] block dimension of the sparse BSR matrix.

  • info[out] structure that holds the information collected during the analysis step.

  • analysis[in] rocsparse_analysis_policy_reuse or rocsparse_analysis_policy_force.

  • solve[in] rocsparse_solve_policy_auto.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, nnzb or block_dim is invalid.

  • rocsparse_status_invalid_pointerdescr, bsr_row_ptr, bsr_col_ind, info or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedtrans == rocsparse_operation_conjugate_transpose or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_bsrsv_solve()#

rocsparse_status rocsparse_sbsrsv_solve(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nnzb, const float *alpha, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, const float *x, float *y, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_dbsrsv_solve(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nnzb, const double *alpha, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, const double *x, double *y, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_cbsrsv_solve(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, const rocsparse_float_complex *x, rocsparse_float_complex *y, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_zbsrsv_solve(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, const rocsparse_double_complex *x, rocsparse_double_complex *y, rocsparse_solve_policy policy, void *temp_buffer)#

Sparse triangular solve using BSR storage format.

rocsparse_bsrsv_solve solves a sparse triangular linear system of a sparse \(m \times m\) matrix, defined in BSR storage format, a dense solution vector \(y\) and the right-hand side \(x\) that is multiplied by \(\alpha\), such that

\[ op(A) \cdot y = \alpha \cdot x, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans == rocsparse_operation_none} \\ A^T, & \text{if trans == rocsparse_operation_transpose} \\ A^H, & \text{if trans == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

rocsparse_bsrsv_solve requires a user allocated temporary buffer. Its size is returned by rocsparse_sbsrsv_buffer_size(), rocsparse_dbsrsv_buffer_size(), rocsparse_cbsrsv_buffer_size() or rocsparse_zbsrsv_buffer_size(). Furthermore, analysis meta data is required. It can be obtained by rocsparse_sbsrsv_analysis(), rocsparse_dbsrsv_analysis(), rocsparse_cbsrsv_analysis() or rocsparse_zbsrsv_analysis(). rocsparse_bsrsv_solve reports the first zero pivot (either numerical or structural zero). The zero pivot status can be checked calling rocsparse_bsrsv_zero_pivot(). If rocsparse_diag_type == rocsparse_diag_type_unit, no zero pivot will be reported, even if \(A_{j,j} = 0\) for some \(j\).

Example

Consider the lower triangular \(m \times m\) matrix \(L\), stored in BSR storage format with unit diagonal. The following example solves \(L \cdot y = x\).

// Create rocSPARSE handle
rocsparse_handle handle;
rocsparse_create_handle(&handle);

// Create matrix descriptor
rocsparse_mat_descr descr;
rocsparse_create_mat_descr(&descr);
rocsparse_set_mat_fill_mode(descr, rocsparse_fill_mode_lower);
rocsparse_set_mat_diag_type(descr, rocsparse_diag_type_unit);

// Create matrix info structure
rocsparse_mat_info info;
rocsparse_create_mat_info(&info);

// Obtain required buffer size
size_t buffer_size;
rocsparse_dbsrsv_buffer_size(handle,
                             rocsparse_direction_column,
                             rocsparse_operation_none,
                             mb,
                             nnzb,
                             descr,
                             bsr_val,
                             bsr_row_ptr,
                             bsr_col_ind,
                             block_dim,
                             info,
                             &buffer_size);

// Allocate temporary buffer
void* temp_buffer;
hipMalloc(&temp_buffer, buffer_size);

// Perform analysis step
rocsparse_dbsrsv_analysis(handle,
                          rocsparse_direction_column,
                          rocsparse_operation_none,
                          mb,
                          nnzb,
                          descr,
                          bsr_val,
                          bsr_row_ptr,
                          bsr_col_ind,
                          block_dim,
                          info,
                          rocsparse_analysis_policy_reuse,
                          rocsparse_solve_policy_auto,
                          temp_buffer);

// Solve Ly = x
rocsparse_dbsrsv_solve(handle,
                       rocsparse_direction_column,
                       rocsparse_operation_none,
                       mb,
                       nnzb,
                       &alpha,
                       descr,
                       bsr_val,
                       bsr_row_ptr,
                       bsr_col_ind,
                       block_dim,
                       info,
                       x,
                       y,
                       rocsparse_solve_policy_auto,
                       temp_buffer);

// No zero pivot should be found, with L having unit diagonal

// Clean up
hipFree(temp_buffer);
rocsparse_destroy_mat_info(info);
rocsparse_destroy_mat_descr(descr);
rocsparse_destroy_handle(handle);

Note

The sparse BSR matrix has to be sorted.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Currently, only trans == rocsparse_operation_none and trans == rocsparse_operation_transpose is supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] matrix storage of BSR blocks.

  • trans[in] matrix operation type.

  • mb[in] number of block rows of the sparse BSR matrix.

  • nnzb[in] number of non-zero blocks of the sparse BSR matrix.

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse BSR matrix.

  • bsr_val[in] array of nnzb blocks of the sparse BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnz containing the block column indices of the sparse BSR matrix.

  • block_dim[in] block dimension of the sparse BSR matrix.

  • info[in] structure that holds the information collected during the analysis step.

  • x[in] array of m elements, holding the right-hand side.

  • y[out] array of m elements, holding the solution.

  • policy[in] rocsparse_solve_policy_auto.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, nnzb or block_dim is invalid.

  • rocsparse_status_invalid_pointerdescr, alpha, bsr_val, bsr_row_ptr, bsr_col_ind, x or y pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedtrans == rocsparse_operation_conjugate_transpose or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_bsrsv_clear()#

rocsparse_status rocsparse_bsrsv_clear(rocsparse_handle handle, rocsparse_mat_info info)#

Sparse triangular solve using BSR storage format.

rocsparse_bsrsv_clear deallocates all memory that was allocated by rocsparse_sbsrsv_analysis(), rocsparse_dbsrsv_analysis(), rocsparse_cbsrsv_analysis() or rocsparse_zbsrsv_analysis(). This is especially useful, if memory is an issue and the analysis data is not required for further computation, e.g. when switching to another sparse matrix format. Calling rocsparse_bsrsv_clear is optional. All allocated resources will be cleared, when the opaque rocsparse_mat_info struct is destroyed using rocsparse_destroy_mat_info().

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[inout] structure that holds the information collected during the analysis step.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo pointer is invalid.

  • rocsparse_status_memory_error – the buffer holding the meta data could not be deallocated.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_coomv()#

rocsparse_status rocsparse_scoomv(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const float *alpha, const rocsparse_mat_descr descr, const float *coo_val, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, const float *x, const float *beta, float *y)#
rocsparse_status rocsparse_dcoomv(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const double *alpha, const rocsparse_mat_descr descr, const double *coo_val, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, const double *x, const double *beta, double *y)#
rocsparse_status rocsparse_ccoomv(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_float_complex *coo_val, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, const rocsparse_float_complex *x, const rocsparse_float_complex *beta, rocsparse_float_complex *y)#
rocsparse_status rocsparse_zcoomv(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_double_complex *coo_val, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, const rocsparse_double_complex *x, const rocsparse_double_complex *beta, rocsparse_double_complex *y)#

Sparse matrix vector multiplication using COO storage format.

rocsparse_coomv multiplies the scalar \(\alpha\) with a sparse \(m \times n\) matrix, defined in COO storage format, and the dense vector \(x\) and adds the result to the dense vector \(y\) that is multiplied by the scalar \(\beta\), such that

\[ y := \alpha \cdot op(A) \cdot x + \beta \cdot y, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans == rocsparse_operation_none} \\ A^T, & \text{if trans == rocsparse_operation_transpose} \\ A^H, & \text{if trans == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

The COO matrix has to be sorted by row indices. This can be achieved by using rocsparse_coosort_by_row().

for(i = 0; i < m; ++i)
{
    y[i] = beta * y[i];
}

for(i = 0; i < nnz; ++i)
{
    y[coo_row_ind[i]] += alpha * coo_val[i] * x[coo_col_ind[i]];
}

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans[in] matrix operation type.

  • m[in] number of rows of the sparse COO matrix.

  • n[in] number of columns of the sparse COO matrix.

  • nnz[in] number of non-zero entries of the sparse COO matrix.

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse COO matrix. Currently, only rocsparse_matrix_type_general is supported.

  • coo_val[in] array of nnz elements of the sparse COO matrix.

  • coo_row_ind[in] array of nnz elements containing the row indices of the sparse COO matrix.

  • coo_col_ind[in] array of nnz elements containing the column indices of the sparse COO matrix.

  • x[in] array of n elements ( \(op(A) = A\)) or m elements ( \(op(A) = A^T\) or \(op(A) = A^H\)).

  • beta[in] scalar \(\beta\).

  • y[inout] array of m elements ( \(op(A) = A\)) or n elements ( \(op(A) = A^T\) or \(op(A) = A^H\)).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or nnz is invalid.

  • rocsparse_status_invalid_pointerdescr, alpha, coo_val, coo_row_ind, coo_col_ind, x, beta or y pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csrmv_analysis()#

rocsparse_status rocsparse_scsrmv_analysis(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info)#
rocsparse_status rocsparse_dcsrmv_analysis(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info)#
rocsparse_status rocsparse_ccsrmv_analysis(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info)#
rocsparse_status rocsparse_zcsrmv_analysis(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info)#

Sparse matrix vector multiplication using CSR storage format.

rocsparse_csrmv_analysis performs the analysis step for rocsparse_scsrmv(), rocsparse_dcsrmv(), rocsparse_ccsrmv() and rocsparse_zcsrmv(). It is expected that this function will be executed only once for a given matrix and particular operation type. The gathered analysis meta data can be cleared by rocsparse_csrmv_clear().

Note

If the matrix sparsity pattern changes, the gathered information will become invalid.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans[in] matrix operation type.

  • m[in] number of rows of the sparse CSR matrix.

  • n[in] number of columns of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • descr[in] descriptor of the sparse CSR matrix.

  • csr_val[in] array of nnz elements of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • info[out] structure that holds the information collected during the analysis step.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or nnz is invalid.

  • rocsparse_status_invalid_pointerdescr, csr_val, csr_row_ptr, csr_col_ind or info pointer is invalid.

  • rocsparse_status_memory_error – the buffer for the gathered information could not be allocated.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implemented – if rocsparse_matrix_type is not one of rocsparse_matrix_type_general, rocsparse_matrix_type_symmetric, or rocsparse_matrix_type_triangular.

rocsparse_csrmv()#

rocsparse_status rocsparse_scsrmv(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const float *alpha, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, const float *x, const float *beta, float *y)#
rocsparse_status rocsparse_dcsrmv(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const double *alpha, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, const double *x, const double *beta, double *y)#
rocsparse_status rocsparse_ccsrmv(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, const rocsparse_float_complex *x, const rocsparse_float_complex *beta, rocsparse_float_complex *y)#
rocsparse_status rocsparse_zcsrmv(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, const rocsparse_double_complex *x, const rocsparse_double_complex *beta, rocsparse_double_complex *y)#

Sparse matrix vector multiplication using CSR storage format.

rocsparse_csrmv multiplies the scalar \(\alpha\) with a sparse \(m \times n\) matrix, defined in CSR storage format, and the dense vector \(x\) and adds the result to the dense vector \(y\) that is multiplied by the scalar \(\beta\), such that

\[ y := \alpha \cdot op(A) \cdot x + \beta \cdot y, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans == rocsparse_operation_none} \\ A^T, & \text{if trans == rocsparse_operation_transpose} \\ A^H, & \text{if trans == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

The info parameter is optional and contains information collected by rocsparse_scsrmv_analysis(), rocsparse_dcsrmv_analysis(), rocsparse_ccsrmv_analysis() or rocsparse_zcsrmv_analysis(). If present, the information will be used to speed up the csrmv computation. If info == NULL, general csrmv routine will be used instead.

for(i = 0; i < m; ++i)
{
    y[i] = beta * y[i];

    for(j = csr_row_ptr[i]; j < csr_row_ptr[i + 1]; ++j)
    {
        y[i] = y[i] + alpha * csr_val[j] * x[csr_col_ind[j]];
    }
}

Example

This example performs a sparse matrix vector multiplication in CSR format using additional meta data to improve performance.

// Create matrix info structure
rocsparse_mat_info info;
rocsparse_create_mat_info(&info);

// Perform analysis step to obtain meta data
rocsparse_scsrmv_analysis(handle,
                          rocsparse_operation_none,
                          m,
                          n,
                          nnz,
                          descr,
                          csr_val,
                          csr_row_ptr,
                          csr_col_ind,
                          info);

// Compute y = Ax
rocsparse_scsrmv(handle,
                 rocsparse_operation_none,
                 m,
                 n,
                 nnz,
                 &alpha,
                 descr,
                 csr_val,
                 csr_row_ptr,
                 csr_col_ind,
                 info,
                 x,
                 &beta,
                 y);

// Do more work
// ...

// Clean up
rocsparse_destroy_mat_info(info);

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans[in] matrix operation type.

  • m[in] number of rows of the sparse CSR matrix.

  • n[in] number of columns of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse CSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val[in] array of nnz elements of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • info[in] information collected by rocsparse_scsrmv_analysis(), rocsparse_dcsrmv_analysis(), rocsparse_ccsrmv_analysis() or rocsparse_dcsrmv_analysis(), can be NULL if no information is available.

  • x[in] array of n elements ( \(op(A) == A\)) or m elements ( \(op(A) == A^T\) or \(op(A) == A^H\)).

  • beta[in] scalar \(\beta\).

  • y[inout] array of m elements ( \(op(A) == A\)) or n elements ( \(op(A) == A^T\) or \(op(A) == A^H\)).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or nnz is invalid.

  • rocsparse_status_invalid_pointerdescr, alpha, csr_val, csr_row_ptr, csr_col_ind, x, beta or y pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csrmv_analysis_clear()#

rocsparse_status rocsparse_csrmv_clear(rocsparse_handle handle, rocsparse_mat_info info)#

Sparse matrix vector multiplication using CSR storage format.

rocsparse_csrmv_clear deallocates all memory that was allocated by rocsparse_scsrmv_analysis(), rocsparse_dcsrmv_analysis(), rocsparse_ccsrmv_analysis() or rocsparse_zcsrmv_analysis(). This is especially useful, if memory is an issue and the analysis data is not required anymore for further computation, e.g. when switching to another sparse matrix format.

Note

Calling rocsparse_csrmv_clear is optional. All allocated resources will be cleared, when the opaque rocsparse_mat_info struct is destroyed using rocsparse_destroy_mat_info().

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[inout] structure that holds the information collected during analysis step.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo pointer is invalid.

  • rocsparse_status_memory_error – the buffer for the gathered information could not be deallocated.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_csrsv_zero_pivot()#

rocsparse_status rocsparse_csrsv_zero_pivot(rocsparse_handle handle, const rocsparse_mat_descr descr, rocsparse_mat_info info, rocsparse_int *position)#

Sparse triangular solve using CSR storage format.

rocsparse_csrsv_zero_pivot returns rocsparse_status_zero_pivot, if either a structural or numerical zero has been found during rocsparse_scsrsv_solve(), rocsparse_dcsrsv_solve(), rocsparse_ccsrsv_solve() or rocsparse_zcsrsv_solve() computation. The first zero pivot \(j\) at \(A_{j,j}\) is stored in position, using same index base as the CSR matrix.

position can be in host or device memory. If no zero pivot has been found, position is set to -1 and rocsparse_status_success is returned instead.

Note

rocsparse_csrsv_zero_pivot is a blocking function. It might influence performance negatively.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • descr[in] descriptor of the sparse CSR matrix.

  • info[in] structure that holds the information collected during the analysis step.

  • position[inout] pointer to zero pivot \(j\), can be in host or device memory.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo or position pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_zero_pivot – zero pivot has been found.

rocsparse_csrsv_buffer_size()#

rocsparse_status rocsparse_scsrsv_buffer_size(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_dcsrsv_buffer_size(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_ccsrsv_buffer_size(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_zcsrsv_buffer_size(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, size_t *buffer_size)#

Sparse triangular solve using CSR storage format.

rocsparse_csrsv_buffer_size returns the size of the temporary storage buffer that is required by rocsparse_scsrsv_analysis(), rocsparse_dcsrsv_analysis(), rocsparse_ccsrsv_analysis(), rocsparse_zcsrsv_analysis(), rocsparse_scsrsv_solve(), rocsparse_dcsrsv_solve(), rocsparse_ccsrsv_solve() and rocsparse_zcsrsv_solve(). The temporary storage buffer must be allocated by the user. The size of the temporary storage buffer is identical to the size returned by rocsparse_scsrilu0_buffer_size(), rocsparse_dcsrilu0_buffer_size(), rocsparse_ccsrilu0_buffer_size() and rocsparse_zcsrilu0_buffer_size() if the matrix sparsity pattern is identical. The user allocated buffer can thus be shared between subsequent calls to those functions.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or nnz is invalid.

  • rocsparse_status_invalid_pointerdescr, csr_val, csr_row_ptr, csr_col_ind, info or buffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedtrans == rocsparse_operation_conjugate_transpose or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csrsv_analysis()#

rocsparse_status rocsparse_scsrsv_analysis(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_dcsrsv_analysis(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_ccsrsv_analysis(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_zcsrsv_analysis(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#

Sparse triangular solve using CSR storage format.

rocsparse_csrsv_analysis performs the analysis step for rocsparse_scsrsv_solve(), rocsparse_dcsrsv_solve(), rocsparse_ccsrsv_solve() and rocsparse_zcsrsv_solve(). It is expected that this function will be executed only once for a given matrix and particular operation type. The analysis meta data can be cleared by rocsparse_csrsv_clear().

rocsparse_csrsv_analysis can share its meta data with rocsparse_scsrsm_analysis(), rocsparse_dcsrsm_analysis(), rocsparse_ccsrsm_analysis(), rocsparse_zcsrsm_analysis(), rocsparse_scsrilu0_analysis(), rocsparse_dcsrilu0_analysis(), rocsparse_ccsrilu0_analysis(), rocsparse_zcsrilu0_analysis(), rocsparse_scsric0_analysis(), rocsparse_dcsric0_analysis(), rocsparse_ccsric0_analysis() and rocsparse_zcsric0_analysis(). Selecting rocsparse_analysis_policy_reuse policy can greatly improve computation performance of meta data. However, the user need to make sure that the sparsity pattern remains unchanged. If this cannot be assured, rocsparse_analysis_policy_force has to be used.

Note

If the matrix sparsity pattern changes, the gathered information will become invalid.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans[in] matrix operation type.

  • m[in] number of rows of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • descr[in] descriptor of the sparse CSR matrix.

  • csr_val[in] array of nnz elements of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • info[out] structure that holds the information collected during the analysis step.

  • analysis[in] rocsparse_analysis_policy_reuse or rocsparse_analysis_policy_force.

  • solve[in] rocsparse_solve_policy_auto.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or nnz is invalid.

  • rocsparse_status_invalid_pointerdescr, csr_row_ptr, csr_col_ind, info or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedtrans == rocsparse_operation_conjugate_transpose or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csrsv_solve()#

rocsparse_status rocsparse_scsrsv_solve(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int nnz, const float *alpha, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, const float *x, float *y, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_dcsrsv_solve(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int nnz, const double *alpha, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, const double *x, double *y, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_ccsrsv_solve(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int nnz, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, const rocsparse_float_complex *x, rocsparse_float_complex *y, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_zcsrsv_solve(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int nnz, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, const rocsparse_double_complex *x, rocsparse_double_complex *y, rocsparse_solve_policy policy, void *temp_buffer)#

Sparse triangular solve using CSR storage format.

rocsparse_csrsv_solve solves a sparse triangular linear system of a sparse \(m \times m\) matrix, defined in CSR storage format, a dense solution vector \(y\) and the right-hand side \(x\) that is multiplied by \(\alpha\), such that

\[ op(A) \cdot y = \alpha \cdot x, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans == rocsparse_operation_none} \\ A^T, & \text{if trans == rocsparse_operation_transpose} \\ A^H, & \text{if trans == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

rocsparse_csrsv_solve requires a user allocated temporary buffer. Its size is returned by rocsparse_scsrsv_buffer_size(), rocsparse_dcsrsv_buffer_size(), rocsparse_ccsrsv_buffer_size() or rocsparse_zcsrsv_buffer_size(). Furthermore, analysis meta data is required. It can be obtained by rocsparse_scsrsv_analysis(), rocsparse_dcsrsv_analysis(), rocsparse_ccsrsv_analysis() or rocsparse_zcsrsv_analysis(). rocsparse_csrsv_solve reports the first zero pivot (either numerical or structural zero). The zero pivot status can be checked calling rocsparse_csrsv_zero_pivot(). If rocsparse_diag_type == rocsparse_diag_type_unit, no zero pivot will be reported, even if \(A_{j,j} = 0\) for some \(j\).

Example

Consider the lower triangular \(m \times m\) matrix \(L\), stored in CSR storage format with unit diagonal. The following example solves \(L \cdot y = x\).

// Create rocSPARSE handle
rocsparse_handle handle;
rocsparse_create_handle(&handle);

// Create matrix descriptor
rocsparse_mat_descr descr;
rocsparse_create_mat_descr(&descr);
rocsparse_set_mat_fill_mode(descr, rocsparse_fill_mode_lower);
rocsparse_set_mat_diag_type(descr, rocsparse_diag_type_unit);

// Create matrix info structure
rocsparse_mat_info info;
rocsparse_create_mat_info(&info);

// Obtain required buffer size
size_t buffer_size;
rocsparse_dcsrsv_buffer_size(handle,
                             rocsparse_operation_none,
                             m,
                             nnz,
                             descr,
                             csr_val,
                             csr_row_ptr,
                             csr_col_ind,
                             info,
                             &buffer_size);

// Allocate temporary buffer
void* temp_buffer;
hipMalloc(&temp_buffer, buffer_size);

// Perform analysis step
rocsparse_dcsrsv_analysis(handle,
                          rocsparse_operation_none,
                          m,
                          nnz,
                          descr,
                          csr_val,
                          csr_row_ptr,
                          csr_col_ind,
                          info,
                          rocsparse_analysis_policy_reuse,
                          rocsparse_solve_policy_auto,
                          temp_buffer);

// Solve Ly = x
rocsparse_dcsrsv_solve(handle,
                       rocsparse_operation_none,
                       m,
                       nnz,
                       &alpha,
                       descr,
                       csr_val,
                       csr_row_ptr,
                       csr_col_ind,
                       info,
                       x,
                       y,
                       rocsparse_solve_policy_auto,
                       temp_buffer);

// No zero pivot should be found, with L having unit diagonal

// Clean up
hipFree(temp_buffer);
rocsparse_destroy_mat_info(info);
rocsparse_destroy_mat_descr(descr);
rocsparse_destroy_handle(handle);

Note

The sparse CSR matrix has to be sorted. This can be achieved by calling rocsparse_csrsort().

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Currently, only trans == rocsparse_operation_none and trans == rocsparse_operation_transpose is supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans[in] matrix operation type.

  • m[in] number of rows of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse CSR matrix.

  • csr_val[in] array of nnz elements of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • info[in] structure that holds the information collected during the analysis step.

  • x[in] array of m elements, holding the right-hand side.

  • y[out] array of m elements, holding the solution.

  • policy[in] rocsparse_solve_policy_auto.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or nnz is invalid.

  • rocsparse_status_invalid_pointerdescr, alpha, csr_val, csr_row_ptr, csr_col_ind, x or y pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedtrans == rocsparse_operation_conjugate_transpose or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csrsv_clear()#

rocsparse_status rocsparse_csrsv_clear(rocsparse_handle handle, const rocsparse_mat_descr descr, rocsparse_mat_info info)#

Sparse triangular solve using CSR storage format.

rocsparse_csrsv_clear deallocates all memory that was allocated by rocsparse_scsrsv_analysis(), rocsparse_dcsrsv_analysis(), rocsparse_ccsrsv_analysis() or rocsparse_zcsrsv_analysis(). This is especially useful, if memory is an issue and the analysis data is not required for further computation, e.g. when switching to another sparse matrix format. Calling rocsparse_csrsv_clear is optional. All allocated resources will be cleared, when the opaque rocsparse_mat_info struct is destroyed using rocsparse_destroy_mat_info().

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • descr[in] descriptor of the sparse CSR matrix.

  • info[inout] structure that holds the information collected during the analysis step.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo pointer is invalid.

  • rocsparse_status_memory_error – the buffer holding the meta data could not be deallocated.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_ellmv()#

rocsparse_status rocsparse_sellmv(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, const float *alpha, const rocsparse_mat_descr descr, const float *ell_val, const rocsparse_int *ell_col_ind, rocsparse_int ell_width, const float *x, const float *beta, float *y)#
rocsparse_status rocsparse_dellmv(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, const double *alpha, const rocsparse_mat_descr descr, const double *ell_val, const rocsparse_int *ell_col_ind, rocsparse_int ell_width, const double *x, const double *beta, double *y)#
rocsparse_status rocsparse_cellmv(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_float_complex *ell_val, const rocsparse_int *ell_col_ind, rocsparse_int ell_width, const rocsparse_float_complex *x, const rocsparse_float_complex *beta, rocsparse_float_complex *y)#
rocsparse_status rocsparse_zellmv(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_double_complex *ell_val, const rocsparse_int *ell_col_ind, rocsparse_int ell_width, const rocsparse_double_complex *x, const rocsparse_double_complex *beta, rocsparse_double_complex *y)#

Sparse matrix vector multiplication using ELL storage format.

rocsparse_ellmv multiplies the scalar \(\alpha\) with a sparse \(m \times n\) matrix, defined in ELL storage format, and the dense vector \(x\) and adds the result to the dense vector \(y\) that is multiplied by the scalar \(\beta\), such that

\[ y := \alpha \cdot op(A) \cdot x + \beta \cdot y, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans == rocsparse_operation_none} \\ A^T, & \text{if trans == rocsparse_operation_transpose} \\ A^H, & \text{if trans == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

for(i = 0; i < m; ++i)
{
    y[i] = beta * y[i];

    for(p = 0; p < ell_width; ++p)
    {
        idx = p * m + i;

        if((ell_col_ind[idx] >= 0) && (ell_col_ind[idx] < n))
        {
            y[i] = y[i] + alpha * ell_val[idx] * x[ell_col_ind[idx]];
        }
    }
}

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans[in] matrix operation type.

  • m[in] number of rows of the sparse ELL matrix.

  • n[in] number of columns of the sparse ELL matrix.

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse ELL matrix. Currently, only rocsparse_matrix_type_general is supported.

  • ell_val[in] array that contains the elements of the sparse ELL matrix. Padded elements should be zero.

  • ell_col_ind[in] array that contains the column indices of the sparse ELL matrix. Padded column indices should be -1.

  • ell_width[in] number of non-zero elements per row of the sparse ELL matrix.

  • x[in] array of n elements ( \(op(A) == A\)) or m elements ( \(op(A) == A^T\) or \(op(A) == A^H\)).

  • beta[in] scalar \(\beta\).

  • y[inout] array of m elements ( \(op(A) == A\)) or n elements ( \(op(A) == A^T\) or \(op(A) == A^H\)).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or ell_width is invalid.

  • rocsparse_status_invalid_pointerdescr, alpha, ell_val, ell_col_ind, x, beta or y pointer is invalid.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_hybmv()#

rocsparse_status rocsparse_shybmv(rocsparse_handle handle, rocsparse_operation trans, const float *alpha, const rocsparse_mat_descr descr, const rocsparse_hyb_mat hyb, const float *x, const float *beta, float *y)#
rocsparse_status rocsparse_dhybmv(rocsparse_handle handle, rocsparse_operation trans, const double *alpha, const rocsparse_mat_descr descr, const rocsparse_hyb_mat hyb, const double *x, const double *beta, double *y)#
rocsparse_status rocsparse_chybmv(rocsparse_handle handle, rocsparse_operation trans, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_hyb_mat hyb, const rocsparse_float_complex *x, const rocsparse_float_complex *beta, rocsparse_float_complex *y)#
rocsparse_status rocsparse_zhybmv(rocsparse_handle handle, rocsparse_operation trans, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_hyb_mat hyb, const rocsparse_double_complex *x, const rocsparse_double_complex *beta, rocsparse_double_complex *y)#

Sparse matrix vector multiplication using HYB storage format.

rocsparse_hybmv multiplies the scalar \(\alpha\) with a sparse \(m \times n\) matrix, defined in HYB storage format, and the dense vector \(x\) and adds the result to the dense vector \(y\) that is multiplied by the scalar \(\beta\), such that

\[ y := \alpha \cdot op(A) \cdot x + \beta \cdot y, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans == rocsparse_operation_none} \\ A^T, & \text{if trans == rocsparse_operation_transpose} \\ A^H, & \text{if trans == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans[in] matrix operation type.

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse HYB matrix. Currently, only rocsparse_matrix_type_general is supported.

  • hyb[in] matrix in HYB storage format.

  • x[in] array of n elements ( \(op(A) == A\)) or m elements ( \(op(A) == A^T\) or \(op(A) == A^H\)).

  • beta[in] scalar \(\beta\).

  • y[inout] array of m elements ( \(op(A) == A\)) or n elements ( \(op(A) == A^T\) or \(op(A) == A^H\)).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizehyb structure was not initialized with valid matrix sizes.

  • rocsparse_status_invalid_pointerdescr, alpha, hyb, x, beta or y pointer is invalid.

  • rocsparse_status_invalid_valuehyb structure was not initialized with a valid partitioning type.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_memory_error – the buffer could not be allocated.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedtrans != rocsparse_operation_none or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_gebsrmv()#

rocsparse_status rocsparse_sgebsrmv(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const float *alpha, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const float *x, const float *beta, float *y)#
rocsparse_status rocsparse_dgebsrmv(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const double *alpha, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const double *x, const double *beta, double *y)#
rocsparse_status rocsparse_cgebsrmv(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const rocsparse_float_complex *x, const rocsparse_float_complex *beta, rocsparse_float_complex *y)#
rocsparse_status rocsparse_zgebsrmv(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const rocsparse_double_complex *x, const rocsparse_double_complex *beta, rocsparse_double_complex *y)#

Sparse matrix vector multiplication using GEBSR storage format.

rocsparse_gebsrmv multiplies the scalar \(\alpha\) with a sparse \((mb \cdot \text{row_block_dim}) \times (nb \cdot \text{col_block_dim})\) matrix, defined in GEBSR storage format, and the dense vector \(x\) and adds the result to the dense vector \(y\) that is multiplied by the scalar \(\beta\), such that

\[ y := \alpha \cdot op(A) \cdot x + \beta \cdot y, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans == rocsparse_operation_none} \\ A^T, & \text{if trans == rocsparse_operation_transpose} \\ A^H, & \text{if trans == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Currently, only trans == rocsparse_operation_none is supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] matrix storage of GEBSR blocks.

  • trans[in] matrix operation type.

  • mb[in] number of block rows of the sparse GEBSR matrix.

  • nb[in] number of block columns of the sparse GEBSR matrix.

  • nnzb[in] number of non-zero blocks of the sparse GEBSR matrix.

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse GEBSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • bsr_val[in] array of nnzb blocks of the sparse GEBSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse GEBSR matrix.

  • bsr_col_ind[in] array of nnz containing the block column indices of the sparse GEBSR matrix.

  • row_block_dim[in] row block dimension of the sparse GEBSR matrix.

  • col_block_dim[in] column block dimension of the sparse GEBSR matrix.

  • x[in] array of nb*col_block_dim elements ( \(op(A) = A\)) or mb*row_block_dim elements ( \(op(A) = A^T\) or \(op(A) = A^H\)).

  • beta[in] scalar \(\beta\).

  • y[inout] array of mb*row_block_dim elements ( \(op(A) = A\)) or nb*col_block_dim elements ( \(op(A) = A^T\) or \(op(A) = A^H\)).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, nb, nnzb, row_block_dim or col_block_dim is invalid.

  • rocsparse_status_invalid_pointerdescr, alpha, bsr_val, bsr_row_ind, bsr_col_ind, x, beta or y pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_not_implementedtrans != rocsparse_operation_none or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_gemvi_buffer_size()#

rocsparse_status rocsparse_sgemvi_buffer_size(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, size_t *buffer_size)#
rocsparse_status rocsparse_dgemvi_buffer_size(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, size_t *buffer_size)#
rocsparse_status rocsparse_cgemvi_buffer_size(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, size_t *buffer_size)#
rocsparse_status rocsparse_zgemvi_buffer_size(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, size_t *buffer_size)#

Dense matrix sparse vector multiplication.

rocsparse_gemvi_buffer_size returns the size of the temporary storage buffer required by rocsparse_sgemvi(), rocsparse_dgemvi(), rocsparse_cgemvi() or rocsparse_zgemvi(). The temporary storage buffer must be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans[in] matrix operation type.

  • m[in] number of rows of the dense matrix.

  • n[in] number of columns of the dense matrix.

  • nnz[in] number of non-zero entries in the sparse vector.

  • buffer_size[out] temporary storage buffer size.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n, or nnz is invalid.

  • rocsparse_status_invalid_pointerbuffer_size pointer is invalid.

  • rocsparse_status_not_implementedtrans != rocsparse_operation_none or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_gemvi()#

rocsparse_status rocsparse_sgemvi(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, const float *alpha, const float *A, rocsparse_int lda, rocsparse_int nnz, const float *x_val, const rocsparse_int *x_ind, const float *beta, float *y, rocsparse_index_base idx_base, void *temp_buffer)#
rocsparse_status rocsparse_dgemvi(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, const double *alpha, const double *A, rocsparse_int lda, rocsparse_int nnz, const double *x_val, const rocsparse_int *x_ind, const double *beta, double *y, rocsparse_index_base idx_base, void *temp_buffer)#
rocsparse_status rocsparse_cgemvi(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, const rocsparse_float_complex *alpha, const rocsparse_float_complex *A, rocsparse_int lda, rocsparse_int nnz, const rocsparse_float_complex *x_val, const rocsparse_int *x_ind, const rocsparse_float_complex *beta, rocsparse_float_complex *y, rocsparse_index_base idx_base, void *temp_buffer)#
rocsparse_status rocsparse_zgemvi(rocsparse_handle handle, rocsparse_operation trans, rocsparse_int m, rocsparse_int n, const rocsparse_double_complex *alpha, const rocsparse_double_complex *A, rocsparse_int lda, rocsparse_int nnz, const rocsparse_double_complex *x_val, const rocsparse_int *x_ind, const rocsparse_double_complex *beta, rocsparse_double_complex *y, rocsparse_index_base idx_base, void *temp_buffer)#

Dense matrix sparse vector multiplication.

rocsparse_gemvi multiplies the scalar \(\alpha\) with a dense \(m \times n\) matrix \(A\) and the sparse vector \(x\) and adds the result to the dense vector \(y\) that is multiplied by the scalar \(\beta\), such that

\[ y := \alpha \cdot op(A) \cdot x + \beta \cdot y, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans == rocsparse_operation_none} \\ A^T, & \text{if trans == rocsparse_operation_transpose} \\ A^H, & \text{if trans == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

rocsparse_gemvi requires a user allocated temporary buffer. Its size is returned by rocsparse_sgemvi_buffer_size(), rocsparse_dgemvi_buffer_size(), rocsparse_cgemvi_buffer_size() or rocsparse_zgemvi_buffer_size().

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Currently, only trans == rocsparse_operation_none is supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans[in] matrix operation type.

  • m[in] number of rows of the dense matrix.

  • n[in] number of columns of the dense matrix.

  • alpha[in] scalar \(\alpha\).

  • A[in] pointer to the dense matrix.

  • lda[in] leading dimension of the dense matrix

  • nnz[in] number of non-zero entries in the sparse vector

  • x_val[in] array of nnz elements containing the values of the sparse vector

  • x_ind[in] array of nnz elements containing the indices of the sparse vector

  • beta[in] scalar \(\beta\).

  • y[inout] array of m elements ( \(op(A) == A\)) or n elements ( \(op(A) == A^T\) or \(op(A) == A^H\)).

  • idx_base[in] rocsparse_index_base_zero or rocsparse_index_base_one.

  • temp_buffer[in] temporary storage buffer

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n, lda or nnz is invalid.

  • rocsparse_status_invalid_pointeralpha, A, x_val, x_ind, beta, y or temp_buffer pointer is invalid.

  • rocsparse_status_not_implementedtrans != rocsparse_operation_none or rocsparse_matrix_type != rocsparse_matrix_type_general.

Sparse Level 3 Functions#

This module holds all sparse level 3 routines.

The sparse level 3 routines describe operations between a matrix in sparse format and multiple vectors in dense format that can also be seen as a dense matrix.

rocsparse_bsrmm()#

rocsparse_status rocsparse_sbsrmm(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int mb, rocsparse_int n, rocsparse_int kb, rocsparse_int nnzb, const float *alpha, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, const float *B, rocsparse_int ldb, const float *beta, float *C, rocsparse_int ldc)#
rocsparse_status rocsparse_dbsrmm(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int mb, rocsparse_int n, rocsparse_int kb, rocsparse_int nnzb, const double *alpha, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, const double *B, rocsparse_int ldb, const double *beta, double *C, rocsparse_int ldc)#
rocsparse_status rocsparse_cbsrmm(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int mb, rocsparse_int n, rocsparse_int kb, rocsparse_int nnzb, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, const rocsparse_float_complex *B, rocsparse_int ldb, const rocsparse_float_complex *beta, rocsparse_float_complex *C, rocsparse_int ldc)#
rocsparse_status rocsparse_zbsrmm(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int mb, rocsparse_int n, rocsparse_int kb, rocsparse_int nnzb, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, const rocsparse_double_complex *B, rocsparse_int ldb, const rocsparse_double_complex *beta, rocsparse_double_complex *C, rocsparse_int ldc)#

Sparse matrix dense matrix multiplication using BSR storage format.

rocsparse_bsrmm multiplies the scalar \(\alpha\) with a sparse \(mb \times kb\) matrix \(A\), defined in BSR storage format, and the dense \(k \times n\) matrix \(B\) (where \(k = block\_dim \times kb\)) and adds the result to the dense \(m \times n\) matrix \(C\) (where \(m = block\_dim \times mb\)) that is multiplied by the scalar \(\beta\), such that

\[ C := \alpha \cdot op(A) \cdot op(B) + \beta \cdot C, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans_A == rocsparse_operation_none} \\ \end{array} \right. \end{split}\]
and
\[\begin{split} op(B) = \left\{ \begin{array}{ll} B, & \text{if trans_B == rocsparse_operation_none} \\ B^T, & \text{if trans_B == rocsparse_operation_transpose} \\ \end{array} \right. \end{split}\]

Example

This example multiplies a BSR matrix with a dense matrix.

//     1 2 0 3 0 0
// A = 0 4 5 0 0 0
//     0 0 0 7 8 0
//     0 0 1 2 4 1

rocsparse_int block_dim = 2;
rocsparse_int mb   = 2;
rocsparse_int kb   = 3;
rocsparse_int nnzb = 4;
rocsparse_direction dir = rocsparse_direction_row;

bsr_row_ptr[mb+1]                 = {0, 2, 4};                                        // device memory
bsr_col_ind[nnzb]                 = {0, 1, 1, 2};                                     // device memory
bsr_val[nnzb*block_dim*block_dim] = {1, 2, 0, 4, 0, 3, 5, 0, 0, 7, 1, 2, 8, 0, 4, 1}; // device memory

// Set dimension n of B
rocsparse_int n = 64;
rocsparse_int m = mb * block_dim;
rocsparse_int k = kb * block_dim;

// Allocate and generate dense matrix B
std::vector<float> hB(k * n);
for(rocsparse_int i = 0; i < k * n; ++i)
{
    hB[i] = static_cast<float>(rand()) / RAND_MAX;
}

// Copy B to the device
float* B;
hipMalloc((void**)&B, sizeof(float) * k * n);
hipMemcpy(B, hB.data(), sizeof(float) * k * n, hipMemcpyHostToDevice);

// alpha and beta
float alpha = 1.0f;
float beta  = 0.0f;

// Allocate memory for the resulting matrix C
float* C;
hipMalloc((void**)&C, sizeof(float) * m * n);

// Perform the matrix multiplication
rocsparse_sbsrmm(handle,
                 dir,
                 rocsparse_operation_none,
                 rocsparse_operation_none,
                 mb,
                 n,
                 kb,
                 nnzb,
                 &alpha,
                 descr,
                 bsr_val,
                 bsr_row_ptr,
                 bsr_col_ind,
                 block_dim,
                 B,
                 k,
                 &beta,
                 C,
                 m);

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Currently, only trans_A == rocsparse_operation_none is supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] the storage format of the blocks. Can be rocsparse_direction_row or rocsparse_direction_column.

  • trans_A[in] matrix \(A\) operation type. Currently, only rocsparse_operation_none is supported.

  • trans_B[in] matrix \(B\) operation type. Currently, only rocsparse_operation_none and rocsparse_operation_transpose are supported.

  • mb[in] number of block rows of the sparse BSR matrix \(A\).

  • n[in] number of columns of the dense matrix \(op(B)\) and \(C\).

  • kb[in] number of block columns of the sparse BSR matrix \(A\).

  • nnzb[in] number of non-zero blocks of the sparse BSR matrix \(A\).

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse BSR matrix \(A\). Currently, only rocsparse_matrix_type_general is supported.

  • bsr_val[in] array of nnzb*block_dim*block_dim elements of the sparse BSR matrix \(A\).

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix \(A\).

  • bsr_col_ind[in] array of nnzb elements containing the block column indices of the sparse BSR matrix \(A\).

  • block_dim[in] size of the blocks in the sparse BSR matrix.

  • B[in] array of dimension \(ldb \times n\) ( \(op(B) == B\)), \(ldb \times k\) otherwise.

  • ldb[in] leading dimension of \(B\), must be at least \(\max{(1, k)}\) ( \( op(B) == B\)) where \(k = block\_dim \times kb\), \(\max{(1, n)}\) otherwise.

  • beta[in] scalar \(\beta\).

  • C[inout] array of dimension \(ldc \times n\).

  • ldc[in] leading dimension of \(C\), must be at least \(\max{(1, m)}\) ( \( op(A) == A\)) where \(m = block\_dim \times mb\), \(\max{(1, k)}\) where \(k = block\_dim \times kb\) otherwise.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, n, kb, nnzb, ldb or ldc is invalid.

  • rocsparse_status_invalid_pointerdescr, alpha, bsr_val, bsr_row_ptr, bsr_col_ind, B, beta or C pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_not_implementedtrans_A != rocsparse_operation_none or trans_B == rocsparse_operation_conjugate_transpose or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_gebsrmm()#

rocsparse_status rocsparse_sgebsrmm(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int mb, rocsparse_int n, rocsparse_int kb, rocsparse_int nnzb, const float *alpha, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const float *B, rocsparse_int ldb, const float *beta, float *C, rocsparse_int ldc)#
rocsparse_status rocsparse_dgebsrmm(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int mb, rocsparse_int n, rocsparse_int kb, rocsparse_int nnzb, const double *alpha, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const double *B, rocsparse_int ldb, const double *beta, double *C, rocsparse_int ldc)#
rocsparse_status rocsparse_cgebsrmm(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int mb, rocsparse_int n, rocsparse_int kb, rocsparse_int nnzb, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const rocsparse_float_complex *B, rocsparse_int ldb, const rocsparse_float_complex *beta, rocsparse_float_complex *C, rocsparse_int ldc)#
rocsparse_status rocsparse_zgebsrmm(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int mb, rocsparse_int n, rocsparse_int kb, rocsparse_int nnzb, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const rocsparse_double_complex *B, rocsparse_int ldb, const rocsparse_double_complex *beta, rocsparse_double_complex *C, rocsparse_int ldc)#

Sparse matrix dense matrix multiplication using GEneral BSR storage format.

rocsparse_gebsrmm multiplies the scalar \(\alpha\) with a sparse \(mb \times kb\) matrix \(A\), defined in GEneral BSR storage format, and the dense \(k \times n\) matrix \(B\) (where \(k = col_block\_dim \times kb\)) and adds the result to the dense \(m \times n\) matrix \(C\) (where \(m = row_block\_dim \times mb\)) that is multiplied by the scalar \(\beta\), such that

\[ C := \alpha \cdot op(A) \cdot op(B) + \beta \cdot C, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans_A == rocsparse_operation_none} \\ \end{array} \right. \end{split}\]
and
\[\begin{split} op(B) = \left\{ \begin{array}{ll} B, & \text{if trans_B == rocsparse_operation_none} \\ B^T, & \text{if trans_B == rocsparse_operation_transpose} \\ \end{array} \right. \end{split}\]

Example

This example multiplies a GEneral BSR matrix with a dense matrix.

//     1 2 0 3 0 0
// A = 0 4 5 0 0 0
//     0 0 0 7 8 0
//     0 0 1 2 4 1

rocsparse_int row_block_dim = 2;
rocsparse_int col_block_dim = 3;
rocsparse_int mb   = 2;
rocsparse_int kb   = 2;
rocsparse_int nnzb = 4;
rocsparse_direction dir = rocsparse_direction_row;

bsr_row_ptr[mb+1]                 = {0, 2, 4};                                        // device memory
bsr_col_ind[nnzb]                 = {0, 1, 0, 1};                                     // device memory
bsr_val[nnzb*row_block_dim*col_block_dim] = {1, 2, 0, 0, 4, 5, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 7, 8, 0, 2, 4, 1}; // device memory

// Set dimension n of B
rocsparse_int n = 64;
rocsparse_int m = mb * row_block_dim;
rocsparse_int k = kb * col_block_dim;

// Allocate and generate dense matrix B
std::vector<float> hB(k * n);
for(rocsparse_int i = 0; i < k * n; ++i)
{
    hB[i] = static_cast<float>(rand()) / RAND_MAX;
}

// Copy B to the device
float* B;
hipMalloc((void**)&B, sizeof(float) * k * n);
hipMemcpy(B, hB.data(), sizeof(float) * k * n, hipMemcpyHostToDevice);

// alpha and beta
float alpha = 1.0f;
float beta  = 0.0f;

// Allocate memory for the resulting matrix C
float* C;
hipMalloc((void**)&C, sizeof(float) * m * n);

// Perform the matrix multiplication
rocsparse_sgebsrmm(handle,
                   dir,
                   rocsparse_operation_none,
                   rocsparse_operation_none,
                   mb,
                   n,
                   kb,
                   nnzb,
                   &alpha,
                   descr,
                   bsr_val,
                   bsr_row_ptr,
                   bsr_col_ind,
                   row_block_dim,
                   col_block_dim,
                   B,
                   k,
                   &beta,
                   C,
                   m);

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Currently, only trans_A == rocsparse_operation_none is supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] the storage format of the blocks. Can be rocsparse_direction_row or rocsparse_direction_column.

  • trans_A[in] matrix \(A\) operation type. Currently, only rocsparse_operation_none is supported.

  • trans_B[in] matrix \(B\) operation type. Currently, only rocsparse_operation_none and rocsparse_operation_transpose are supported.

  • mb[in] number of block rows of the sparse GEneral BSR matrix \(A\).

  • n[in] number of columns of the dense matrix \(op(B)\) and \(C\).

  • kb[in] number of block columns of the sparse GEneral BSR matrix \(A\).

  • nnzb[in] number of non-zero blocks of the sparse GEneral BSR matrix \(A\).

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse GEneral BSR matrix \(A\). Currently, only rocsparse_matrix_type_general is supported.

  • bsr_val[in] array of nnzb*row_block_dim*col_block_dim elements of the sparse GEneral BSR matrix \(A\).

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse GEneral BSR matrix \(A\).

  • bsr_col_ind[in] array of nnzb elements containing the block column indices of the sparse GEneral BSR matrix \(A\).

  • row_block_dim[in] row size of the blocks in the sparse GEneral BSR matrix.

  • col_block_dim[in] column size of the blocks in the sparse GEneral BSR matrix.

  • B[in] array of dimension \(ldb \times n\) ( \(op(B) == B\)), \(ldb \times k\) otherwise.

  • ldb[in] leading dimension of \(B\), must be at least \(\max{(1, k)}\) ( \( op(B) == B\)) where \(k = col\_block\_dim \times kb\), \(\max{(1, n)}\) otherwise.

  • beta[in] scalar \(\beta\).

  • C[inout] array of dimension \(ldc \times n\).

  • ldc[in] leading dimension of \(C\), must be at least \(\max{(1, m)}\) ( \( op(A) == A\)) where \(m = row\_block\_dim \times mb\), \(\max{(1, k)}\) where \(k = col\_block\_dim \times kb\) otherwise.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, n, kb, nnzb, ldb, ldc, row_block_dim or col_block_dim is invalid.

  • rocsparse_status_invalid_pointerdescr, alpha, bsr_val, bsr_row_ptr, bsr_col_ind, B, beta or C pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_not_implementedtrans_A != rocsparse_operation_none or trans_B == rocsparse_operation_conjugate_transpose or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csrmm()#

rocsparse_status rocsparse_scsrmm(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, rocsparse_int nnz, const float *alpha, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const float *B, rocsparse_int ldb, const float *beta, float *C, rocsparse_int ldc)#
rocsparse_status rocsparse_dcsrmm(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, rocsparse_int nnz, const double *alpha, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const double *B, rocsparse_int ldb, const double *beta, double *C, rocsparse_int ldc)#
rocsparse_status rocsparse_ccsrmm(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, rocsparse_int nnz, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_float_complex *B, rocsparse_int ldb, const rocsparse_float_complex *beta, rocsparse_float_complex *C, rocsparse_int ldc)#
rocsparse_status rocsparse_zcsrmm(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, rocsparse_int nnz, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_double_complex *B, rocsparse_int ldb, const rocsparse_double_complex *beta, rocsparse_double_complex *C, rocsparse_int ldc)#

Sparse matrix dense matrix multiplication using CSR storage format.

rocsparse_csrmm multiplies the scalar \(\alpha\) with a sparse \(m \times k\) matrix \(A\), defined in CSR storage format, and the dense \(k \times n\) matrix \(B\) and adds the result to the dense \(m \times n\) matrix \(C\) that is multiplied by the scalar \(\beta\), such that

\[ C := \alpha \cdot op(A) \cdot op(B) + \beta \cdot C, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans_A == rocsparse_operation_none} \\ A^T, & \text{if trans_A == rocsparse_operation_transpose} \\ A^H, & \text{if trans_A == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]
and
\[\begin{split} op(B) = \left\{ \begin{array}{ll} B, & \text{if trans_B == rocsparse_operation_none} \\ B^T, & \text{if trans_B == rocsparse_operation_transpose} \\ B^H, & \text{if trans_B == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

for(i = 0; i < ldc; ++i)
{
    for(j = 0; j < n; ++j)
    {
        C[i][j] = beta * C[i][j];

        for(k = csr_row_ptr[i]; k < csr_row_ptr[i + 1]; ++k)
        {
            C[i][j] += alpha * csr_val[k] * B[csr_col_ind[k]][j];
        }
    }
}

Example

This example multiplies a CSR matrix with a dense matrix.

//     1 2 0 3 0
// A = 0 4 5 0 0
//     6 0 0 7 8

rocsparse_int m   = 3;
rocsparse_int k   = 5;
rocsparse_int nnz = 8;

csr_row_ptr[m+1] = {0, 3, 5, 8};             // device memory
csr_col_ind[nnz] = {0, 1, 3, 1, 2, 0, 3, 4}; // device memory
csr_val[nnz]     = {1, 2, 3, 4, 5, 6, 7, 8}; // device memory

// Set dimension n of B
rocsparse_int n = 64;

// Allocate and generate dense matrix B
std::vector<float> hB(k * n);
for(rocsparse_int i = 0; i < k * n; ++i)
{
    hB[i] = static_cast<float>(rand()) / RAND_MAX;
}

// Copy B to the device
float* B;
hipMalloc((void**)&B, sizeof(float) * k * n);
hipMemcpy(B, hB.data(), sizeof(float) * k * n, hipMemcpyHostToDevice);

// alpha and beta
float alpha = 1.0f;
float beta  = 0.0f;

// Allocate memory for the resulting matrix C
float* C;
hipMalloc((void**)&C, sizeof(float) * m * n);

// Perform the matrix multiplication
rocsparse_scsrmm(handle,
                 rocsparse_operation_none,
                 rocsparse_operation_none,
                 m,
                 n,
                 k,
                 nnz,
                 &alpha,
                 descr,
                 csr_val,
                 csr_row_ptr,
                 csr_col_ind,
                 B,
                 k,
                 &beta,
                 C,
                 m);

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans_A[in] matrix \(A\) operation type.

  • trans_B[in] matrix \(B\) operation type.

  • m[in] number of rows of the sparse CSR matrix \(A\).

  • n[in] number of columns of the dense matrix \(op(B)\) and \(C\).

  • k[in] number of columns of the sparse CSR matrix \(A\).

  • nnz[in] number of non-zero entries of the sparse CSR matrix \(A\).

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse CSR matrix \(A\). Currently, only rocsparse_matrix_type_general is supported.

  • csr_val[in] array of nnz elements of the sparse CSR matrix \(A\).

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(A\).

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix \(A\).

  • B[in] array of dimension \(ldb \times n\) ( \(op(B) == B\)), \(ldb \times k\) otherwise.

  • ldb[in] leading dimension of \(B\), must be at least \(\max{(1, k)}\) ( \(op(B) == B\)), \(\max{(1, n)}\) otherwise.

  • beta[in] scalar \(\beta\).

  • C[inout] array of dimension \(ldc \times n\).

  • ldc[in] leading dimension of \(C\), must be at least \(\max{(1, m)}\) ( \(op(A) == A\)), \(\max{(1, k)}\) otherwise.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n, k, nnz, ldb or ldc is invalid.

  • rocsparse_status_invalid_pointerdescr, alpha, csr_val, csr_row_ptr, csr_col_ind, B, beta or C pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csrsm_zero_pivot()#

rocsparse_status rocsparse_csrsm_zero_pivot(rocsparse_handle handle, rocsparse_mat_info info, rocsparse_int *position)#

Sparse triangular system solve using CSR storage format.

rocsparse_csrsm_zero_pivot returns rocsparse_status_zero_pivot, if either a structural or numerical zero has been found during rocsparse_scsrsm_solve(), rocsparse_dcsrsm_solve(), rocsparse_ccsrsm_solve() or rocsparse_zcsrsm_solve() computation. The first zero pivot \(j\) at \(A_{j,j}\) is stored in position, using same index base as the CSR matrix.

position can be in host or device memory. If no zero pivot has been found, position is set to -1 and rocsparse_status_success is returned instead.

Note

rocsparse_csrsm_zero_pivot is a blocking function. It might influence performance negatively.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[in] structure that holds the information collected during the analysis step.

  • position[inout] pointer to zero pivot \(j\), can be in host or device memory.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo or position pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_zero_pivot – zero pivot has been found.

rocsparse_csrsm_buffer_size()#

rocsparse_status rocsparse_scsrsm_buffer_size(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int nrhs, rocsparse_int nnz, const float *alpha, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const float *B, rocsparse_int ldb, rocsparse_mat_info info, rocsparse_solve_policy policy, size_t *buffer_size)#
rocsparse_status rocsparse_dcsrsm_buffer_size(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int nrhs, rocsparse_int nnz, const double *alpha, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const double *B, rocsparse_int ldb, rocsparse_mat_info info, rocsparse_solve_policy policy, size_t *buffer_size)#
rocsparse_status rocsparse_ccsrsm_buffer_size(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int nrhs, rocsparse_int nnz, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_float_complex *B, rocsparse_int ldb, rocsparse_mat_info info, rocsparse_solve_policy policy, size_t *buffer_size)#
rocsparse_status rocsparse_zcsrsm_buffer_size(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int nrhs, rocsparse_int nnz, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_double_complex *B, rocsparse_int ldb, rocsparse_mat_info info, rocsparse_solve_policy policy, size_t *buffer_size)#

Sparse triangular system solve using CSR storage format.

rocsparse_csrsm_buffer_size returns the size of the temporary storage buffer that is required by rocsparse_scsrsm_analysis(), rocsparse_dcsrsm_analysis(), rocsparse_ccsrsm_analysis(), rocsparse_zcsrsm_analysis(), rocsparse_scsrsm_solve(), rocsparse_dcsrsm_solve(), rocsparse_ccsrsm_solve() and rocsparse_zcsrsm_solve(). The temporary storage buffer must be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans_A[in] matrix A operation type.

  • trans_B[in] matrix B operation type.

  • m[in] number of rows of the sparse CSR matrix A.

  • nrhs[in] number of columns of the dense matrix op(B).

  • nnz[in] number of non-zero entries of the sparse CSR matrix A.

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse CSR matrix A.

  • csr_val[in] array of nnz elements of the sparse CSR matrix A.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix A.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix A.

  • B[in] array of m \(\times\) nrhs elements of the rhs matrix B.

  • ldb[in] leading dimension of rhs matrix B.

  • info[in] structure that holds the information collected during the analysis step.

  • policy[in] rocsparse_solve_policy_auto.

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_scsrsm_analysis(), rocsparse_dcsrsm_analysis(), rocsparse_ccsrsm_analysis(), rocsparse_zcsrsm_analysis(), rocsparse_scsrsm_solve(), rocsparse_dcsrsm_solve(), rocsparse_ccsrsm_solve() and rocsparse_zcsrsm_solve().

Return values:

rocsparse_csrsm_analysis()#

rocsparse_status rocsparse_scsrsm_analysis(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int nrhs, rocsparse_int nnz, const float *alpha, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const float *B, rocsparse_int ldb, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_dcsrsm_analysis(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int nrhs, rocsparse_int nnz, const double *alpha, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const double *B, rocsparse_int ldb, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_ccsrsm_analysis(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int nrhs, rocsparse_int nnz, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_float_complex *B, rocsparse_int ldb, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_zcsrsm_analysis(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int nrhs, rocsparse_int nnz, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_double_complex *B, rocsparse_int ldb, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#

Sparse triangular system solve using CSR storage format.

rocsparse_csrsm_analysis performs the analysis step for rocsparse_scsrsm_solve(), rocsparse_dcsrsm_solve(), rocsparse_ccsrsm_solve() and rocsparse_zcsrsm_solve(). It is expected that this function will be executed only once for a given matrix and particular operation type. The analysis meta data can be cleared by rocsparse_csrsm_clear().

rocsparse_csrsm_analysis can share its meta data with rocsparse_scsrilu0_analysis(), rocsparse_dcsrilu0_analysis(), rocsparse_ccsrilu0_analysis(), rocsparse_zcsrilu0_analysis(), rocsparse_scsric0_analysis(), rocsparse_dcsric0_analysis(), rocsparse_ccsric0_analysis(), rocsparse_zcsric0_analysis(), rocsparse_scsrsv_analysis(), rocsparse_dcsrsv_analysis(), rocsparse_ccsrsv_analysis() and rocsparse_zcsrsv_analysis(). Selecting rocsparse_analysis_policy_reuse policy can greatly improve computation performance of meta data. However, the user need to make sure that the sparsity pattern remains unchanged. If this cannot be assured, rocsparse_analysis_policy_force has to be used.

Note

If the matrix sparsity pattern changes, the gathered information will become invalid.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans_A[in] matrix A operation type.

  • trans_B[in] matrix B operation type.

  • m[in] number of rows of the sparse CSR matrix A.

  • nrhs[in] number of columns of the dense matrix op(B).

  • nnz[in] number of non-zero entries of the sparse CSR matrix A.

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse CSR matrix A.

  • csr_val[in] array of nnz elements of the sparse CSR matrix A.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix A.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix A.

  • B[in] array of m \(\times\) nrhs elements of the rhs matrix B.

  • ldb[in] leading dimension of rhs matrix B.

  • info[out] structure that holds the information collected during the analysis step.

  • analysis[in] rocsparse_analysis_policy_reuse or rocsparse_analysis_policy_force.

  • solve[in] rocsparse_solve_policy_auto.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:

rocsparse_csrsm_solve()#

rocsparse_status rocsparse_scsrsm_solve(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int nrhs, rocsparse_int nnz, const float *alpha, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, float *B, rocsparse_int ldb, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_dcsrsm_solve(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int nrhs, rocsparse_int nnz, const double *alpha, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, double *B, rocsparse_int ldb, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_ccsrsm_solve(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int nrhs, rocsparse_int nnz, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_float_complex *B, rocsparse_int ldb, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_zcsrsm_solve(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int nrhs, rocsparse_int nnz, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_double_complex *B, rocsparse_int ldb, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#

Sparse triangular system solve using CSR storage format.

rocsparse_csrsm_solve solves a sparse triangular linear system of a sparse \(m \times m\) matrix, defined in CSR storage format, a dense solution matrix \(X\) and the right-hand side matrix \(B\) that is multiplied by \(\alpha\), such that

\[ op(A) \cdot op(X) = \alpha \cdot op(B), \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans_A == rocsparse_operation_none} \\ A^T, & \text{if trans_A == rocsparse_operation_transpose} \\ A^H, & \text{if trans_A == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]
,
\[\begin{split} op(B) = \left\{ \begin{array}{ll} B, & \text{if trans_B == rocsparse_operation_none} \\ B^T, & \text{if trans_B == rocsparse_operation_transpose} \\ B^H, & \text{if trans_B == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]
and
\[\begin{split} op(X) = \left\{ \begin{array}{ll} X, & \text{if trans_B == rocsparse_operation_none} \\ X^T, & \text{if trans_B == rocsparse_operation_transpose} \\ X^H, & \text{if trans_B == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

rocsparse_csrsm_solve requires a user allocated temporary buffer. Its size is returned by rocsparse_scsrsm_buffer_size(), rocsparse_dcsrsm_buffer_size(), rocsparse_ccsrsm_buffer_size() or rocsparse_zcsrsm_buffer_size(). Furthermore, analysis meta data is required. It can be obtained by rocsparse_scsrsm_analysis(), rocsparse_dcsrsm_analysis(), rocsparse_ccsrsm_analysis() or rocsparse_zcsrsm_analysis(). rocsparse_csrsm_solve reports the first zero pivot (either numerical or structural zero). The zero pivot status can be checked calling rocsparse_csrsm_zero_pivot(). If rocsparse_diag_type == rocsparse_diag_type_unit, no zero pivot will be reported, even if \(A_{j,j} = 0\) for some \(j\).

Example

Consider the lower triangular \(m \times m\) matrix \(L\), stored in CSR storage format with unit diagonal. The following example solves \(L \cdot X = B\).

// Create rocSPARSE handle
rocsparse_handle handle;
rocsparse_create_handle(&handle);

// Create matrix descriptor
rocsparse_mat_descr descr;
rocsparse_create_mat_descr(&descr);
rocsparse_set_mat_fill_mode(descr, rocsparse_fill_mode_lower);
rocsparse_set_mat_diag_type(descr, rocsparse_diag_type_unit);

// Create matrix info structure
rocsparse_mat_info info;
rocsparse_create_mat_info(&info);

// Obtain required buffer size
size_t buffer_size;
rocsparse_dcsrsm_buffer_size(handle,
                             rocsparse_operation_none,
                             rocsparse_operation_none,
                             m,
                             nrhs,
                             nnz,
                             &alpha,
                             descr,
                             csr_val,
                             csr_row_ptr,
                             csr_col_ind,
                             B,
                             ldb,
                             info,
                             rocsparse_solve_policy_auto,
                             &buffer_size);

// Allocate temporary buffer
void* temp_buffer;
hipMalloc(&temp_buffer, buffer_size);

// Perform analysis step
rocsparse_dcsrsm_analysis(handle,
                          rocsparse_operation_none,
                          rocsparse_operation_none,
                          m,
                          nrhs,
                          nnz,
                          &alpha,
                          descr,
                          csr_val,
                          csr_row_ptr,
                          csr_col_ind,
                          B,
                          ldb,
                          info,
                          rocsparse_analysis_policy_reuse,
                          rocsparse_solve_policy_auto,
                          temp_buffer);

// Solve LX = B
rocsparse_dcsrsm_solve(handle,
                       rocsparse_operation_none,
                       rocsparse_operation_none,
                       m,
                       nrhs,
                       nnz,
                       &alpha,
                       descr,
                       csr_val,
                       csr_row_ptr,
                       csr_col_ind,
                       B,
                       ldb,
                       info,
                       rocsparse_solve_policy_auto,
                       temp_buffer);

// No zero pivot should be found, with L having unit diagonal

// Clean up
hipFree(temp_buffer);
rocsparse_destroy_mat_info(info);
rocsparse_destroy_mat_descr(descr);
rocsparse_destroy_handle(handle);

Note

The sparse CSR matrix has to be sorted. This can be achieved by calling rocsparse_csrsort().

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Currently, only trans_A != rocsparse_operation_conjugate_transpose and trans_B != rocsparse_operation_conjugate_transpose is supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans_A[in] matrix A operation type.

  • trans_B[in] matrix B operation type.

  • m[in] number of rows of the sparse CSR matrix A.

  • nrhs[in] number of columns of the dense matrix op(B).

  • nnz[in] number of non-zero entries of the sparse CSR matrix A.

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse CSR matrix A.

  • csr_val[in] array of nnz elements of the sparse CSR matrix A.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix A.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix A.

  • B[inout] array of m \(\times\) nrhs elements of the rhs matrix B.

  • ldb[in] leading dimension of rhs matrix B.

  • info[in] structure that holds the information collected during the analysis step.

  • policy[in] rocsparse_solve_policy_auto.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:

rocsparse_csrsm_clear()#

rocsparse_status rocsparse_csrsm_clear(rocsparse_handle handle, rocsparse_mat_info info)#

Sparse triangular system solve using CSR storage format.

rocsparse_csrsm_clear deallocates all memory that was allocated by rocsparse_scsrsm_analysis(), rocsparse_dcsrsm_analysis(), rocsparse_ccsrsm_analysis() or rocsparse_zcsrsm_analysis(). This is especially useful, if memory is an issue and the analysis data is not required for further computation, e.g. when switching to another sparse matrix format. Calling rocsparse_csrsm_clear is optional. All allocated resources will be cleared, when the opaque rocsparse_mat_info struct is destroyed using rocsparse_destroy_mat_info().

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[inout] structure that holds the information collected during the analysis step.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo pointer is invalid.

  • rocsparse_status_memory_error – the buffer holding the meta data could not be deallocated.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_bsrsm_zero_pivot()#

rocsparse_status rocsparse_bsrsm_zero_pivot(rocsparse_handle handle, rocsparse_mat_info info, rocsparse_int *position)#

Sparse triangular system solve using BSR storage format.

rocsparse_bsrsm_zero_pivot returns rocsparse_status_zero_pivot, if either a structural or numerical zero has been found during rocsparse_sbsrsm_solve(), rocsparse_dbsrsm_solve(), rocsparse_cbsrsm_solve() or rocsparse_zbsrsm_solve() computation. The first zero pivot \(j\) at \(A_{j,j}\) is stored in position, using same index base as the BSR matrix.

position can be in host or device memory. If no zero pivot has been found, position is set to -1 and rocsparse_status_success is returned instead.

Note

rocsparse_bsrsm_zero_pivot is a blocking function. It might influence performance negatively.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[in] structure that holds the information collected during the analysis step.

  • position[inout] pointer to zero pivot \(j\), can be in host or device memory.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo or position pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_zero_pivot – zero pivot has been found.

rocsparse_bsrsm_buffer_size()#

rocsparse_status rocsparse_sbsrsm_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_X, rocsparse_int mb, rocsparse_int nrhs, rocsparse_int nnzb, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_dbsrsm_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_X, rocsparse_int mb, rocsparse_int nrhs, rocsparse_int nnzb, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_cbsrsm_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_X, rocsparse_int mb, rocsparse_int nrhs, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_zbsrsm_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_X, rocsparse_int mb, rocsparse_int nrhs, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, size_t *buffer_size)#

Sparse triangular system solve using BSR storage format.

rocsparse_bsrsm_buffer_size returns the size of the temporary storage buffer that is required by rocsparse_sbsrsm_analysis(), rocsparse_dbsrsm_analysis(), rocsparse_cbsrsm_analysis(), rocsparse_zbsrsm_analysis(), rocsparse_sbsrsm_solve(), rocsparse_dbsrsm_solve(), rocsparse_cbsrsm_solve() and rocsparse_zbsrsm_solve(). The temporary storage buffer must be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] matrix storage of BSR blocks.

  • trans_A[in] matrix A operation type.

  • trans_X[in] matrix X operation type.

  • mb[in] number of block rows of the sparse BSR matrix A.

  • nrhs[in] number of columns of the dense matrix op(X).

  • nnzb[in] number of non-zero blocks of the sparse BSR matrix A.

  • descr[in] descriptor of the sparse BSR matrix A.

  • bsr_val[in] array of nnzb blocks of the sparse BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnzb containing the block column indices of the sparse BSR matrix.

  • block_dim[in] block dimension of the sparse BSR matrix.

  • info[in] structure that holds the information collected during the analysis step.

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_sbsrsm_analysis(), rocsparse_dbsrsm_analysis(), rocsparse_cbsrsm_analysis(), rocsparse_zbsrsm_analysis(), rocsparse_sbsrsm_solve(), rocsparse_dbsrsm_solve(), rocsparse_cbsrsm_solve() and rocsparse_zbsrsm_solve().

Return values:

rocsparse_bsrsm_analysis()#

rocsparse_status rocsparse_sbsrsm_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_X, rocsparse_int mb, rocsparse_int nrhs, rocsparse_int nnzb, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_dbsrsm_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_X, rocsparse_int mb, rocsparse_int nrhs, rocsparse_int nnzb, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_cbsrsm_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_X, rocsparse_int mb, rocsparse_int nrhs, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_zbsrsm_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_X, rocsparse_int mb, rocsparse_int nrhs, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#

Sparse triangular system solve using BSR storage format.

rocsparse_bsrsm_analysis performs the analysis step for rocsparse_sbsrsm_solve(), rocsparse_dbsrsm_solve(), rocsparse_cbsrsm_solve() and rocsparse_zbsrsm_solve(). It is expected that this function will be executed only once for a given matrix and particular operation type. The analysis meta data can be cleared by rocsparse_bsrsm_clear().

rocsparse_bsrsm_analysis can share its meta data with rocsparse_sbsrilu0_analysis(), rocsparse_dbsrilu0_analysis(), rocsparse_cbsrilu0_analysis(), rocsparse_zbsrilu0_analysis(), rocsparse_sbsric0_analysis(), rocsparse_dbsric0_analysis(), rocsparse_cbsric0_analysis(), rocsparse_zbsric0_analysis(), rocsparse_sbsrsv_analysis(), rocsparse_dbsrsv_analysis(), rocsparse_cbsrsv_analysis() and rocsparse_zbsrsv_analysis(). Selecting rocsparse_analysis_policy_reuse policy can greatly improve computation performance of meta data. However, the user need to make sure that the sparsity pattern remains unchanged. If this cannot be assured, rocsparse_analysis_policy_force has to be used.

Note

If the matrix sparsity pattern changes, the gathered information will become invalid.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] matrix storage of BSR blocks.

  • trans_A[in] matrix A operation type.

  • trans_X[in] matrix X operation type.

  • mb[in] number of block rows of the sparse BSR matrix A.

  • nrhs[in] number of columns of the dense matrix op(X).

  • nnzb[in] number of non-zero blocks of the sparse BSR matrix A.

  • descr[in] descriptor of the sparse BSR matrix A.

  • bsr_val[in] array of nnzb blocks of the sparse BSR matrix A.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix A.

  • bsr_col_ind[in] array of nnzb containing the block column indices of the sparse BSR matrix A.

  • block_dim[in] block dimension of the sparse BSR matrix A.

  • info[out] structure that holds the information collected during the analysis step.

  • analysis[in] rocsparse_analysis_policy_reuse or rocsparse_analysis_policy_force.

  • solve[in] rocsparse_solve_policy_auto.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:

rocsparse_bsrsm_solve()#

rocsparse_status rocsparse_sbsrsm_solve(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_X, rocsparse_int mb, rocsparse_int nrhs, rocsparse_int nnzb, const float *alpha, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, const float *B, rocsparse_int ldb, float *X, rocsparse_int ldx, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_dbsrsm_solve(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_X, rocsparse_int mb, rocsparse_int nrhs, rocsparse_int nnzb, const double *alpha, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, const double *B, rocsparse_int ldb, double *X, rocsparse_int ldx, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_cbsrsm_solve(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_X, rocsparse_int mb, rocsparse_int nrhs, rocsparse_int nnzb, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, const rocsparse_float_complex *B, rocsparse_int ldb, rocsparse_float_complex *X, rocsparse_int ldx, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_zbsrsm_solve(rocsparse_handle handle, rocsparse_direction dir, rocsparse_operation trans_A, rocsparse_operation trans_X, rocsparse_int mb, rocsparse_int nrhs, rocsparse_int nnzb, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, const rocsparse_double_complex *B, rocsparse_int ldb, rocsparse_double_complex *X, rocsparse_int ldx, rocsparse_solve_policy policy, void *temp_buffer)#

Sparse triangular system solve using BSR storage format.

rocsparse_bsrsm_solve solves a sparse triangular linear system of a sparse \(m \times m\) matrix, defined in BSR storage format, a dense solution matrix \(X\) and the right-hand side matrix \(B\) that is multiplied by \(\alpha\), such that

\[ op(A) \cdot op(X) = \alpha \cdot op(B), \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans_A == rocsparse_operation_none} \\ A^T, & \text{if trans_A == rocsparse_operation_transpose} \\ A^H, & \text{if trans_A == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]
,
\[\begin{split} op(X) = \left\{ \begin{array}{ll} X, & \text{if trans_X == rocsparse_operation_none} \\ X^T, & \text{if trans_X == rocsparse_operation_transpose} \\ X^H, & \text{if trans_X == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

rocsparse_bsrsm_solve requires a user allocated temporary buffer. Its size is returned by rocsparse_sbsrsm_buffer_size(), rocsparse_dbsrsm_buffer_size(), rocsparse_cbsrsm_buffer_size() or rocsparse_zbsrsm_buffer_size(). Furthermore, analysis meta data is required. It can be obtained by rocsparse_sbsrsm_analysis(), rocsparse_dbsrsm_analysis(), rocsparse_cbsrsm_analysis() or rocsparse_zbsrsm_analysis(). rocsparse_bsrsm_solve reports the first zero pivot (either numerical or structural zero). The zero pivot status can be checked calling rocsparse_bsrsm_zero_pivot(). If rocsparse_diag_type == rocsparse_diag_type_unit, no zero pivot will be reported, even if \(A_{j,j} = 0\) for some \(j\).

Note

The sparse BSR matrix has to be sorted.

Note

Operation type of B and X must match, if \(op(B)=B, op(X)=X\).

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Currently, only trans_A != rocsparse_operation_conjugate_transpose and trans_X != rocsparse_operation_conjugate_transpose is supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] matrix storage of BSR blocks.

  • trans_A[in] matrix A operation type.

  • trans_X[in] matrix X operation type.

  • mb[in] number of block rows of the sparse BSR matrix A.

  • nrhs[in] number of columns of the dense matrix op(X).

  • nnzb[in] number of non-zero blocks of the sparse BSR matrix A.

  • alpha[in] scalar \(\alpha\).

  • descr[in] descriptor of the sparse BSR matrix A.

  • bsr_val[in] array of nnzb blocks of the sparse BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnzb containing the block column indices of the sparse BSR matrix.

  • block_dim[in] block dimension of the sparse BSR matrix.

  • info[in] structure that holds the information collected during the analysis step.

  • B[in] rhs matrix B with leading dimension ldb.

  • ldb[in] leading dimension of rhs matrix B.

  • X[out] solution matrix X with leading dimension ldx.

  • ldx[in] leading dimension of solution matrix X.

  • policy[in] rocsparse_solve_policy_auto.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:

rocsparse_bsrsm_clear()#

rocsparse_status rocsparse_bsrsm_clear(rocsparse_handle handle, rocsparse_mat_info info)#

Sparse triangular system solve using BSR storage format.

rocsparse_bsrsm_clear deallocates all memory that was allocated by rocsparse_sbsrsm_analysis(), rocsparse_dbsrsm_analysis(), rocsparse_cbsrsm_analysis() or rocsparse_zbsrsm_analysis(). This is especially useful, if memory is an issue and the analysis data is not required for further computation, e.g. when switching to another sparse matrix format. Calling rocsparse_bsrsm_clear is optional. All allocated resources will be cleared, when the opaque rocsparse_mat_info struct is destroyed using rocsparse_destroy_mat_info().

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[inout] structure that holds the information collected during the analysis step.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo pointer is invalid.

  • rocsparse_status_memory_error – the buffer holding the meta data could not be deallocated.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_gemmi()#

rocsparse_status rocsparse_sgemmi(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, rocsparse_int nnz, const float *alpha, const float *A, rocsparse_int lda, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const float *beta, float *C, rocsparse_int ldc)#
rocsparse_status rocsparse_dgemmi(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, rocsparse_int nnz, const double *alpha, const double *A, rocsparse_int lda, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const double *beta, double *C, rocsparse_int ldc)#
rocsparse_status rocsparse_cgemmi(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, rocsparse_int nnz, const rocsparse_float_complex *alpha, const rocsparse_float_complex *A, rocsparse_int lda, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_float_complex *beta, rocsparse_float_complex *C, rocsparse_int ldc)#
rocsparse_status rocsparse_zgemmi(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, rocsparse_int nnz, const rocsparse_double_complex *alpha, const rocsparse_double_complex *A, rocsparse_int lda, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_double_complex *beta, rocsparse_double_complex *C, rocsparse_int ldc)#

Dense matrix sparse matrix multiplication using CSR storage format.

rocsparse_gemmi multiplies the scalar \(\alpha\) with a dense \(m \times k\) matrix \(A\) and the sparse \(k \times n\) matrix \(B\), defined in CSR storage format and adds the result to the dense \(m \times n\) matrix \(C\) that is multiplied by the scalar \(\beta\), such that

\[ C := \alpha \cdot op(A) \cdot op(B) + \beta \cdot C \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans_A == rocsparse_operation_none} \\ A^T, & \text{if trans_A == rocsparse_operation_transpose} \\ A^H, & \text{if trans_A == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]
and
\[\begin{split} op(B) = \left\{ \begin{array}{ll} B, & \text{if trans_B == rocsparse_operation_none} \\ B^T, & \text{if trans_B == rocsparse_operation_transpose} \\ B^H, & \text{if trans_B == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

Example

This example multiplies a dense matrix with a CSC matrix.

rocsparse_int m   = 2;
rocsparse_int n   = 5;
rocsparse_int k   = 3;
rocsparse_int nnz = 8;
rocsparse_int lda = m;
rocsparse_int ldc = m;

// Matrix A (m x k)
// (  9.0  10.0  11.0 )
// ( 12.0  13.0  14.0 )

// Matrix B (k x n)
// ( 1.0  2.0  0.0  3.0  0.0 )
// ( 0.0  4.0  5.0  0.0  0.0 )
// ( 6.0  0.0  0.0  7.0  8.0 )

// Matrix C (m x n)
// ( 15.0  16.0  17.0  18.0  19.0 )
// ( 20.0  21.0  22.0  23.0  24.0 )

A[lda * k]           = {9.0, 12.0, 10.0, 13.0, 11.0, 14.0};      // device memory
csc_col_ptr_B[n + 1] = {0, 2, 4, 5, 7, 8};                       // device memory
csc_row_ind_B[nnz]   = {0, 0, 1, 1, 2, 3, 3, 4};                 // device memory
csc_val_B[nnz]       = {1.0, 6.0, 2.0, 4.0, 5.0, 3.0, 7.0, 8.0}; // device memory
C[ldc * n]           = {15.0, 20.0, 16.0, 21.0, 17.0, 22.0,      // device memory
                        18.0, 23.0, 19.0, 24.0};

// alpha and beta
float alpha = 1.0f;
float beta  = 0.0f;

// Perform the matrix multiplication
rocsparse_sgemmi(handle,
                 rocsparse_operation_none,
                 rocsparse_operation_transpose,
                 m,
                 n,
                 k,
                 nnz,
                 &alpha,
                 A,
                 lda,
                 descr_B,
                 csc_val_B,
                 csc_col_ptr_B,
                 csc_row_ind_B,
                 &beta,
                 C,
                 ldc);

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans_A[in] matrix \(A\) operation type.

  • trans_B[in] matrix \(B\) operation type.

  • m[in] number of rows of the dense matrix \(A\).

  • n[in] number of columns of the sparse CSR matrix \(op(B)\) and \(C\).

  • k[in] number of columns of the dense matrix \(A\).

  • nnz[in] number of non-zero entries of the sparse CSR matrix \(B\).

  • alpha[in] scalar \(\alpha\).

  • A[in] array of dimension \(lda \times k\) ( \(op(A) == A\)) or \(lda \times m\) ( \(op(A) == A^T\) or \(op(A) == A^H\)).

  • lda[in] leading dimension of \(A\), must be at least \(m\) ( \(op(A) == A\)) or \(k\) ( \(op(A) == A^T\) or \(op(A) == A^H\)).

  • descr[in] descriptor of the sparse CSR matrix \(B\). Currently, only rocsparse_matrix_type_general is supported.

  • csr_val[in] array of nnz elements of the sparse CSR matrix \(B\).

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(B\).

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix \(B\).

  • beta[in] scalar \(\beta\).

  • C[inout] array of dimension \(ldc \times n\) that holds the values of \(C\).

  • ldc[in] leading dimension of \(C\), must be at least \(m\).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n, k, nnz, lda or ldc is invalid.

  • rocsparse_status_invalid_pointeralpha, A, csr_val, csr_row_ptr, csr_col_ind, beta or C pointer is invalid.

Sparse Extra Functions#

This module holds all sparse extra routines.

The sparse extra routines describe operations that manipulate sparse matrices.

rocsparse_csrgeam_nnz()#

rocsparse_status rocsparse_csrgeam_nnz(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const rocsparse_mat_descr descr_C, rocsparse_int *csr_row_ptr_C, rocsparse_int *nnz_C)#

Sparse matrix sparse matrix addition using CSR storage format.

rocsparse_csrgeam_nnz computes the total CSR non-zero elements and the CSR row offsets, that point to the start of every row of the sparse CSR matrix, of the resulting matrix C. It is assumed that csr_row_ptr_C has been allocated with size m+1.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Currently, only rocsparse_matrix_type_general is supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSR matrix \(A\), \(B\) and \(C\).

  • n[in] number of columns of the sparse CSR matrix \(A\), \(B\) and \(C\).

  • descr_A[in] descriptor of the sparse CSR matrix \(A\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_A[in] number of non-zero entries of the sparse CSR matrix \(A\).

  • csr_row_ptr_A[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(A\).

  • csr_col_ind_A[in] array of nnz_A elements containing the column indices of the sparse CSR matrix \(A\).

  • descr_B[in] descriptor of the sparse CSR matrix \(B\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_B[in] number of non-zero entries of the sparse CSR matrix \(B\).

  • csr_row_ptr_B[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(B\).

  • csr_col_ind_B[in] array of nnz_B elements containing the column indices of the sparse CSR matrix \(B\).

  • descr_C[in] descriptor of the sparse CSR matrix \(C\). Currenty, only rocsparse_matrix_type_general is supported.

  • csr_row_ptr_C[out] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(C\).

  • nnz_C[out] pointer to the number of non-zero entries of the sparse CSR matrix \(C\). nnz_C can be a host or device pointer.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n, nnz_A or nnz_B is invalid.

  • rocsparse_status_invalid_pointerdescr_A, csr_row_ptr_A, csr_col_ind_A, descr_B, csr_row_ptr_B, csr_col_ind_B, descr_C, csr_row_ptr_C or nnz_C is invalid.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csrgeam()#

rocsparse_status rocsparse_scsrgeam(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const float *alpha, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const float *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const float *beta, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const float *csr_val_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const rocsparse_mat_descr descr_C, float *csr_val_C, const rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C)#
rocsparse_status rocsparse_dcsrgeam(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const double *alpha, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const double *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const double *beta, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const double *csr_val_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const rocsparse_mat_descr descr_C, double *csr_val_C, const rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C)#
rocsparse_status rocsparse_ccsrgeam(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const rocsparse_float_complex *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_float_complex *beta, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const rocsparse_float_complex *csr_val_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const rocsparse_mat_descr descr_C, rocsparse_float_complex *csr_val_C, const rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C)#
rocsparse_status rocsparse_zcsrgeam(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const rocsparse_double_complex *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_double_complex *beta, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const rocsparse_double_complex *csr_val_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const rocsparse_mat_descr descr_C, rocsparse_double_complex *csr_val_C, const rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C)#

Sparse matrix sparse matrix addition using CSR storage format.

rocsparse_csrgeam multiplies the scalar \(\alpha\) with the sparse \(m \times n\) matrix \(A\), defined in CSR storage format, multiplies the scalar \(\beta\) with the sparse \(m \times n\) matrix \(B\), defined in CSR storage format, and adds both resulting matrices to obtain the sparse \(m \times n\) matrix \(C\), defined in CSR storage format, such that

\[ C := \alpha \cdot A + \beta \cdot B. \]

It is assumed that csr_row_ptr_C has already been filled and that csr_val_C and csr_col_ind_C are allocated by the user. csr_row_ptr_C and allocation size of csr_col_ind_C and csr_val_C is defined by the number of non-zero elements of the sparse CSR matrix C. Both can be obtained by rocsparse_csrgeam_nnz().

Example

This example adds two CSR matrices.

// Initialize scalar multipliers
float alpha = 1.0f;
float beta  = 1.0f;

// Create matrix descriptors
rocsparse_mat_descr descr_A;
rocsparse_mat_descr descr_B;
rocsparse_mat_descr descr_C;

rocsparse_create_mat_descr(&descr_A);
rocsparse_create_mat_descr(&descr_B);
rocsparse_create_mat_descr(&descr_C);

// Set pointer mode
rocsparse_set_pointer_mode(handle, rocsparse_pointer_mode_host);

// Obtain number of total non-zero entries in C and row pointers of C
rocsparse_int nnz_C;
hipMalloc((void**)&csr_row_ptr_C, sizeof(rocsparse_int) * (m + 1));

rocsparse_csrgeam_nnz(handle,
                      m,
                      n,
                      descr_A,
                      nnz_A,
                      csr_row_ptr_A,
                      csr_col_ind_A,
                      descr_B,
                      nnz_B,
                      csr_row_ptr_B,
                      csr_col_ind_B,
                      descr_C,
                      csr_row_ptr_C,
                      &nnz_C);

// Compute column indices and values of C
hipMalloc((void**)&csr_col_ind_C, sizeof(rocsparse_int) * nnz_C);
hipMalloc((void**)&csr_val_C, sizeof(float) * nnz_C);

rocsparse_scsrgeam(handle,
                   m,
                   n,
                   &alpha,
                   descr_A,
                   nnz_A,
                   csr_val_A,
                   csr_row_ptr_A,
                   csr_col_ind_A,
                   &beta,
                   descr_B,
                   nnz_B,
                   csr_val_B,
                   csr_row_ptr_B,
                   csr_col_ind_B,
                   descr_C,
                   csr_val_C,
                   csr_row_ptr_C,
                   csr_col_ind_C);

Note

Both scalars \(\alpha\) and \(beta\) have to be valid.

Note

Currently, only rocsparse_matrix_type_general is supported.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSR matrix \(A\), \(B\) and \(C\).

  • n[in] number of columns of the sparse CSR matrix \(A\), \(B\) and \(C\).

  • alpha[in] scalar \(\alpha\).

  • descr_A[in] descriptor of the sparse CSR matrix \(A\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_A[in] number of non-zero entries of the sparse CSR matrix \(A\).

  • csr_val_A[in] array of nnz_A elements of the sparse CSR matrix \(A\).

  • csr_row_ptr_A[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(A\).

  • csr_col_ind_A[in] array of nnz_A elements containing the column indices of the sparse CSR matrix \(A\).

  • beta[in] scalar \(\beta\).

  • descr_B[in] descriptor of the sparse CSR matrix \(B\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_B[in] number of non-zero entries of the sparse CSR matrix \(B\).

  • csr_val_B[in] array of nnz_B elements of the sparse CSR matrix \(B\).

  • csr_row_ptr_B[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(B\).

  • csr_col_ind_B[in] array of nnz_B elements containing the column indices of the sparse CSR matrix \(B\).

  • descr_C[in] descriptor of the sparse CSR matrix \(C\). Currenty, only rocsparse_matrix_type_general is supported.

  • csr_val_C[out] array of elements of the sparse CSR matrix \(C\).

  • csr_row_ptr_C[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(C\).

  • csr_col_ind_C[out] array of elements containing the column indices of the sparse CSR matrix \(C\).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n, nnz_A or nnz_B is invalid.

  • rocsparse_status_invalid_pointeralpha, descr_A, csr_val_A, csr_row_ptr_A, csr_col_ind_A, beta, descr_B, csr_val_B, csr_row_ptr_B, csr_col_ind_B, descr_C, csr_val_C, csr_row_ptr_C or csr_col_ind_C is invalid.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csrgemm_buffer_size()#

rocsparse_status rocsparse_scsrgemm_buffer_size(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, const float *alpha, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const float *beta, const rocsparse_mat_descr descr_D, rocsparse_int nnz_D, const rocsparse_int *csr_row_ptr_D, const rocsparse_int *csr_col_ind_D, rocsparse_mat_info info_C, size_t *buffer_size)#
rocsparse_status rocsparse_dcsrgemm_buffer_size(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, const double *alpha, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const double *beta, const rocsparse_mat_descr descr_D, rocsparse_int nnz_D, const rocsparse_int *csr_row_ptr_D, const rocsparse_int *csr_col_ind_D, rocsparse_mat_info info_C, size_t *buffer_size)#
rocsparse_status rocsparse_ccsrgemm_buffer_size(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const rocsparse_float_complex *beta, const rocsparse_mat_descr descr_D, rocsparse_int nnz_D, const rocsparse_int *csr_row_ptr_D, const rocsparse_int *csr_col_ind_D, rocsparse_mat_info info_C, size_t *buffer_size)#
rocsparse_status rocsparse_zcsrgemm_buffer_size(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const rocsparse_double_complex *beta, const rocsparse_mat_descr descr_D, rocsparse_int nnz_D, const rocsparse_int *csr_row_ptr_D, const rocsparse_int *csr_col_ind_D, rocsparse_mat_info info_C, size_t *buffer_size)#

Sparse matrix sparse matrix multiplication using CSR storage format.

rocsparse_csrgemm_buffer_size returns the size of the temporary storage buffer that is required by rocsparse_csrgemm_nnz(), rocsparse_scsrgemm(), rocsparse_dcsrgemm(), rocsparse_ccsrgemm() and rocsparse_zcsrgemm(). The temporary storage buffer must be allocated by the user.

Note

Please note, that for matrix products with more than 4096 non-zero entries per row, additional temporary storage buffer is allocated by the algorithm.

Note

Please note, that for matrix products with more than 8192 intermediate products per row, additional temporary storage buffer is allocated by the algorithm.

Note

Currently, only trans_A == trans_B == rocsparse_operation_none is supported.

Note

Currently, only rocsparse_matrix_type_general is supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans_A[in] matrix \(A\) operation type.

  • trans_B[in] matrix \(B\) operation type.

  • m[in] number of rows of the sparse CSR matrix \(op(A)\) and \(C\).

  • n[in] number of columns of the sparse CSR matrix \(op(B)\) and \(C\).

  • k[in] number of columns of the sparse CSR matrix \(op(A)\) and number of rows of the sparse CSR matrix \(op(B)\).

  • alpha[in] scalar \(\alpha\).

  • descr_A[in] descriptor of the sparse CSR matrix \(A\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_A[in] number of non-zero entries of the sparse CSR matrix \(A\).

  • csr_row_ptr_A[in] array of m+1 elements ( \(op(A) == A\), k+1 otherwise) that point to the start of every row of the sparse CSR matrix \(op(A)\).

  • csr_col_ind_A[in] array of nnz_A elements containing the column indices of the sparse CSR matrix \(A\).

  • descr_B[in] descriptor of the sparse CSR matrix \(B\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_B[in] number of non-zero entries of the sparse CSR matrix \(B\).

  • csr_row_ptr_B[in] array of k+1 elements ( \(op(B) == B\), m+1 otherwise) that point to the start of every row of the sparse CSR matrix \(op(B)\).

  • csr_col_ind_B[in] array of nnz_B elements containing the column indices of the sparse CSR matrix \(B\).

  • beta[in] scalar \(\beta\).

  • descr_D[in] descriptor of the sparse CSR matrix \(D\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_D[in] number of non-zero entries of the sparse CSR matrix \(D\).

  • csr_row_ptr_D[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(D\).

  • csr_col_ind_D[in] array of nnz_D elements containing the column indices of the sparse CSR matrix \(D\).

  • info_C[inout] structure that holds meta data for the sparse CSR matrix \(C\).

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_csrgemm_nnz(), rocsparse_scsrgemm(), rocsparse_dcsrgemm(), rocsparse_ccsrgemm() and rocsparse_zcsrgemm().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n, k, nnz_A, nnz_B or nnz_D is invalid.

  • rocsparse_status_invalid_pointeralpha and beta are invalid, descr_A, csr_row_ptr_A, csr_col_ind_A, descr_B, csr_row_ptr_B or csr_col_ind_B are invalid if alpha is valid, descr_D, csr_row_ptr_D or csr_col_ind_D is invalid if beta is valid, info_C or buffer_size is invalid.

  • rocsparse_status_not_implementedtrans_A != rocsparse_operation_none, trans_B != rocsparse_operation_none, or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csrgemm_nnz()#

rocsparse_status rocsparse_csrgemm_nnz(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const rocsparse_mat_descr descr_D, rocsparse_int nnz_D, const rocsparse_int *csr_row_ptr_D, const rocsparse_int *csr_col_ind_D, const rocsparse_mat_descr descr_C, rocsparse_int *csr_row_ptr_C, rocsparse_int *nnz_C, const rocsparse_mat_info info_C, void *temp_buffer)#

Sparse matrix sparse matrix multiplication using CSR storage format.

rocsparse_csrgemm_nnz computes the total CSR non-zero elements and the CSR row offsets, that point to the start of every row of the sparse CSR matrix, of the resulting multiplied matrix C. It is assumed that csr_row_ptr_C has been allocated with size m+1. The required buffer size can be obtained by rocsparse_scsrgemm_buffer_size(), rocsparse_dcsrgemm_buffer_size(), rocsparse_ccsrgemm_buffer_size() and rocsparse_zcsrgemm_buffer_size(), respectively.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Please note, that for matrix products with more than 8192 intermediate products per row, additional temporary storage buffer is allocated by the algorithm.

Note

This function supports unsorted CSR matrices as input, while output will be sorted. Please note that matrices B and D can only be unsorted up to 8192 intermediate products per row. If this number is exceeded, rocsparse_status_requires_sorted_storage will be returned.

Note

Currently, only trans_A == trans_B == rocsparse_operation_none is supported.

Note

Currently, only rocsparse_matrix_type_general is supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans_A[in] matrix \(A\) operation type.

  • trans_B[in] matrix \(B\) operation type.

  • m[in] number of rows of the sparse CSR matrix \(op(A)\) and \(C\).

  • n[in] number of columns of the sparse CSR matrix \(op(B)\) and \(C\).

  • k[in] number of columns of the sparse CSR matrix \(op(A)\) and number of rows of the sparse CSR matrix \(op(B)\).

  • descr_A[in] descriptor of the sparse CSR matrix \(A\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_A[in] number of non-zero entries of the sparse CSR matrix \(A\).

  • csr_row_ptr_A[in] array of m+1 elements ( \(op(A) == A\), k+1 otherwise) that point to the start of every row of the sparse CSR matrix \(op(A)\).

  • csr_col_ind_A[in] array of nnz_A elements containing the column indices of the sparse CSR matrix \(A\).

  • descr_B[in] descriptor of the sparse CSR matrix \(B\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_B[in] number of non-zero entries of the sparse CSR matrix \(B\).

  • csr_row_ptr_B[in] array of k+1 elements ( \(op(B) == B\), m+1 otherwise) that point to the start of every row of the sparse CSR matrix \(op(B)\).

  • csr_col_ind_B[in] array of nnz_B elements containing the column indices of the sparse CSR matrix \(B\).

  • descr_D[in] descriptor of the sparse CSR matrix \(D\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_D[in] number of non-zero entries of the sparse CSR matrix \(D\).

  • csr_row_ptr_D[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(D\).

  • csr_col_ind_D[in] array of nnz_D elements containing the column indices of the sparse CSR matrix \(D\).

  • descr_C[in] descriptor of the sparse CSR matrix \(C\). Currenty, only rocsparse_matrix_type_general is supported.

  • csr_row_ptr_C[out] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(C\).

  • nnz_C[out] pointer to the number of non-zero entries of the sparse CSR matrix \(C\).

  • info_C[in] structure that holds meta data for the sparse CSR matrix \(C\).

  • temp_buffer[in] temporary storage buffer allocated by the user, size is returned by rocsparse_scsrgemm_buffer_size(), rocsparse_dcsrgemm_buffer_size(), rocsparse_ccsrgemm_buffer_size() or rocsparse_zcsrgemm_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n, k, nnz_A, nnz_B or nnz_D is invalid.

  • rocsparse_status_invalid_pointerdescr_A, csr_row_ptr_A, csr_col_ind_A, descr_B, csr_row_ptr_B, csr_col_ind_B, descr_D, csr_row_ptr_D, csr_col_ind_D, descr_C, csr_row_ptr_C, nnz_C, info_C or temp_buffer is invalid.

  • rocsparse_status_memory_error – additional buffer for long rows could not be allocated.

  • rocsparse_status_not_implementedtrans_A != rocsparse_operation_none, trans_B != rocsparse_operation_none, or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csrgemm_symbolic()#

rocsparse_status rocsparse_csrgemm_symbolic(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const rocsparse_mat_descr descr_D, rocsparse_int nnz_D, const rocsparse_int *csr_row_ptr_D, const rocsparse_int *csr_col_ind_D, const rocsparse_mat_descr descr_C, rocsparse_int nnz_C, const rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C, const rocsparse_mat_info info_C, void *temp_buffer)#

Sparse matrix sparse matrix symbolic multiplication using CSR storage format.

rocsparse_csrgemm_symbolic multiplies two sparsity patterns and add an extra one:

\[ opA \cdot op(B) + D \]
with \(m \times k\) matrix \(A\), defined in CSR storage format, the sparse \(k \times n\) matrix \(B\), defined in CSR storage format and the sparse \(m \times n\) matrix \(D\). The * final result is stored in the sparse \(m \times n\) matrix \(C\), defined in CSR storage format, such that
\[ C := op(A) \cdot op(B) + D, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans_A == rocsparse_operation_none} \\ A^T, & \text{if trans_A == rocsparse_operation_transpose} \\ A^H, & \text{if trans_A == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]
and
\[\begin{split} op(B) = \left\{ \begin{array}{ll} B, & \text{if trans_B == rocsparse_operation_none} \\ B^T, & \text{if trans_B == rocsparse_operation_transpose} \\ B^H, & \text{if trans_B == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

It is assumed that csr_row_ptr_C has already been filled and that and csr_col_ind_C is allocated by the user. csr_row_ptr_C and allocation size of csr_col_ind_C is defined by the number of non-zero elements of the sparse CSR matrix C. Both can be obtained by rocsparse_csrgemm_nnz(). The required buffer size for the computation can be obtained by rocsparse_scsrgemm_buffer_size(), rocsparse_dcsrgemm_buffer_size(), rocsparse_ccsrgemm_buffer_size() and rocsparse_zcsrgemm_buffer_size(), respectively.

Example

This example multiplies symbolically two CSR matrices and adds the result to another CSR matrix.

// Initialize scalar multipliers
float alpha = 2.0f;
float beta  = 1.0f;

// Create matrix descriptors
rocsparse_mat_descr descr_A;
rocsparse_mat_descr descr_B;
rocsparse_mat_descr descr_C;
rocsparse_mat_descr descr_D;

rocsparse_create_mat_descr(&descr_A);
rocsparse_create_mat_descr(&descr_B);
rocsparse_create_mat_descr(&descr_C);
rocsparse_create_mat_descr(&descr_D);

// Create matrix info structure
rocsparse_mat_info info_C;
rocsparse_create_mat_info(&info_C);

// Set pointer mode
rocsparse_set_pointer_mode(handle, rocsparse_pointer_mode_host);

// Query rocsparse for the required buffer size
size_t buffer_size;

rocsparse_scsrgemm_buffer_size(handle,
                               rocsparse_operation_none,
                               rocsparse_operation_none,
                               m,
                               n,
                               k,
                               &alpha,
                               descr_A,
                               nnz_A,
                               csr_row_ptr_A,
                               csr_col_ind_A,
                               descr_B,
                               nnz_B,
                               csr_row_ptr_B,
                               csr_col_ind_B,
                               &beta,
                               descr_D,
                               nnz_D,
                               csr_row_ptr_D,
                               csr_col_ind_D,
                               info_C,
                               &buffer_size);

// Allocate buffer
void* buffer;
hipMalloc(&buffer, buffer_size);

// Obtain number of total non-zero entries in C and row pointers of C
rocsparse_int nnz_C;
hipMalloc((void**)&csr_row_ptr_C, sizeof(rocsparse_int) * (m + 1));

rocsparse_csrgemm_nnz(handle,
                      rocsparse_operation_none,
                      rocsparse_operation_none,
                      m,
                      n,
                      k,
                      descr_A,
                      nnz_A,
                      csr_row_ptr_A,
                      csr_col_ind_A,
                      descr_B,
                      nnz_B,
                      csr_row_ptr_B,
                      csr_col_ind_B,
                      descr_D,
                      nnz_D,
                      csr_row_ptr_D,
                      csr_col_ind_D,
                      descr_C,
                      csr_row_ptr_C,
                      &nnz_C,
                      info_C,
                      buffer);

// Compute column indices of C
hipMalloc((void**)&csr_col_ind_C, sizeof(rocsparse_int) * nnz_C);

rocsparse_csrgemm_symbolic(handle,
                   rocsparse_operation_none,
                   rocsparse_operation_none,
                   m,
                   n,
                   k,
                   descr_A,
                   nnz_A,
                   csr_row_ptr_A,
                   csr_col_ind_A,
                   descr_B,
                   nnz_B,
                   csr_row_ptr_B,
                   csr_col_ind_B,
                   descr_D,
                   nnz_D,
                   csr_row_ptr_D,
                   csr_col_ind_D,
                   descr_C,
                   nnz_C,
                   csr_row_ptr_C,
                   csr_col_ind_C,
                   info_C,
                   buffer);

Note

Currently, only trans_A == rocsparse_operation_none is supported.

Note

Currently, only trans_B == rocsparse_operation_none is supported.

Note

Currently, only rocsparse_matrix_type_general is supported.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Please note, that for matrix products with more than 4096 non-zero entries per row, additional temporary storage buffer is allocated by the algorithm.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans_A[in] matrix \(A\) operation type.

  • trans_B[in] matrix \(B\) operation type.

  • m[in] number of rows of the sparse CSR matrix \(op(A)\) and \(C\).

  • n[in] number of columns of the sparse CSR matrix \(op(B)\) and \(C\).

  • k[in] number of columns of the sparse CSR matrix \(op(A)\) and number of rows of the sparse CSR matrix \(op(B)\).

  • descr_A[in] descriptor of the sparse CSR matrix \(A\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_A[in] number of non-zero entries of the sparse CSR matrix \(A\).

  • csr_row_ptr_A[in] array of m+1 elements ( \(op(A) == A\), k+1 otherwise) that point to the start of every row of the sparse CSR matrix \(op(A)\).

  • csr_col_ind_A[in] array of nnz_A elements containing the column indices of the sparse CSR matrix \(A\).

  • descr_B[in] descriptor of the sparse CSR matrix \(B\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_B[in] number of non-zero entries of the sparse CSR matrix \(B\).

  • csr_row_ptr_B[in] array of k+1 elements ( \(op(B) == B\), m+1 otherwise) that point to the start of every row of the sparse CSR matrix \(op(B)\).

  • csr_col_ind_B[in] array of nnz_B elements containing the column indices of the sparse CSR matrix \(B\).

  • descr_D[in] descriptor of the sparse CSR matrix \(D\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_D[in] number of non-zero entries of the sparse CSR matrix \(D\).

  • csr_row_ptr_D[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(D\).

  • csr_col_ind_D[in] array of nnz_D elements containing the column indices of the sparse CSR matrix \(D\).

  • descr_C[in] descriptor of the sparse CSR matrix \(C\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_C[in] number of non-zero entries of the sparse CSR matrix \(C\).

  • csr_row_ptr_C[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(C\).

  • csr_col_ind_C[out] array of nnz_C elements containing the column indices of the sparse CSR matrix \(C\).

  • info_C[in] structure that holds meta data for the sparse CSR matrix \(C\).

  • temp_buffer[in] temporary storage buffer allocated by the user, size is returned by rocsparse_scsrgemm_buffer_size(), rocsparse_dcsrgemm_buffer_size(), rocsparse_ccsrgemm_buffer_size() or rocsparse_zcsrgemm_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n, k, nnz_A, nnz_B or nnz_D is invalid.

  • rocsparse_status_invalid_pointerdescr_A, csr_row_ptr_A, csr_col_ind_A, descr_B, csr_row_ptr_B or csr_col_ind_B, descr_D, csr_row_ptr_D, csr_col_ind_D csr_row_ptr_C, csr_col_ind_C, info_C or temp_buffer is invalid.

  • rocsparse_status_memory_error – additional buffer for long rows could not be allocated.

  • rocsparse_status_not_implementedtrans_A != rocsparse_operation_none, trans_B != rocsparse_operation_none, or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csrgemm()#

rocsparse_status rocsparse_scsrgemm(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, const float *alpha, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const float *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const float *csr_val_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const float *beta, const rocsparse_mat_descr descr_D, rocsparse_int nnz_D, const float *csr_val_D, const rocsparse_int *csr_row_ptr_D, const rocsparse_int *csr_col_ind_D, const rocsparse_mat_descr descr_C, float *csr_val_C, const rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C, const rocsparse_mat_info info_C, void *temp_buffer)#
rocsparse_status rocsparse_dcsrgemm(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, const double *alpha, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const double *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const double *csr_val_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const double *beta, const rocsparse_mat_descr descr_D, rocsparse_int nnz_D, const double *csr_val_D, const rocsparse_int *csr_row_ptr_D, const rocsparse_int *csr_col_ind_D, const rocsparse_mat_descr descr_C, double *csr_val_C, const rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C, const rocsparse_mat_info info_C, void *temp_buffer)#
rocsparse_status rocsparse_ccsrgemm(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const rocsparse_float_complex *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const rocsparse_float_complex *csr_val_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const rocsparse_float_complex *beta, const rocsparse_mat_descr descr_D, rocsparse_int nnz_D, const rocsparse_float_complex *csr_val_D, const rocsparse_int *csr_row_ptr_D, const rocsparse_int *csr_col_ind_D, const rocsparse_mat_descr descr_C, rocsparse_float_complex *csr_val_C, const rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C, const rocsparse_mat_info info_C, void *temp_buffer)#
rocsparse_status rocsparse_zcsrgemm(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const rocsparse_double_complex *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const rocsparse_double_complex *csr_val_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const rocsparse_double_complex *beta, const rocsparse_mat_descr descr_D, rocsparse_int nnz_D, const rocsparse_double_complex *csr_val_D, const rocsparse_int *csr_row_ptr_D, const rocsparse_int *csr_col_ind_D, const rocsparse_mat_descr descr_C, rocsparse_double_complex *csr_val_C, const rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C, const rocsparse_mat_info info_C, void *temp_buffer)#

Sparse matrix sparse matrix multiplication using CSR storage format.

rocsparse_csrgemm multiplies the scalar \(\alpha\) with the sparse \(m \times k\) matrix \(A\), defined in CSR storage format, and the sparse \(k \times n\) matrix \(B\), defined in CSR storage format, and adds the result to the sparse \(m \times n\) matrix \(D\) that is multiplied by \(\beta\). The final result is stored in the sparse \(m \times n\) matrix \(C\), defined in CSR storage format, such that

\[ C := \alpha \cdot op(A) \cdot op(B) + \beta \cdot D, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans_A == rocsparse_operation_none} \\ A^T, & \text{if trans_A == rocsparse_operation_transpose} \\ A^H, & \text{if trans_A == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]
and
\[\begin{split} op(B) = \left\{ \begin{array}{ll} B, & \text{if trans_B == rocsparse_operation_none} \\ B^T, & \text{if trans_B == rocsparse_operation_transpose} \\ B^H, & \text{if trans_B == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

It is assumed that csr_row_ptr_C has already been filled and that csr_val_C and csr_col_ind_C are allocated by the user. csr_row_ptr_C and allocation size of csr_col_ind_C and csr_val_C is defined by the number of non-zero elements of the sparse CSR matrix C. Both can be obtained by rocsparse_csrgemm_nnz(). The required buffer size for the computation can be obtained by rocsparse_scsrgemm_buffer_size(), rocsparse_dcsrgemm_buffer_size(), rocsparse_ccsrgemm_buffer_size() and rocsparse_zcsrgemm_buffer_size(), respectively.

Example

This example multiplies two CSR matrices with a scalar alpha and adds the result to another CSR matrix.

// Initialize scalar multipliers
float alpha = 2.0f;
float beta  = 1.0f;

// Create matrix descriptors
rocsparse_mat_descr descr_A;
rocsparse_mat_descr descr_B;
rocsparse_mat_descr descr_C;
rocsparse_mat_descr descr_D;

rocsparse_create_mat_descr(&descr_A);
rocsparse_create_mat_descr(&descr_B);
rocsparse_create_mat_descr(&descr_C);
rocsparse_create_mat_descr(&descr_D);

// Create matrix info structure
rocsparse_mat_info info_C;
rocsparse_create_mat_info(&info_C);

// Set pointer mode
rocsparse_set_pointer_mode(handle, rocsparse_pointer_mode_host);

// Query rocsparse for the required buffer size
size_t buffer_size;

rocsparse_scsrgemm_buffer_size(handle,
                               rocsparse_operation_none,
                               rocsparse_operation_none,
                               m,
                               n,
                               k,
                               &alpha,
                               descr_A,
                               nnz_A,
                               csr_row_ptr_A,
                               csr_col_ind_A,
                               descr_B,
                               nnz_B,
                               csr_row_ptr_B,
                               csr_col_ind_B,
                               &beta,
                               descr_D,
                               nnz_D,
                               csr_row_ptr_D,
                               csr_col_ind_D,
                               info_C,
                               &buffer_size);

// Allocate buffer
void* buffer;
hipMalloc(&buffer, buffer_size);

// Obtain number of total non-zero entries in C and row pointers of C
rocsparse_int nnz_C;
hipMalloc((void**)&csr_row_ptr_C, sizeof(rocsparse_int) * (m + 1));

rocsparse_csrgemm_nnz(handle,
                      rocsparse_operation_none,
                      rocsparse_operation_none,
                      m,
                      n,
                      k,
                      descr_A,
                      nnz_A,
                      csr_row_ptr_A,
                      csr_col_ind_A,
                      descr_B,
                      nnz_B,
                      csr_row_ptr_B,
                      csr_col_ind_B,
                      descr_D,
                      nnz_D,
                      csr_row_ptr_D,
                      csr_col_ind_D,
                      descr_C,
                      csr_row_ptr_C,
                      &nnz_C,
                      info_C,
                      buffer);

// Compute column indices and values of C
hipMalloc((void**)&csr_col_ind_C, sizeof(rocsparse_int) * nnz_C);
hipMalloc((void**)&csr_val_C, sizeof(float) * nnz_C);

rocsparse_scsrgemm(handle,
                   rocsparse_operation_none,
                   rocsparse_operation_none,
                   m,
                   n,
                   k,
                   &alpha,
                   descr_A,
                   nnz_A,
                   csr_val_A,
                   csr_row_ptr_A,
                   csr_col_ind_A,
                   descr_B,
                   nnz_B,
                   csr_val_B,
                   csr_row_ptr_B,
                   csr_col_ind_B,
                   &beta,
                   descr_D,
                   nnz_D,
                   csr_val_D,
                   csr_row_ptr_D,
                   csr_col_ind_D,
                   descr_C,
                   csr_val_C,
                   csr_row_ptr_C,
                   csr_col_ind_C,
                   info_C,
                   buffer);

Note

If \(\alpha == 0\), then \(C = \beta \cdot D\) will be computed.

Note

If \(\beta == 0\), then \(C = \alpha \cdot op(A) \cdot op(B)\) will be computed.

Note

\(\alpha == beta == 0\) is invalid.

Note

Currently, only trans_A == rocsparse_operation_none is supported.

Note

Currently, only trans_B == rocsparse_operation_none is supported.

Note

Currently, only rocsparse_matrix_type_general is supported.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Please note, that for matrix products with more than 4096 non-zero entries per row, additional temporary storage buffer is allocated by the algorithm.

Note

This function supports unsorted CSR matrices as input, while output will be sorted. Please note that matrices B and D can only be unsorted up to 4096 non-zero entries per row. If this number is exceeded, rocsparse_status_requires_sorted_storage will be returned.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans_A[in] matrix \(A\) operation type.

  • trans_B[in] matrix \(B\) operation type.

  • m[in] number of rows of the sparse CSR matrix \(op(A)\) and \(C\).

  • n[in] number of columns of the sparse CSR matrix \(op(B)\) and \(C\).

  • k[in] number of columns of the sparse CSR matrix \(op(A)\) and number of rows of the sparse CSR matrix \(op(B)\).

  • alpha[in] scalar \(\alpha\).

  • descr_A[in] descriptor of the sparse CSR matrix \(A\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_A[in] number of non-zero entries of the sparse CSR matrix \(A\).

  • csr_val_A[in] array of nnz_A elements of the sparse CSR matrix \(A\).

  • csr_row_ptr_A[in] array of m+1 elements ( \(op(A) == A\), k+1 otherwise) that point to the start of every row of the sparse CSR matrix \(op(A)\).

  • csr_col_ind_A[in] array of nnz_A elements containing the column indices of the sparse CSR matrix \(A\).

  • descr_B[in] descriptor of the sparse CSR matrix \(B\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_B[in] number of non-zero entries of the sparse CSR matrix \(B\).

  • csr_val_B[in] array of nnz_B elements of the sparse CSR matrix \(B\).

  • csr_row_ptr_B[in] array of k+1 elements ( \(op(B) == B\), m+1 otherwise) that point to the start of every row of the sparse CSR matrix \(op(B)\).

  • csr_col_ind_B[in] array of nnz_B elements containing the column indices of the sparse CSR matrix \(B\).

  • beta[in] scalar \(\beta\).

  • descr_D[in] descriptor of the sparse CSR matrix \(D\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_D[in] number of non-zero entries of the sparse CSR matrix \(D\).

  • csr_val_D[in] array of nnz_D elements of the sparse CSR matrix \(D\).

  • csr_row_ptr_D[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(D\).

  • csr_col_ind_D[in] array of nnz_D elements containing the column indices of the sparse CSR matrix \(D\).

  • descr_C[in] descriptor of the sparse CSR matrix \(C\). Currenty, only rocsparse_matrix_type_general is supported.

  • csr_val_C[out] array of nnz_C elements of the sparse CSR matrix \(C\).

  • csr_row_ptr_C[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(C\).

  • csr_col_ind_C[out] array of nnz_C elements containing the column indices of the sparse CSR matrix \(C\).

  • info_C[in] structure that holds meta data for the sparse CSR matrix \(C\).

  • temp_buffer[in] temporary storage buffer allocated by the user, size is returned by rocsparse_scsrgemm_buffer_size(), rocsparse_dcsrgemm_buffer_size(), rocsparse_ccsrgemm_buffer_size() or rocsparse_zcsrgemm_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n, k, nnz_A, nnz_B or nnz_D is invalid.

  • rocsparse_status_invalid_pointeralpha and beta are invalid, descr_A, csr_val_A, csr_row_ptr_A, csr_col_ind_A, descr_B, csr_val_B, csr_row_ptr_B or csr_col_ind_B are invalid if alpha is valid, descr_D, csr_val_D, csr_row_ptr_D or csr_col_ind_D is invalid if beta is valid, csr_val_C, csr_row_ptr_C, csr_col_ind_C, info_C or temp_buffer is invalid.

  • rocsparse_status_memory_error – additional buffer for long rows could not be allocated.

  • rocsparse_status_not_implementedtrans_A != rocsparse_operation_none, trans_B != rocsparse_operation_none, or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csrgemm_numeric()#

rocsparse_status rocsparse_scsrgemm_numeric(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, const float *alpha, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const float *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const float *csr_val_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const float *beta, const rocsparse_mat_descr descr_D, rocsparse_int nnz_D, const float *csr_val_D, const rocsparse_int *csr_row_ptr_D, const rocsparse_int *csr_col_ind_D, const rocsparse_mat_descr descr_C, rocsparse_int nnz_C, float *csr_val_C, const rocsparse_int *csr_row_ptr_C, const rocsparse_int *csr_col_ind_C, const rocsparse_mat_info info_C, void *temp_buffer)#
rocsparse_status rocsparse_dcsrgemm_numeric(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, const double *alpha, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const double *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const double *csr_val_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const double *beta, const rocsparse_mat_descr descr_D, rocsparse_int nnz_D, const double *csr_val_D, const rocsparse_int *csr_row_ptr_D, const rocsparse_int *csr_col_ind_D, const rocsparse_mat_descr descr_C, rocsparse_int nnz_C, double *csr_val_C, const rocsparse_int *csr_row_ptr_C, const rocsparse_int *csr_col_ind_C, const rocsparse_mat_info info_C, void *temp_buffer)#
rocsparse_status rocsparse_ccsrgemm_numeric(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, const rocsparse_float_complex *alpha, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const rocsparse_float_complex *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const rocsparse_float_complex *csr_val_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const rocsparse_float_complex *beta, const rocsparse_mat_descr descr_D, rocsparse_int nnz_D, const rocsparse_float_complex *csr_val_D, const rocsparse_int *csr_row_ptr_D, const rocsparse_int *csr_col_ind_D, const rocsparse_mat_descr descr_C, rocsparse_int nnz_C, rocsparse_float_complex *csr_val_C, const rocsparse_int *csr_row_ptr_C, const rocsparse_int *csr_col_ind_C, const rocsparse_mat_info info_C, void *temp_buffer)#
rocsparse_status rocsparse_zcsrgemm_numeric(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, rocsparse_int m, rocsparse_int n, rocsparse_int k, const rocsparse_double_complex *alpha, const rocsparse_mat_descr descr_A, rocsparse_int nnz_A, const rocsparse_double_complex *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const rocsparse_mat_descr descr_B, rocsparse_int nnz_B, const rocsparse_double_complex *csr_val_B, const rocsparse_int *csr_row_ptr_B, const rocsparse_int *csr_col_ind_B, const rocsparse_double_complex *beta, const rocsparse_mat_descr descr_D, rocsparse_int nnz_D, const rocsparse_double_complex *csr_val_D, const rocsparse_int *csr_row_ptr_D, const rocsparse_int *csr_col_ind_D, const rocsparse_mat_descr descr_C, rocsparse_int nnz_C, rocsparse_double_complex *csr_val_C, const rocsparse_int *csr_row_ptr_C, const rocsparse_int *csr_col_ind_C, const rocsparse_mat_info info_C, void *temp_buffer)#

Sparse matrix sparse matrix numeric multiplication using CSR storage format.

rocsparse_csrgemm_numeric multiplies the scalar \(\alpha\) with the sparse \(m \times k\) matrix \(A\), defined in CSR storage format, and the sparse \(k \times n\) matrix \(B\), defined in CSR storage format, and adds the result to the sparse \(m \times n\) matrix \(D\) that is multiplied by \(\beta\). The final result is stored in the sparse \(m \times n\) matrix \(C\), predefined in CSR storage format, such that

\[ C := \alpha \cdot op(A) \cdot op(B) + \beta \cdot D, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans_A == rocsparse_operation_none} \\ A^T, & \text{if trans_A == rocsparse_operation_transpose} \\ A^H, & \text{if trans_A == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]
and
\[\begin{split} op(B) = \left\{ \begin{array}{ll} B, & \text{if trans_B == rocsparse_operation_none} \\ B^T, & \text{if trans_B == rocsparse_operation_transpose} \\ B^H, & \text{if trans_B == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

The symbolic part of the csr matrix C can be obtained by rocsparse_csrgemm_symbolic(). It is assumed that csr_row_ptr_C and csr_col_ind_C have already been filled and that csr_val_C is allocated by the user. csr_row_ptr_C and allocation size of csr_col_ind_C and csr_val_C is defined by the number of non-zero elements of the sparse CSR matrix C. Both can be obtained by rocsparse_csrgemm_nnz(). The required buffer size for the computation can be obtained by rocsparse_scsrgemm_buffer_size(), rocsparse_dcsrgemm_buffer_size(), rocsparse_ccsrgemm_buffer_size() and rocsparse_zcsrgemm_buffer_size(), respectively.

Example

This example multiplies two CSR matrices with a scalar alpha and adds the result to another CSR matrix.

// Initialize scalar multipliers
float alpha = 2.0f;
float beta  = 1.0f;

// Create matrix descriptors
rocsparse_mat_descr descr_A;
rocsparse_mat_descr descr_B;
rocsparse_mat_descr descr_C;
rocsparse_mat_descr descr_D;

rocsparse_create_mat_descr(&descr_A);
rocsparse_create_mat_descr(&descr_B);
rocsparse_create_mat_descr(&descr_C);
rocsparse_create_mat_descr(&descr_D);

// Create matrix info structure
rocsparse_mat_info info_C;
rocsparse_create_mat_info(&info_C);

// Set pointer mode
rocsparse_set_pointer_mode(handle, rocsparse_pointer_mode_host);

// Query rocsparse for the required buffer size
size_t buffer_size;

rocsparse_scsrgemm_buffer_size(handle,
                               rocsparse_operation_none,
                               rocsparse_operation_none,
                               m,
                               n,
                               k,
                               &alpha,
                               descr_A,
                               nnz_A,
                               csr_row_ptr_A,
                               csr_col_ind_A,
                               descr_B,
                               nnz_B,
                               csr_row_ptr_B,
                               csr_col_ind_B,
                               &beta,
                               descr_D,
                               nnz_D,
                               csr_row_ptr_D,
                               csr_col_ind_D,
                               info_C,
                               &buffer_size);

// Allocate buffer
void* buffer;
hipMalloc(&buffer, buffer_size);

// Obtain number of total non-zero entries in C and row pointers of C
rocsparse_int nnz_C;
hipMalloc((void**)&csr_row_ptr_C, sizeof(rocsparse_int) * (m + 1));

rocsparse_csrgemm_nnz(handle,
                      rocsparse_operation_none,
                      rocsparse_operation_none,
                      m,
                      n,
                      k,
                      descr_A,
                      nnz_A,
                      csr_row_ptr_A,
                      csr_col_ind_A,
                      descr_B,
                      nnz_B,
                      csr_row_ptr_B,
                      csr_col_ind_B,
                      descr_D,
                      nnz_D,
                      csr_row_ptr_D,
                      csr_col_ind_D,
                      descr_C,
                      csr_row_ptr_C,
                      &nnz_C,
                      info_C,
                      buffer);

// Compute column indices and values of C
hipMalloc((void**)&csr_col_ind_C, sizeof(rocsparse_int) * nnz_C);
rocsparse_csrgemm_symbolic(handle,
                   rocsparse_operation_none,
                   rocsparse_operation_none,
                   m,
                   n,
                   k,
                   descr_A,
                   nnz_A,
                   csr_row_ptr_A,
                   csr_col_ind_A,
                   descr_B,
                   nnz_B,
                   csr_row_ptr_B,
                   csr_col_ind_B,
                   descr_D,
                   nnz_D,
                   csr_row_ptr_D,
                   csr_col_ind_D,
                   descr_C,
                   nnz_C,
                   csr_row_ptr_C,
                   csr_col_ind_C,
                   info_C,
                   buffer);
hipMalloc((void**)&csr_val_C, sizeof(float) * nnz_C);

rocsparse_scsrgemm_numeric(handle,
                   rocsparse_operation_none,
                   rocsparse_operation_none,
                   m,
                   n,
                   k,
                   &alpha,
                   descr_A,
                   nnz_A,
                   csr_val_A,
                   csr_row_ptr_A,
                   csr_col_ind_A,
                   descr_B,
                   nnz_B,
                   csr_val_B,
                   csr_row_ptr_B,
                   csr_col_ind_B,
                   &beta,
                   descr_D,
                   nnz_D,
                   csr_val_D,
                   csr_row_ptr_D,
                   csr_col_ind_D,
                   descr_C,
                   nnz_C,
                   csr_val_C,
                   csr_row_ptr_C,
                   csr_col_ind_C,
                   info_C,
                   buffer);

Note

If \(\alpha == 0\), then \(C = \beta \cdot D\) will be computed.

Note

If \(\beta == 0\), then \(C = \alpha \cdot op(A) \cdot op(B)\) will be computed.

Note

\(\alpha == beta == 0\) is invalid.

Note

Currently, only trans_A == rocsparse_operation_none is supported.

Note

Currently, only trans_B == rocsparse_operation_none is supported.

Note

Currently, only rocsparse_matrix_type_general is supported.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Please note, that for matrix products with more than 4096 non-zero entries per row, additional temporary storage buffer is allocated by the algorithm.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans_A[in] matrix \(A\) operation type.

  • trans_B[in] matrix \(B\) operation type.

  • m[in] number of rows of the sparse CSR matrix \(op(A)\) and \(C\).

  • n[in] number of columns of the sparse CSR matrix \(op(B)\) and \(C\).

  • k[in] number of columns of the sparse CSR matrix \(op(A)\) and number of rows of the sparse CSR matrix \(op(B)\).

  • alpha[in] scalar \(\alpha\).

  • descr_A[in] descriptor of the sparse CSR matrix \(A\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_A[in] number of non-zero entries of the sparse CSR matrix \(A\).

  • csr_val_A[in] array of nnz_A elements of the sparse CSR matrix \(A\).

  • csr_row_ptr_A[in] array of m+1 elements ( \(op(A) == A\), k+1 otherwise) that point to the start of every row of the sparse CSR matrix \(op(A)\).

  • csr_col_ind_A[in] array of nnz_A elements containing the column indices of the sparse CSR matrix \(A\).

  • descr_B[in] descriptor of the sparse CSR matrix \(B\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_B[in] number of non-zero entries of the sparse CSR matrix \(B\).

  • csr_val_B[in] array of nnz_B elements of the sparse CSR matrix \(B\).

  • csr_row_ptr_B[in] array of k+1 elements ( \(op(B) == B\), m+1 otherwise) that point to the start of every row of the sparse CSR matrix \(op(B)\).

  • csr_col_ind_B[in] array of nnz_B elements containing the column indices of the sparse CSR matrix \(B\).

  • beta[in] scalar \(\beta\).

  • descr_D[in] descriptor of the sparse CSR matrix \(D\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_D[in] number of non-zero entries of the sparse CSR matrix \(D\).

  • csr_val_D[in] array of nnz_D elements of the sparse CSR matrix \(D\).

  • csr_row_ptr_D[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(D\).

  • csr_col_ind_D[in] array of nnz_D elements containing the column indices of the sparse CSR matrix \(D\).

  • descr_C[in] descriptor of the sparse CSR matrix \(C\). Currenty, only rocsparse_matrix_type_general is supported.

  • nnz_C[in] number of non-zero entries of the sparse CSR matrix \(C\).

  • csr_val_C[out] array of nnz_C elements of the sparse CSR matrix \(C\).

  • csr_row_ptr_C[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix \(C\).

  • csr_col_ind_C[in] array of nnz_C elements containing the column indices of the sparse CSR matrix \(C\).

  • info_C[in] structure that holds meta data for the sparse CSR matrix \(C\).

  • temp_buffer[in] temporary storage buffer allocated by the user, size is returned by rocsparse_scsrgemm_buffer_size(), rocsparse_dcsrgemm_buffer_size(), rocsparse_ccsrgemm_buffer_size() or rocsparse_zcsrgemm_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n, k, nnz_A, nnz_B or nnz_D is invalid.

  • rocsparse_status_invalid_pointeralpha and beta are invalid, descr_A, csr_val_A, csr_row_ptr_A, csr_col_ind_A, descr_B, csr_val_B, csr_row_ptr_B or csr_col_ind_B are invalid if alpha is valid, descr_D, csr_val_D, csr_row_ptr_D or csr_col_ind_D is invalid if beta is valid, csr_val_C, csr_row_ptr_C, csr_col_ind_C, info_C or temp_buffer is invalid.

  • rocsparse_status_memory_error – additional buffer for long rows could not be allocated.

  • rocsparse_status_not_implementedtrans_A != rocsparse_operation_none, trans_B != rocsparse_operation_none, or rocsparse_matrix_type != rocsparse_matrix_type_general.

Preconditioner Functions#

This module holds all sparse preconditioners.

The sparse preconditioners describe manipulations on a matrix in sparse format to obtain a sparse preconditioner matrix.

rocsparse_bsric0_zero_pivot()#

rocsparse_status rocsparse_bsric0_zero_pivot(rocsparse_handle handle, rocsparse_mat_info info, rocsparse_int *position)#

Incomplete Cholesky factorization with 0 fill-ins and no pivoting using BSR storage format.

rocsparse_bsric0_zero_pivot returns rocsparse_status_zero_pivot, if either a structural or numerical zero has been found during rocsparse_sbsric0(), rocsparse_dbsric0(), rocsparse_cbsric0() or rocsparse_zbsric0() computation. The first zero pivot \(j\) at \(A_{j,j}\) is stored in position, using same index base as the BSR matrix.

position can be in host or device memory. If no zero pivot has been found, position is set to -1 and rocsparse_status_success is returned instead.

Note

If a zero pivot is found, position=j means that either the diagonal block A(j,j) is missing (structural zero) or the diagonal block A(j,j) is not positive definite (numerical zero).

Note

rocsparse_bsric0_zero_pivot is a blocking function. It might influence performance negatively.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[in] structure that holds the information collected during the analysis step.

  • position[inout] pointer to zero pivot \(j\), can be in host or device memory.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo or position pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_zero_pivot – zero pivot has been found.

rocsparse_bsric0_buffer_size()#

rocsparse_status rocsparse_sbsric0_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_dbsric0_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_cbsric0_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_zbsric0_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, size_t *buffer_size)#

Incomplete Cholesky factorization with 0 fill-ins and no pivoting using BSR storage format.

rocsparse_bsric0_buffer_size returns the size of the temporary storage buffer that is required by rocsparse_sbsric0_analysis(), rocsparse_dbsric0_analysis(), rocsparse_cbsric0_analysis(), rocsparse_zbsric0_analysis(), rocsparse_sbsric0(), rocsparse_dbsric0(), rocsparse_sbsric0() and rocsparse_dbsric0(). The temporary storage buffer must be allocated by the user. The size of the temporary storage buffer is identical to the size returned by rocsparse_sbsrsv_buffer_size(), rocsparse_dbsrsv_buffer_size(), rocsparse_cbsrsv_buffer_size(), rocsparse_zbsrsv_buffer_size(), rocsparse_sbsrilu0_buffer_size(), rocsparse_dbsrilu0_buffer_size(), rocsparse_cbsrilu0_buffer_size() and rocsparse_zbsrilu0_buffer_size() if the matrix sparsity pattern is identical. The user allocated buffer can thus be shared between subsequent calls to those functions.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] direction that specifies whether to count nonzero elements by rocsparse_direction_row or by rocsparse_direction_row.

  • mb[in] number of block rows in the sparse BSR matrix.

  • nnzb[in] number of non-zero block entries of the sparse BSR matrix.

  • descr[in] descriptor of the sparse BSR matrix.

  • bsr_val[in] array of length nnzb*block_dim*block_dim containing the values of the sparse BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnzb elements containing the block column indices of the sparse BSR matrix.

  • block_dim[in] the block dimension of the BSR matrix. Between 1 and m where m=mb*block_dim.

  • info[out] structure that holds the information collected during the analysis step.

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_sbsric0_analysis(), rocsparse_dbsric0_analysis(), rocsparse_cbsric0_analysis(), rocsparse_zbsric0_analysis(), rocsparse_sbsric0(), rocsparse_dbsric0(), rocsparse_cbsric0() and rocsparse_zbsric0().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, nnzb, or block_dim is invalid.

  • rocsparse_status_invalid_pointerdescr, bsr_val, bsr_row_ptr, bsr_col_ind, info or buffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_bsric0_analysis()#

rocsparse_status rocsparse_sbsric0_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_dbsric0_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_cbsric0_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_zbsric0_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#

Incomplete Cholesky factorization with 0 fill-ins and no pivoting using BSR storage format.

rocsparse_bsric0_analysis performs the analysis step for rocsparse_sbsric0() rocsparse_dbsric0(), rocsparse_cbsric0(), and rocsparse_zbsric0(). It is expected that this function will be executed only once for a given matrix and particular operation type. The analysis meta data can be cleared by rocsparse_bsric0_clear().

rocsparse_bsric0_analysis can share its meta data with rocsparse_sbsrilu0_analysis(), rocsparse_dbsrilu0_analysis(), rocsparse_cbsrilu0_analysis(), rocsparse_zbsrilu0_analysis(), rocsparse_sbsrsv_analysis(), rocsparse_dbsrsv_analysis(), rocsparse_cbsrsv_analysis(), rocsparse_zbsrsv_analysis(), rocsparse_sbsrsm_analysis(), rocsparse_dbsrsm_analysis(), rocsparse_cbsrsm_analysis() and rocsparse_zbsrsm_analysis(). Selecting rocsparse_analysis_policy_reuse policy can greatly improve computation performance of meta data. However, the user need to make sure that the sparsity pattern remains unchanged. If this cannot be assured, rocsparse_analysis_policy_force has to be used.

Note

If the matrix sparsity pattern changes, the gathered information will become invalid.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] direction that specified whether to count nonzero elements by rocsparse_direction_row or by rocsparse_direction_row.

  • mb[in] number of block rows in the sparse BSR matrix.

  • nnzb[in] number of non-zero block entries of the sparse BSR matrix.

  • descr[in] descriptor of the sparse BSR matrix.

  • bsr_val[in] array of length nnzb*block_dim*block_dim containing the values of the sparse BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnzb elements containing the block column indices of the sparse BSR matrix.

  • block_dim[in] the block dimension of the BSR matrix. Between 1 and m where m=mb*block_dim.

  • info[out] structure that holds the information collected during the analysis step.

  • analysis[in] rocsparse_analysis_policy_reuse or rocsparse_analysis_policy_force.

  • solve[in] rocsparse_solve_policy_auto.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, nnzb, or block_dim is invalid.

  • rocsparse_status_invalid_pointerdescr, bsr_val, bsr_row_ptr, bsr_col_ind, info or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_bsric0()#

rocsparse_status rocsparse_sbsric0(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_dbsric0(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_cbsric0(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_zbsric0(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#

Incomplete Cholesky factorization with 0 fill-ins and no pivoting using BSR storage format.

rocsparse_bsric0 computes the incomplete Cholesky factorization with 0 fill-ins and no pivoting of a sparse \(mb \times mb\) BSR matrix \(A\), such that

\[ A \approx LL^T \]

rocsparse_bsric0 requires a user allocated temporary buffer. Its size is returned by rocsparse_sbsric0_buffer_size(), rocsparse_dbsric0_buffer_size(), rocsparse_cbsric0_buffer_size() or rocsparse_zbsric0_buffer_size(). Furthermore, analysis meta data is required. It can be obtained by rocsparse_sbsric0_analysis(), rocsparse_dbsric0_analysis(), rocsparse_cbsric0_analysis() or rocsparse_zbsric0_analysis(). rocsparse_bsric0 reports the first zero pivot (either numerical or structural zero). The zero pivot status can be obtained by calling rocsparse_bsric0_zero_pivot().

Example

Consider the sparse \(m \times m\) matrix \(A\), stored in BSR storage format. The following example computes the incomplete Cholesky factorization \(M \approx LL^T\) and solves the preconditioned system \(My = x\).

// Create rocSPARSE handle
rocsparse_handle handle;
rocsparse_create_handle(&handle);

// Create matrix descriptor for M
rocsparse_mat_descr descr_M;
rocsparse_create_mat_descr(&descr_M);

// Create matrix descriptor for L
rocsparse_mat_descr descr_L;
rocsparse_create_mat_descr(&descr_L);
rocsparse_set_mat_fill_mode(descr_L, rocsparse_fill_mode_lower);
rocsparse_set_mat_diag_type(descr_L, rocsparse_diag_type_unit);

// Create matrix descriptor for L'
rocsparse_mat_descr descr_Lt;
rocsparse_create_mat_descr(&descr_Lt);
rocsparse_set_mat_fill_mode(descr_Lt, rocsparse_fill_mode_upper);
rocsparse_set_mat_diag_type(descr_Lt, rocsparse_diag_type_non_unit);

// Create matrix info structure
rocsparse_mat_info info;
rocsparse_create_mat_info(&info);

// Obtain required buffer size
size_t buffer_size_M;
size_t buffer_size_L;
size_t buffer_size_Lt;
rocsparse_dbsric0_buffer_size(handle,
                               rocsparse_direction_row,
                               mb,
                               nnzb,
                               descr_M,
                               bsr_val,
                               bsr_row_ptr,
                               bsr_col_ind,
                               block_dim,
                               info,
                               &buffer_size_M);
rocsparse_dbsrsv_buffer_size(handle,
                             rocsparse_direction_row,
                             rocsparse_operation_none,
                             mb,
                             nnzb,
                             descr_L,
                             bsr_val,
                             bsr_row_ptr,
                             bsr_col_ind,
                             block_dim,
                             info,
                             &buffer_size_L);
rocsparse_dbsrsv_buffer_size(handle,
                             rocsparse_direction_row,
                             rocsparse_operation_transpose,
                             mb,
                             nnzb,
                             descr_Lt,
                             bsr_val,
                             bsr_row_ptr,
                             bsr_col_ind,
                             block_dim,
                             info,
                             &buffer_size_Lt);

size_t buffer_size = max(buffer_size_M, max(buffer_size_L, buffer_size_Lt));

// Allocate temporary buffer
void* temp_buffer;
hipMalloc(&temp_buffer, buffer_size);

// Perform analysis steps, using rocsparse_analysis_policy_reuse to improve
// computation performance
rocsparse_dbsric0_analysis(handle,
                            rocsparse_direction_row,
                            mb,
                            nnzb,
                            descr_M,
                            bsr_val,
                            bsr_row_ptr,
                            bsr_col_ind,
                            block_dim,
                            info,
                            rocsparse_analysis_policy_reuse,
                            rocsparse_solve_policy_auto,
                            temp_buffer);
rocsparse_dbsrsv_analysis(handle,
                          rocsparse_direction_row,
                          rocsparse_operation_none,
                          mb,
                          nnzb,
                          descr_L,
                          bsr_val,
                          bsr_row_ptr,
                          bsr_col_ind,
                          block_dim,
                          info,
                          rocsparse_analysis_policy_reuse,
                          rocsparse_solve_policy_auto,
                          temp_buffer);
rocsparse_dbsrsv_analysis(handle,
                          rocsparse_direction_row,
                          rocsparse_operation_transpose,
                          mb,
                          nnzb,
                          descr_Lt,
                          bsr_val,
                          bsr_row_ptr,
                          bsr_col_ind,
                          block_dim,
                          info,
                          rocsparse_analysis_policy_reuse,
                          rocsparse_solve_policy_auto,
                          temp_buffer);

// Check for zero pivot
rocsparse_int position;
if(rocsparse_status_zero_pivot == rocsparse_bsric0_zero_pivot(handle,
                                                              info,
                                                              &position))
{
    printf("A has structural zero at A(%d,%d)\n", position, position);
}

// Compute incomplete Cholesky factorization M = LL'
rocsparse_dbsric0(handle,
                   rocsparse_direction_row,
                   mb,
                   nnzb,
                   descr_M,
                   bsr_val,
                   bsr_row_ptr,
                   bsr_col_ind,
                   block_dim,
                   info,
                   rocsparse_solve_policy_auto,
                   temp_buffer);

// Check for zero pivot
if(rocsparse_status_zero_pivot == rocsparse_bsric0_zero_pivot(handle,
                                                               info,
                                                               &position))
{
    printf("L has structural and/or numerical zero at L(%d,%d)\n",
           position,
           position);
}

// Solve Lz = x
rocsparse_dbsrsv_solve(handle,
                       rocsparse_direction_row,
                       rocsparse_operation_none,
                       mb,
                       nnzb,
                       &alpha,
                       descr_L,
                       bsr_val,
                       bsr_row_ptr,
                       bsr_col_ind,
                       block_dim,
                       info,
                       x,
                       z,
                       rocsparse_solve_policy_auto,
                       temp_buffer);

// Solve L'y = z
rocsparse_dbsrsv_solve(handle,
                       rocsparse_direction_row,
                       rocsparse_operation_transpose,
                       mb,
                       nnzb,
                       &alpha,
                       descr_Lt,
                       bsr_val,
                       bsr_row_ptr,
                       bsr_col_ind,
                       block_dim,
                       info,
                       z,
                       y,
                       rocsparse_solve_policy_auto,
                       temp_buffer);

// Clean up
hipFree(temp_buffer);
rocsparse_destroy_mat_info(info);
rocsparse_destroy_mat_descr(descr_M);
rocsparse_destroy_mat_descr(descr_L);
rocsparse_destroy_mat_descr(descr_Lt);
rocsparse_destroy_handle(handle);

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] direction that specified whether to count nonzero elements by rocsparse_direction_row or by rocsparse_direction_row.

  • mb[in] number of block rows in the sparse BSR matrix.

  • nnzb[in] number of non-zero block entries of the sparse BSR matrix.

  • descr[in] descriptor of the sparse BSR matrix.

  • bsr_val[inout] array of length nnzb*block_dim*block_dim containing the values of the sparse BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnzb elements containing the block column indices of the sparse BSR matrix.

  • block_dim[in] the block dimension of the BSR matrix. Between 1 and m where m=mb*block_dim.

  • info[in] structure that holds the information collected during the analysis step.

  • policy[in] rocsparse_solve_policy_auto.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, nnzb, or block_dim is invalid.

  • rocsparse_status_invalid_pointerdescr, bsr_val, bsr_row_ptr or bsr_col_ind pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_bsric0_clear()#

rocsparse_status rocsparse_bsric0_clear(rocsparse_handle handle, rocsparse_mat_info info)#

Incomplete Cholesky factorization with 0 fill-ins and no pivoting using BSR storage format.

rocsparse_bsric0_clear deallocates all memory that was allocated by rocsparse_sbsric0_analysis(), rocsparse_dbsric0_analysis(), rocsparse_cbsric0_analysis() or rocsparse_zbsric0_analysis(). This is especially useful, if memory is an issue and the analysis data is not required for further computation.

Note

Calling rocsparse_bsric0_clear is optional. All allocated resources will be cleared, when the opaque rocsparse_mat_info struct is destroyed using rocsparse_destroy_mat_info().

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[inout] structure that holds the information collected during the analysis step.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo pointer is invalid.

  • rocsparse_status_memory_error – the buffer holding the meta data could not be deallocated.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_bsrilu0_zero_pivot()#

rocsparse_status rocsparse_bsrilu0_zero_pivot(rocsparse_handle handle, rocsparse_mat_info info, rocsparse_int *position)#

Incomplete LU factorization with 0 fill-ins and no pivoting using BSR storage format.

rocsparse_bsrilu0_zero_pivot returns rocsparse_status_zero_pivot, if either a structural or numerical zero has been found during rocsparse_sbsrilu0(), rocsparse_dbsrilu0(), rocsparse_cbsrilu0() or rocsparse_zbsrilu0() computation. The first zero pivot \(j\) at \(A_{j,j}\) is stored in position, using same index base as the BSR matrix.

position can be in host or device memory. If no zero pivot has been found, position is set to -1 and rocsparse_status_success is returned instead.

Note

If a zero pivot is found, position \(=j\) means that either the diagonal block \(A_{j,j}\) is missing (structural zero) or the diagonal block \(A_{j,j}\) is not invertible (numerical zero).

Note

rocsparse_bsrilu0_zero_pivot is a blocking function. It might influence performance negatively.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[in] structure that holds the information collected during the analysis step.

  • position[inout] pointer to zero pivot \(j\), can be in host or device memory.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo or position pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_zero_pivot – zero pivot has been found.

rocsparse_bsrilu0_numeric_boost()#

rocsparse_status rocsparse_sbsrilu0_numeric_boost(rocsparse_handle handle, rocsparse_mat_info info, int enable_boost, const float *boost_tol, const float *boost_val)#
rocsparse_status rocsparse_dbsrilu0_numeric_boost(rocsparse_handle handle, rocsparse_mat_info info, int enable_boost, const double *boost_tol, const double *boost_val)#
rocsparse_status rocsparse_cbsrilu0_numeric_boost(rocsparse_handle handle, rocsparse_mat_info info, int enable_boost, const float *boost_tol, const rocsparse_float_complex *boost_val)#
rocsparse_status rocsparse_zbsrilu0_numeric_boost(rocsparse_handle handle, rocsparse_mat_info info, int enable_boost, const double *boost_tol, const rocsparse_double_complex *boost_val)#

Incomplete LU factorization with 0 fill-ins and no pivoting using BSR storage format.

rocsparse_bsrilu0_numeric_boost enables the user to replace a numerical value in an incomplete LU factorization. tol is used to determine whether a numerical value is replaced by boost_val, such that \(A_{j,j} = \text{boost_val}\) if \(\text{tol} \ge \left|A_{j,j}\right|\).

Note

The boost value is enabled by setting enable_boost to 1 or disabled by setting enable_boost to 0.

Note

tol and boost_val can be in host or device memory.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[in] structure that holds the information collected during the analysis step.

  • enable_boost[in] enable/disable numeric boost.

  • boost_tol[in] tolerance to determine whether a numerical value is replaced or not.

  • boost_val[in] boost value to replace a numerical value.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo, tol or boost_val pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_bsrilu0_buffer_size()#

rocsparse_status rocsparse_sbsrilu0_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_dbsrilu0_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_cbsrilu0_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_zbsrilu0_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, size_t *buffer_size)#

Incomplete LU factorization with 0 fill-ins and no pivoting using BSR storage format.

rocsparse_bsrilu0_buffer_size returns the size of the temporary storage buffer that is required by rocsparse_sbsrilu0_analysis(), rocsparse_dbsrilu0_analysis(), rocsparse_cbsrilu0_analysis(), rocsparse_zbsrilu0_analysis(), rocsparse_sbsrilu0(), rocsparse_dbsrilu0(), rocsparse_sbsrilu0() and rocsparse_dbsrilu0(). The temporary storage buffer must be allocated by the user. The size of the temporary storage buffer is identical to the size returned by rocsparse_sbsrsv_buffer_size(), rocsparse_dbsrsv_buffer_size(), rocsparse_cbsrsv_buffer_size(), rocsparse_zbsrsv_buffer_size(), rocsparse_sbsric0_buffer_size(), rocsparse_dbsric0_buffer_size(), rocsparse_cbsric0_buffer_size() and rocsparse_zbsric0_buffer_size() if the matrix sparsity pattern is identical. The user allocated buffer can thus be shared between subsequent calls to those functions.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, nnzb, or block_dim is invalid.

  • rocsparse_status_invalid_pointerdescr, bsr_val, bsr_row_ptr, bsr_col_ind, info or buffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_bsrilu0_analysis()#

rocsparse_status rocsparse_sbsrilu0_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_dbsrilu0_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_cbsrilu0_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_zbsrilu0_analysis(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#

Incomplete LU factorization with 0 fill-ins and no pivoting using BSR storage format.

rocsparse_bsrilu0_analysis performs the analysis step for rocsparse_sbsrilu0() rocsparse_dbsrilu0(), rocsparse_cbsrilu0(), and rocsparse_zbsrilu0(). It is expected that this function will be executed only once for a given matrix. The analysis meta data can be cleared by rocsparse_bsrilu0_clear().

rocsparse_bsrilu0_analysis can share its meta data with rocsparse_sbsric0_analysis(), rocsparse_dbsric0_analysis(), rocsparse_cbsric0_analysis(), rocsparse_zbsric0_analysis(), rocsparse_sbsrsv_analysis(), rocsparse_dbsrsv_analysis(), rocsparse_cbsrsv_analysis(), rocsparse_zbsrsv_analysis(), rocsparse_sbsrsm_analysis(), rocsparse_dbsrsm_analysis(), rocsparse_cbsrsm_analysis() and rocsparse_zbsrsm_analysis(). Selecting rocsparse_analysis_policy_reuse policy can greatly improve computation performance of meta data. However, the user need to make sure that the sparsity pattern remains unchanged. If this cannot be assured, rocsparse_analysis_policy_force has to be used.

Note

If the matrix sparsity pattern changes, the gathered information will become invalid.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] direction that specified whether to count nonzero elements by rocsparse_direction_row or by rocsparse_direction_row.

  • mb[in] number of block rows in the sparse BSR matrix.

  • nnzb[in] number of non-zero block entries of the sparse BSR matrix.

  • descr[in] descriptor of the sparse BSR matrix.

  • bsr_val[in] array of length nnzb*block_dim*block_dim containing the values of the sparse BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnzb elements containing the block column indices of the sparse BSR matrix.

  • block_dim[in] the block dimension of the BSR matrix. Between 1 and m where m=mb*block_dim.

  • info[out] structure that holds the information collected during the analysis step.

  • analysis[in] rocsparse_analysis_policy_reuse or rocsparse_analysis_policy_force.

  • solve[in] rocsparse_solve_policy_auto.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, nnzb, or block_dim is invalid.

  • rocsparse_status_invalid_pointerdescr, bsr_val, bsr_row_ptr, bsr_col_ind, info or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_bsrilu0()#

rocsparse_status rocsparse_sbsrilu0(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_dbsrilu0(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_cbsrilu0(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_zbsrilu0(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nnzb, const rocsparse_mat_descr descr, rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#

Incomplete LU factorization with 0 fill-ins and no pivoting using BSR storage format.

rocsparse_bsrilu0 computes the incomplete LU factorization with 0 fill-ins and no pivoting of a sparse \(mb \times mb\) BSR matrix \(A\), such that

\[ A \approx LU \]

rocsparse_bsrilu0 requires a user allocated temporary buffer. Its size is returned by rocsparse_sbsrilu0_buffer_size(), rocsparse_dbsrilu0_buffer_size(), rocsparse_cbsrilu0_buffer_size() or rocsparse_zbsrilu0_buffer_size(). Furthermore, analysis meta data is required. It can be obtained by rocsparse_sbsrilu0_analysis(), rocsparse_dbsrilu0_analysis(), rocsparse_cbsrilu0_analysis() or rocsparse_zbsrilu0_analysis(). rocsparse_bsrilu0 reports the first zero pivot (either numerical or structural zero). The zero pivot status can be obtained by calling rocsparse_bsrilu0_zero_pivot().

Example

Consider the sparse \(m \times m\) matrix \(A\), stored in BSR storage format. The following example computes the incomplete LU factorization \(M \approx LU\) and solves the preconditioned system \(My = x\).

// Create rocSPARSE handle
rocsparse_handle handle;
rocsparse_create_handle(&handle);

// Create matrix descriptor for M
rocsparse_mat_descr descr_M;
rocsparse_create_mat_descr(&descr_M);

// Create matrix descriptor for L
rocsparse_mat_descr descr_L;
rocsparse_create_mat_descr(&descr_L);
rocsparse_set_mat_fill_mode(descr_L, rocsparse_fill_mode_lower);
rocsparse_set_mat_diag_type(descr_L, rocsparse_diag_type_unit);

// Create matrix descriptor for U
rocsparse_mat_descr descr_U;
rocsparse_create_mat_descr(&descr_U);
rocsparse_set_mat_fill_mode(descr_U, rocsparse_fill_mode_upper);
rocsparse_set_mat_diag_type(descr_U, rocsparse_diag_type_non_unit);

// Create matrix info structure
rocsparse_mat_info info;
rocsparse_create_mat_info(&info);

// Obtain required buffer size
size_t buffer_size_M;
size_t buffer_size_L;
size_t buffer_size_U;
rocsparse_dbsrilu0_buffer_size(handle,
                               rocsparse_direction_row,
                               mb,
                               nnzb,
                               descr_M,
                               bsr_val,
                               bsr_row_ptr,
                               bsr_col_ind,
                               block_dim,
                               info,
                               &buffer_size_M);
rocsparse_dbsrsv_buffer_size(handle,
                             rocsparse_direction_row,
                             rocsparse_operation_none,
                             mb,
                             nnzb,
                             descr_L,
                             bsr_val,
                             bsr_row_ptr,
                             bsr_col_ind,
                             block_dim,
                             info,
                             &buffer_size_L);
rocsparse_dbsrsv_buffer_size(handle,
                             rocsparse_direction_row,
                             rocsparse_operation_none,
                             mb,
                             nnzb,
                             descr_U,
                             bsr_val,
                             bsr_row_ptr,
                             bsr_col_ind,
                             block_dim,
                             info,
                             &buffer_size_U);

size_t buffer_size = max(buffer_size_M, max(buffer_size_L, buffer_size_U));

// Allocate temporary buffer
void* temp_buffer;
hipMalloc(&temp_buffer, buffer_size);

// Perform analysis steps, using rocsparse_analysis_policy_reuse to improve
// computation performance
rocsparse_dbsrilu0_analysis(handle,
                            rocsparse_direction_row,
                            mb,
                            nnzb,
                            descr_M,
                            bsr_val,
                            bsr_row_ptr,
                            bsr_col_ind,
                            block_dim,
                            info,
                            rocsparse_analysis_policy_reuse,
                            rocsparse_solve_policy_auto,
                            temp_buffer);
rocsparse_dbsrsv_analysis(handle,
                          rocsparse_direction_row,
                          rocsparse_operation_none,
                          mb,
                          nnzb,
                          descr_L,
                          bsr_val,
                          bsr_row_ptr,
                          bsr_col_ind,
                          block_dim,
                          info,
                          rocsparse_analysis_policy_reuse,
                          rocsparse_solve_policy_auto,
                          temp_buffer);
rocsparse_dbsrsv_analysis(handle,
                          rocsparse_direction_row,
                          rocsparse_operation_none,
                          mb,
                          nnzb,
                          descr_U,
                          bsr_val,
                          bsr_row_ptr,
                          bsr_col_ind,
                          block_dim,
                          info,
                          rocsparse_analysis_policy_reuse,
                          rocsparse_solve_policy_auto,
                          temp_buffer);

// Check for zero pivot
rocsparse_int position;
if(rocsparse_status_zero_pivot == rocsparse_bsrilu0_zero_pivot(handle,
                                                              info,
                                                              &position))
{
    printf("A has structural zero at A(%d,%d)\n", position, position);
}

// Compute incomplete LU factorization M = LU
rocsparse_dbsrilu0(handle,
                   rocsparse_direction_row,
                   mb,
                   nnzb,
                   descr_M,
                   bsr_val,
                   bsr_row_ptr,
                   bsr_col_ind,
                   block_dim,
                   info,
                   rocsparse_solve_policy_auto,
                   temp_buffer);

// Check for zero pivot
if(rocsparse_status_zero_pivot == rocsparse_bsrilu0_zero_pivot(handle,
                                                               info,
                                                               &position))
{
    printf("L has structural and/or numerical zero at L(%d,%d)\n",
           position,
           position);
}

// Solve Lz = x
rocsparse_dbsrsv_solve(handle,
                       rocsparse_direction_row,
                       rocsparse_operation_none,
                       mb,
                       nnzb,
                       &alpha,
                       descr_L,
                       bsr_val,
                       bsr_row_ptr,
                       bsr_col_ind,
                       block_dim,
                       info,
                       x,
                       z,
                       rocsparse_solve_policy_auto,
                       temp_buffer);

// Solve Uy = z
rocsparse_dbsrsv_solve(handle,
                       rocsparse_direction_row,
                       rocsparse_operation_none,
                       mb,
                       nnzb,
                       &alpha,
                       descr_U,
                       bsr_val,
                       bsr_row_ptr,
                       bsr_col_ind,
                       block_dim,
                       info,
                       z,
                       y,
                       rocsparse_solve_policy_auto,
                       temp_buffer);

// Clean up
hipFree(temp_buffer);
rocsparse_destroy_mat_info(info);
rocsparse_destroy_mat_descr(descr_M);
rocsparse_destroy_mat_descr(descr_L);
rocsparse_destroy_mat_descr(descr_U);
rocsparse_destroy_handle(handle);

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] direction that specified whether to count nonzero elements by rocsparse_direction_row or by rocsparse_direction_row.

  • mb[in] number of block rows in the sparse BSR matrix.

  • nnzb[in] number of non-zero block entries of the sparse BSR matrix.

  • descr[in] descriptor of the sparse BSR matrix.

  • bsr_val[inout] array of length nnzb*block_dim*block_dim containing the values of the sparse BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnzb elements containing the block column indices of the sparse BSR matrix.

  • block_dim[in] the block dimension of the BSR matrix. Between 1 and m where m=mb*block_dim.

  • info[in] structure that holds the information collected during the analysis step.

  • policy[in] rocsparse_solve_policy_auto.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, nnzb, or block_dim is invalid.

  • rocsparse_status_invalid_pointerdescr, bsr_val, bsr_row_ptr or bsr_col_ind pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_bsrilu0_clear()#

rocsparse_status rocsparse_bsrilu0_clear(rocsparse_handle handle, rocsparse_mat_info info)#

Incomplete LU factorization with 0 fill-ins and no pivoting using BSR storage format.

rocsparse_bsrilu0_clear deallocates all memory that was allocated by rocsparse_sbsrilu0_analysis(), rocsparse_dbsrilu0_analysis(), rocsparse_cbsrilu0_analysis() or rocsparse_zbsrilu0_analysis(). This is especially useful, if memory is an issue and the analysis data is not required for further computation.

Note

Calling rocsparse_bsrilu0_clear is optional. All allocated resources will be cleared, when the opaque rocsparse_mat_info struct is destroyed using rocsparse_destroy_mat_info().

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[inout] structure that holds the information collected during the analysis step.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo pointer is invalid.

  • rocsparse_status_memory_error – the buffer holding the meta data could not be deallocated.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_csric0_zero_pivot()#

rocsparse_status rocsparse_csric0_zero_pivot(rocsparse_handle handle, rocsparse_mat_info info, rocsparse_int *position)#

Incomplete Cholesky factorization with 0 fill-ins and no pivoting using CSR storage format.

rocsparse_csric_zero_pivot returns rocsparse_status_zero_pivot, if either a structural or numerical zero has been found during rocsparse_scsric0() or rocsparse_dcsric0() computation. The first zero pivot \(j\) at \(A_{j,j}\) is stored in position, using same index base as the CSR matrix.

position can be in host or device memory. If no zero pivot has been found, position is set to -1 and rocsparse_status_success is returned instead.

Note

rocsparse_csric0_zero_pivot is a blocking function. It might influence performance negatively.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[in] structure that holds the information collected during the analysis step.

  • position[inout] pointer to zero pivot \(j\), can be in host or device memory.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo or position pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_zero_pivot – zero pivot has been found.

rocsparse_csric0_buffer_size()#

rocsparse_status rocsparse_scsric0_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_dcsric0_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_ccsric0_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_zcsric0_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, size_t *buffer_size)#

Incomplete Cholesky factorization with 0 fill-ins and no pivoting using CSR storage format.

rocsparse_csric0_buffer_size returns the size of the temporary storage buffer that is required by rocsparse_scsric0_analysis(), rocsparse_dcsric0_analysis(), rocsparse_scsric0() and rocsparse_dcsric0(). The temporary storage buffer must be allocated by the user. The size of the temporary storage buffer is identical to the size returned by rocsparse_scsrsv_buffer_size(), rocsparse_dcsrsv_buffer_size(), rocsparse_scsrilu0_buffer_size() and rocsparse_dcsrilu0_buffer_size() if the matrix sparsity pattern is identical. The user allocated buffer can thus be shared between subsequent calls to those functions.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • descr[in] descriptor of the sparse CSR matrix.

  • csr_val[in] array of nnz elements of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • info[out] structure that holds the information collected during the analysis step.

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_scsric0_analysis(), rocsparse_dcsric0_analysis(), rocsparse_scsric0() and rocsparse_dcsric0().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or nnz is invalid.

  • rocsparse_status_invalid_pointerdescr, csr_val, csr_row_ptr, csr_col_ind, info or buffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedtrans != rocsparse_operation_none or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csric0_analysis()#

rocsparse_status rocsparse_scsric0_analysis(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_dcsric0_analysis(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_ccsric0_analysis(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_zcsric0_analysis(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#

Incomplete Cholesky factorization with 0 fill-ins and no pivoting using CSR storage format.

rocsparse_csric0_analysis performs the analysis step for rocsparse_scsric0() and rocsparse_dcsric0(). It is expected that this function will be executed only once for a given matrix and particular operation type. The analysis meta data can be cleared by rocsparse_csric0_clear().

rocsparse_csric0_analysis can share its meta data with rocsparse_scsrilu0_analysis(), rocsparse_dcsrilu0_analysis(), rocsparse_ccsrilu0_analysis(), rocsparse_zcsrilu0_analysis(), rocsparse_scsrsv_analysis(), rocsparse_dcsrsv_analysis(), rocsparse_ccsrsv_analysis(), rocsparse_zcsrsv_analysis(), rocsparse_scsrsm_analysis(), rocsparse_dcsrsm_analysis(), rocsparse_scsrsm_analysis() and rocsparse_dcsrsm_analysis(). Selecting rocsparse_analysis_policy_reuse policy can greatly improve computation performance of meta data. However, the user need to make sure that the sparsity pattern remains unchanged. If this cannot be assured, rocsparse_analysis_policy_force has to be used.

Note

If the matrix sparsity pattern changes, the gathered information will become invalid.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • descr[in] descriptor of the sparse CSR matrix.

  • csr_val[in] array of nnz elements of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • info[out] structure that holds the information collected during the analysis step.

  • analysis[in] rocsparse_analysis_policy_reuse or rocsparse_analysis_policy_force.

  • solve[in] rocsparse_solve_policy_auto.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or nnz is invalid.

  • rocsparse_status_invalid_pointerdescr, csr_val, csr_row_ptr, csr_col_ind, info or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedtrans != rocsparse_operation_none or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csric0()#

rocsparse_status rocsparse_scsric0(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_dcsric0(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_ccsric0(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_zcsric0(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#

Incomplete Cholesky factorization with 0 fill-ins and no pivoting using CSR storage format.

rocsparse_csric0 computes the incomplete Cholesky factorization with 0 fill-ins and no pivoting of a sparse \(m \times m\) CSR matrix \(A\), such that

\[ A \approx LL^T \]

rocsparse_csric0 requires a user allocated temporary buffer. Its size is returned by rocsparse_scsric0_buffer_size() or rocsparse_dcsric0_buffer_size(). Furthermore, analysis meta data is required. It can be obtained by rocsparse_scsric0_analysis() or rocsparse_dcsric0_analysis(). rocsparse_csric0 reports the first zero pivot (either numerical or structural zero). The zero pivot status can be obtained by calling rocsparse_csric0_zero_pivot().

Example

Consider the sparse \(m \times m\) matrix \(A\), stored in CSR storage format. The following example computes the incomplete Cholesky factorization \(M \approx LL^T\) and solves the preconditioned system \(My = x\).

// Create rocSPARSE handle
rocsparse_handle handle;
rocsparse_create_handle(&handle);

// Create matrix descriptor for M
rocsparse_mat_descr descr_M;
rocsparse_create_mat_descr(&descr_M);

// Create matrix descriptor for L
rocsparse_mat_descr descr_L;
rocsparse_create_mat_descr(&descr_L);
rocsparse_set_mat_fill_mode(descr_L, rocsparse_fill_mode_lower);
rocsparse_set_mat_diag_type(descr_L, rocsparse_diag_type_unit);

// Create matrix descriptor for L'
rocsparse_mat_descr descr_Lt;
rocsparse_create_mat_descr(&descr_Lt);
rocsparse_set_mat_fill_mode(descr_Lt, rocsparse_fill_mode_upper);
rocsparse_set_mat_diag_type(descr_Lt, rocsparse_diag_type_non_unit);

// Create matrix info structure
rocsparse_mat_info info;
rocsparse_create_mat_info(&info);

// Obtain required buffer size
size_t buffer_size_M;
size_t buffer_size_L;
size_t buffer_size_Lt;
rocsparse_dcsric0_buffer_size(handle,
                              m,
                              nnz,
                              descr_M,
                              csr_val,
                              csr_row_ptr,
                              csr_col_ind,
                              info,
                              &buffer_size_M);
rocsparse_dcsrsv_buffer_size(handle,
                             rocsparse_operation_none,
                             m,
                             nnz,
                             descr_L,
                             csr_val,
                             csr_row_ptr,
                             csr_col_ind,
                             info,
                             &buffer_size_L);
rocsparse_dcsrsv_buffer_size(handle,
                             rocsparse_operation_transpose,
                             m,
                             nnz,
                             descr_Lt,
                             csr_val,
                             csr_row_ptr,
                             csr_col_ind,
                             info,
                             &buffer_size_Lt);

size_t buffer_size = max(buffer_size_M, max(buffer_size_L, buffer_size_Lt));

// Allocate temporary buffer
void* temp_buffer;
hipMalloc(&temp_buffer, buffer_size);

// Perform analysis steps, using rocsparse_analysis_policy_reuse to improve
// computation performance
rocsparse_dcsric0_analysis(handle,
                           m,
                           nnz,
                           descr_M,
                           csr_val,
                           csr_row_ptr,
                           csr_col_ind,
                           info,
                           rocsparse_analysis_policy_reuse,
                           rocsparse_solve_policy_auto,
                           temp_buffer);
rocsparse_dcsrsv_analysis(handle,
                          rocsparse_operation_none,
                          m,
                          nnz,
                          descr_L,
                          csr_val,
                          csr_row_ptr,
                          csr_col_ind,
                          info,
                          rocsparse_analysis_policy_reuse,
                          rocsparse_solve_policy_auto,
                          temp_buffer);
rocsparse_dcsrsv_analysis(handle,
                          rocsparse_operation_transpose,
                          m,
                          nnz,
                          descr_Lt,
                          csr_val,
                          csr_row_ptr,
                          csr_col_ind,
                          info,
                          rocsparse_analysis_policy_reuse,
                          rocsparse_solve_policy_auto,
                          temp_buffer);

// Check for zero pivot
rocsparse_int position;
if(rocsparse_status_zero_pivot == rocsparse_csric0_zero_pivot(handle,
                                                              info,
                                                              &position))
{
    printf("A has structural zero at A(%d,%d)\n", position, position);
}

// Compute incomplete Cholesky factorization M = LL'
rocsparse_dcsric0(handle,
                  m,
                  nnz,
                  descr_M,
                  csr_val,
                  csr_row_ptr,
                  csr_col_ind,
                  info,
                  rocsparse_solve_policy_auto,
                  temp_buffer);

// Check for zero pivot
if(rocsparse_status_zero_pivot == rocsparse_csric0_zero_pivot(handle,
                                                              info,
                                                              &position))
{
    printf("L has structural and/or numerical zero at L(%d,%d)\n",
           position,
           position);
}

// Solve Lz = x
rocsparse_dcsrsv_solve(handle,
                       rocsparse_operation_none,
                       m,
                       nnz,
                       &alpha,
                       descr_L,
                       csr_val,
                       csr_row_ptr,
                       csr_col_ind,
                       info,
                       x,
                       z,
                       rocsparse_solve_policy_auto,
                       temp_buffer);

// Solve L'y = z
rocsparse_dcsrsv_solve(handle,
                       rocsparse_operation_transpose,
                       m,
                       nnz,
                       &alpha,
                       descr_Lt,
                       csr_val,
                       csr_row_ptr,
                       csr_col_ind,
                       info,
                       z,
                       y,
                       rocsparse_solve_policy_auto,
                       temp_buffer);

// Clean up
hipFree(temp_buffer);
rocsparse_destroy_mat_info(info);
rocsparse_destroy_mat_descr(descr_M);
rocsparse_destroy_mat_descr(descr_L);
rocsparse_destroy_mat_descr(descr_Lt);
rocsparse_destroy_handle(handle);

Note

The sparse CSR matrix has to be sorted. This can be achieved by calling rocsparse_csrsort().

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • descr[in] descriptor of the sparse CSR matrix.

  • csr_val[inout] array of nnz elements of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • info[in] structure that holds the information collected during the analysis step.

  • policy[in] rocsparse_solve_policy_auto.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or nnz is invalid.

  • rocsparse_status_invalid_pointerdescr, csr_val, csr_row_ptr or csr_col_ind pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedtrans != rocsparse_operation_none or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csric0_clear()#

rocsparse_status rocsparse_csric0_clear(rocsparse_handle handle, rocsparse_mat_info info)#

Incomplete Cholesky factorization with 0 fill-ins and no pivoting using CSR storage format.

rocsparse_csric0_clear deallocates all memory that was allocated by rocsparse_scsric0_analysis() or rocsparse_dcsric0_analysis(). This is especially useful, if memory is an issue and the analysis data is not required for further computation.

Note

Calling rocsparse_csric0_clear is optional. All allocated resources will be cleared, when the opaque rocsparse_mat_info struct is destroyed using rocsparse_destroy_mat_info().

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[inout] structure that holds the information collected during the analysis step.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo pointer is invalid.

  • rocsparse_status_memory_error – the buffer holding the meta data could not be deallocated.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_csritilu0_buffer_size()#

rocsparse_status rocsparse_csritilu0_buffer_size(rocsparse_handle handle, rocsparse_itilu0_alg alg, rocsparse_int option, rocsparse_int nmaxiter, rocsparse_int m, rocsparse_int nnz, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_index_base base, rocsparse_datatype datatype, size_t *buffer_size)#

Iterative Incomplete LU factorization with 0 fill-ins and no pivoting using CSR storage format.

rocsparse_csritilu0_buffer_size computes the size in bytes of the buffer that has to be allocated by the user.

Note

The sparse CSR matrix has to be sorted. This can be achieved by calling rocsparse_csrsort().

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • alg[in] algorithm to use, rocsparse_itilu0_alg

  • option[in] combination of enumeration values from rocsparse_itilu0_option.

  • nmaxiter[in] maximum number of iterations.

  • m[in] number of rows of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • idx_base[in] rocsparse_index_base_zero or rocsparse_index_base_one.

  • datatype[in] Type of numerical values, rocsparse_datatype.

  • buffer_size[out] size of the temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or nnz is invalid.

  • rocsparse_status_invalid_valuealg, base or datatype is invalid.

  • rocsparse_status_invalid_pointercsr_row_ptr or csr_col_ind pointer is invalid.

  • rocsparse_status_zero_pivot – if nnz is zero.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_csritilu0_preprocess()#

rocsparse_status rocsparse_csritilu0_preprocess(rocsparse_handle handle, rocsparse_itilu0_alg alg, rocsparse_int option, rocsparse_int nmaxiter, rocsparse_int m, rocsparse_int nnz, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_index_base base, rocsparse_datatype datatype, size_t buffer_size, void *buffer)#

Iterative Incomplete LU factorization with 0 fill-ins and no pivoting using CSR storage format.

rocsparse_csritilu0_preprocess computes the information required to run rocsparse_csritilu0_compute and stores it in the buffer.

Note

The sparse CSR matrix has to be sorted. This can be achieved by calling rocsparse_csrsort().

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • alg[in] algorithm to use, rocsparse_itilu0_alg

  • option[in] combination of enumeration values from rocsparse_itilu0_option.

  • nmaxiter[in] maximum number of iterations.

  • m[in] number of rows of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • idx_base[in] rocsparse_index_base_zero or rocsparse_index_base_one.

  • datatype[in] Type of numerical values, rocsparse_datatype.

  • buffer_size[in] size of the storage buffer allocated by the user.

  • buffer[in] storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valuealg, base or datatype is invalid.

  • rocsparse_status_invalid_sizem or nnz is invalid.

  • rocsparse_status_invalid_pointercsr_row_ptr or csr_col_ind pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_zero_pivot – if missing diagonal element is detected.

rocsparse_csritilu0_history()#

rocsparse_status rocsparse_scsritilu0_history(rocsparse_handle handle, rocsparse_itilu0_alg alg, rocsparse_int *niter, float *data, size_t buffer_size, void *buffer)#
rocsparse_status rocsparse_dcsritilu0_history(rocsparse_handle handle, rocsparse_itilu0_alg alg, rocsparse_int *niter, double *data, size_t buffer_size, void *buffer)#
rocsparse_status rocsparse_ccsritilu0_history(rocsparse_handle handle, rocsparse_itilu0_alg alg, rocsparse_int *niter, float *data, size_t buffer_size, void *buffer)#
rocsparse_status rocsparse_zcsritilu0_history(rocsparse_handle handle, rocsparse_itilu0_alg alg, rocsparse_int *niter, double *data, size_t buffer_size, void *buffer)#

Iterative Incomplete LU factorization with 0 fill-ins and no pivoting using CSR storage format.

rocsparse_csritilu0_history fetches convergence history data.

Note

The sparse CSR matrix has to be sorted. This can be achieved by calling rocsparse_csrsort().

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • alg[in] algorithm to use, rocsparse_itilu0_alg

  • niter[out] number of performed iterations.

  • data[out] norms.

  • buffer_size[in] size of the buffer allocated by the user.

  • buffer[in] buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerniter or data is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_csritilu0_compute()#

rocsparse_status rocsparse_scsritilu0_compute(rocsparse_handle handle, rocsparse_itilu0_alg alg, rocsparse_int option, rocsparse_int *nmaxiter, float tol, rocsparse_int m, rocsparse_int nnz, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const float *csr_val, float *ilu0, rocsparse_index_base base, size_t buffer_size, void *buffer)#
rocsparse_status rocsparse_dcsritilu0_compute(rocsparse_handle handle, rocsparse_itilu0_alg alg, rocsparse_int option, rocsparse_int *nmaxiter, double tol, rocsparse_int m, rocsparse_int nnz, const rocsparse_int *ptr, const rocsparse_int *ind, const double *val, double *ilu0, rocsparse_index_base base, size_t buffer_size, void *buffer)#
rocsparse_status rocsparse_ccsritilu0_compute(rocsparse_handle handle, rocsparse_itilu0_alg alg, rocsparse_int option, rocsparse_int *nmaxiter, float tol, rocsparse_int m, rocsparse_int nnz, const rocsparse_int *ptr, const rocsparse_int *ind, const rocsparse_float_complex *val, rocsparse_float_complex *ilu0, rocsparse_index_base base, size_t buffer_size, void *buffer)#
rocsparse_status rocsparse_zcsritilu0_compute(rocsparse_handle handle, rocsparse_itilu0_alg alg, rocsparse_int option, rocsparse_int *nmaxiter, double tol, rocsparse_int m, rocsparse_int nnz, const rocsparse_int *ptr, const rocsparse_int *ind, const rocsparse_double_complex *val, rocsparse_double_complex *ilu0, rocsparse_index_base base, size_t buffer_size, void *buffer)#

Iterative Incomplete LU factorization with 0 fill-ins and no pivoting using CSR storage format.

rocsparse_csritilu0_compute computes iteratively the incomplete LU factorization with 0 fill-ins and no pivoting of a sparse \(m \times m\) CSR matrix \(A\), such that

\[ A \approx LU \]

rocsparse_csritilu0 requires a user allocated temporary buffer. Its size is returned by rocsparse_csritilu0_buffer_size(). Furthermore, analysis meta data is required. It can be obtained by rocsparse_csritlu0_preprocess().

Note

The sparse CSR matrix has to be sorted. This can be achieved by calling rocsparse_csrsort().

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • alg[in] algorithm to use, rocsparse_itilu0_alg

  • option[in] combination of enumeration values from rocsparse_itilu0_option.

  • nmaxiter[inout] maximum number of iterations.

  • tol[in] tolerance to use for stopping criteria.

  • m[in] number of rows of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • [ino] – csr_val array of nnz elements of the sparse CSR matrix.

  • ilu0[out] incomplete factorization.

  • idx_base[in] rocsparse_index_base_zero or rocsparse_index_base_one.

  • datatype[in] Type of numerical values, rocsparse_datatype.

  • buffer_size[in] size of the storage buffer allocated by the user.

  • buffer[in] storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_valuealg, base or datatype is invalid.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or nnz is invalid.

  • rocsparse_status_invalid_pointercsr_row_ptr or csr_col_ind pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_csrilu0_zero_pivot()#

rocsparse_status rocsparse_csrilu0_zero_pivot(rocsparse_handle handle, rocsparse_mat_info info, rocsparse_int *position)#

Incomplete LU factorization with 0 fill-ins and no pivoting using CSR storage format.

rocsparse_csrilu0_zero_pivot returns rocsparse_status_zero_pivot, if either a structural or numerical zero has been found during rocsparse_scsrilu0(), rocsparse_dcsrilu0(), rocsparse_ccsrilu0() or rocsparse_zcsrilu0() computation. The first zero pivot \(j\) at \(A_{j,j}\) is stored in position, using same index base as the CSR matrix.

position can be in host or device memory. If no zero pivot has been found, position is set to -1 and rocsparse_status_success is returned instead.

Note

rocsparse_csrilu0_zero_pivot is a blocking function. It might influence performance negatively.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[in] structure that holds the information collected during the analysis step.

  • position[inout] pointer to zero pivot \(j\), can be in host or device memory.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo or position pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_zero_pivot – zero pivot has been found.

rocsparse_csrilu0_numeric_boost()#

rocsparse_status rocsparse_scsrilu0_numeric_boost(rocsparse_handle handle, rocsparse_mat_info info, int enable_boost, const float *boost_tol, const float *boost_val)#
rocsparse_status rocsparse_dcsrilu0_numeric_boost(rocsparse_handle handle, rocsparse_mat_info info, int enable_boost, const double *boost_tol, const double *boost_val)#
rocsparse_status rocsparse_ccsrilu0_numeric_boost(rocsparse_handle handle, rocsparse_mat_info info, int enable_boost, const float *boost_tol, const rocsparse_float_complex *boost_val)#
rocsparse_status rocsparse_zcsrilu0_numeric_boost(rocsparse_handle handle, rocsparse_mat_info info, int enable_boost, const double *boost_tol, const rocsparse_double_complex *boost_val)#

Incomplete LU factorization with 0 fill-ins and no pivoting using CSR storage format.

rocsparse_csrilu0_numeric_boost enables the user to replace a numerical value in an incomplete LU factorization. tol is used to determine whether a numerical value is replaced by boost_val, such that \(A_{j,j} = \text{boost_val}\) if \(\text{tol} \ge \left|A_{j,j}\right|\).

Note

The boost value is enabled by setting enable_boost to 1 or disabled by setting enable_boost to 0.

Note

tol and boost_val can be in host or device memory.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[in] structure that holds the information collected during the analysis step.

  • enable_boost[in] enable/disable numeric boost.

  • boost_tol[in] tolerance to determine whether a numerical value is replaced or not.

  • boost_val[in] boost value to replace a numerical value.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo, tol or boost_val pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_csrilu0_buffer_size()#

rocsparse_status rocsparse_scsrilu0_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_dcsrilu0_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_ccsrilu0_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_zcsrilu0_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, size_t *buffer_size)#

Incomplete LU factorization with 0 fill-ins and no pivoting using CSR storage format.

rocsparse_csrilu0_buffer_size returns the size of the temporary storage buffer that is required by rocsparse_scsrilu0_analysis(), rocsparse_dcsrilu0_analysis(), rocsparse_ccsrilu0_analysis(), rocsparse_zcsrilu0_analysis(), rocsparse_scsrilu0(), rocsparse_dcsrilu0(), rocsparse_ccsrilu0() and rocsparse_zcsrilu0(). The temporary storage buffer must be allocated by the user. The size of the temporary storage buffer is identical to the size returned by rocsparse_scsrsv_buffer_size(), rocsparse_dcsrsv_buffer_size(), rocsparse_ccsrsv_buffer_size() and rocsparse_zcsrsv_buffer_size() if the matrix sparsity pattern is identical. The user allocated buffer can thus be shared between subsequent calls to those functions.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or nnz is invalid.

  • rocsparse_status_invalid_pointerdescr, csr_val, csr_row_ptr, csr_col_ind, info or buffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedtrans != rocsparse_operation_none or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csrilu0_analysis()#

rocsparse_status rocsparse_scsrilu0_analysis(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_dcsrilu0_analysis(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_ccsrilu0_analysis(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#
rocsparse_status rocsparse_zcsrilu0_analysis(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_analysis_policy analysis, rocsparse_solve_policy solve, void *temp_buffer)#

Incomplete LU factorization with 0 fill-ins and no pivoting using CSR storage format.

rocsparse_csrilu0_analysis performs the analysis step for rocsparse_scsrilu0(), rocsparse_dcsrilu0(), rocsparse_ccsrilu0() and rocsparse_zcsrilu0(). It is expected that this function will be executed only once for a given matrix and particular operation type. The analysis meta data can be cleared by rocsparse_csrilu0_clear().

rocsparse_csrilu0_analysis can share its meta data with rocsparse_scsric0_analysis(), rocsparse_dcsric0_analysis(), rocsparse_ccsric0_analysis(), rocsparse_zcsric0_analysis(), rocsparse_scsrsv_analysis(), rocsparse_dcsrsv_analysis(), rocsparse_ccsrsv_analysis(), rocsparse_zcsrsv_analysis(), rocsparse_scsrsm_analysis(), rocsparse_dcsrsm_analysis(), rocsparse_scsrsm_analysis() and rocsparse_dcsrsm_analysis(). Selecting rocsparse_analysis_policy_reuse policy can greatly improve computation performance of meta data. However, the user need to make sure that the sparsity pattern remains unchanged. If this cannot be assured, rocsparse_analysis_policy_force has to be used.

Note

If the matrix sparsity pattern changes, the gathered information will become invalid.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • descr[in] descriptor of the sparse CSR matrix.

  • csr_val[in] array of nnz elements of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • info[out] structure that holds the information collected during the analysis step.

  • analysis[in] rocsparse_analysis_policy_reuse or rocsparse_analysis_policy_force.

  • solve[in] rocsparse_solve_policy_auto.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or nnz is invalid.

  • rocsparse_status_invalid_pointerdescr, csr_val, csr_row_ptr, csr_col_ind, info or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedtrans != rocsparse_operation_none or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csrilu0()#

rocsparse_status rocsparse_scsrilu0(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_dcsrilu0(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_ccsrilu0(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#
rocsparse_status rocsparse_zcsrilu0(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, rocsparse_solve_policy policy, void *temp_buffer)#

Incomplete LU factorization with 0 fill-ins and no pivoting using CSR storage format.

rocsparse_csrilu0 computes the incomplete LU factorization with 0 fill-ins and no pivoting of a sparse \(m \times m\) CSR matrix \(A\), such that

\[ A \approx LU \]

rocsparse_csrilu0 requires a user allocated temporary buffer. Its size is returned by rocsparse_scsrilu0_buffer_size(), rocsparse_dcsrilu0_buffer_size(), rocsparse_ccsrilu0_buffer_size() or rocsparse_zcsrilu0_buffer_size(). Furthermore, analysis meta data is required. It can be obtained by rocsparse_scsrilu0_analysis(), rocsparse_dcsrilu0_analysis(), rocsparse_ccsrilu0_analysis() or rocsparse_zcsrilu0_analysis(). rocsparse_csrilu0 reports the first zero pivot (either numerical or structural zero). The zero pivot status can be obtained by calling rocsparse_csrilu0_zero_pivot().

Example

Consider the sparse \(m \times m\) matrix \(A\), stored in CSR storage format. The following example computes the incomplete LU factorization \(M \approx LU\) and solves the preconditioned system \(My = x\).

// Create rocSPARSE handle
rocsparse_handle handle;
rocsparse_create_handle(&handle);

// Create matrix descriptor for M
rocsparse_mat_descr descr_M;
rocsparse_create_mat_descr(&descr_M);

// Create matrix descriptor for L
rocsparse_mat_descr descr_L;
rocsparse_create_mat_descr(&descr_L);
rocsparse_set_mat_fill_mode(descr_L, rocsparse_fill_mode_lower);
rocsparse_set_mat_diag_type(descr_L, rocsparse_diag_type_unit);

// Create matrix descriptor for U
rocsparse_mat_descr descr_U;
rocsparse_create_mat_descr(&descr_U);
rocsparse_set_mat_fill_mode(descr_U, rocsparse_fill_mode_upper);
rocsparse_set_mat_diag_type(descr_U, rocsparse_diag_type_non_unit);

// Create matrix info structure
rocsparse_mat_info info;
rocsparse_create_mat_info(&info);

// Obtain required buffer size
size_t buffer_size_M;
size_t buffer_size_L;
size_t buffer_size_U;
rocsparse_dcsrilu0_buffer_size(handle,
                              m,
                              nnz,
                              descr_M,
                              csr_val,
                              csr_row_ptr,
                              csr_col_ind,
                              info,
                              &buffer_size_M);
rocsparse_dcsrsv_buffer_size(handle,
                             rocsparse_operation_none,
                             m,
                             nnz,
                             descr_L,
                             csr_val,
                             csr_row_ptr,
                             csr_col_ind,
                             info,
                             &buffer_size_L);
rocsparse_dcsrsv_buffer_size(handle,
                             rocsparse_operation_none,
                             m,
                             nnz,
                             descr_U,
                             csr_val,
                             csr_row_ptr,
                             csr_col_ind,
                             info,
                             &buffer_size_U);

size_t buffer_size = max(buffer_size_M, max(buffer_size_L, buffer_size_U));

// Allocate temporary buffer
void* temp_buffer;
hipMalloc(&temp_buffer, buffer_size);

// Perform analysis steps, using rocsparse_analysis_policy_reuse to improve
// computation performance
rocsparse_dcsrilu0_analysis(handle,
                            m,
                            nnz,
                            descr_M,
                            csr_val,
                            csr_row_ptr,
                            csr_col_ind,
                            info,
                            rocsparse_analysis_policy_reuse,
                            rocsparse_solve_policy_auto,
                            temp_buffer);
rocsparse_dcsrsv_analysis(handle,
                          rocsparse_operation_none,
                          m,
                          nnz,
                          descr_L,
                          csr_val,
                          csr_row_ptr,
                          csr_col_ind,
                          info,
                          rocsparse_analysis_policy_reuse,
                          rocsparse_solve_policy_auto,
                          temp_buffer);
rocsparse_dcsrsv_analysis(handle,
                          rocsparse_operation_none,
                          m,
                          nnz,
                          descr_U,
                          csr_val,
                          csr_row_ptr,
                          csr_col_ind,
                          info,
                          rocsparse_analysis_policy_reuse,
                          rocsparse_solve_policy_auto,
                          temp_buffer);

// Check for zero pivot
rocsparse_int position;
if(rocsparse_status_zero_pivot == rocsparse_csrilu0_zero_pivot(handle,
                                                               info,
                                                               &position))
{
    printf("A has structural zero at A(%d,%d)\n", position, position);
}

// Compute incomplete LU factorization
rocsparse_dcsrilu0(handle,
                   m,
                   nnz,
                   descr_M,
                   csr_val,
                   csr_row_ptr,
                   csr_col_ind,
                   info,
                   rocsparse_solve_policy_auto,
                   temp_buffer);

// Check for zero pivot
if(rocsparse_status_zero_pivot == rocsparse_csrilu0_zero_pivot(handle,
                                                               info,
                                                               &position))
{
    printf("U has structural and/or numerical zero at U(%d,%d)\n",
           position,
           position);
}

// Solve Lz = x
rocsparse_dcsrsv_solve(handle,
                       rocsparse_operation_none,
                       m,
                       nnz,
                       &alpha,
                       descr_L,
                       csr_val,
                       csr_row_ptr,
                       csr_col_ind,
                       info,
                       x,
                       z,
                       rocsparse_solve_policy_auto,
                       temp_buffer);

// Solve Uy = z
rocsparse_dcsrsv_solve(handle,
                       rocsparse_operation_none,
                       m,
                       nnz,
                       &alpha,
                       descr_U,
                       csr_val,
                       csr_row_ptr,
                       csr_col_ind,
                       info,
                       z,
                       y,
                       rocsparse_solve_policy_auto,
                       temp_buffer);

// Clean up
hipFree(temp_buffer);
rocsparse_destroy_mat_info(info);
rocsparse_destroy_mat_descr(descr_M);
rocsparse_destroy_mat_descr(descr_L);
rocsparse_destroy_mat_descr(descr_U);
rocsparse_destroy_handle(handle);

Note

The sparse CSR matrix has to be sorted. This can be achieved by calling rocsparse_csrsort().

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • descr[in] descriptor of the sparse CSR matrix.

  • csr_val[inout] array of nnz elements of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • info[in] structure that holds the information collected during the analysis step.

  • policy[in] rocsparse_solve_policy_auto.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or nnz is invalid.

  • rocsparse_status_invalid_pointerdescr, csr_val, csr_row_ptr or csr_col_ind pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedtrans != rocsparse_operation_none or rocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csrilu0_clear()#

rocsparse_status rocsparse_csrilu0_clear(rocsparse_handle handle, rocsparse_mat_info info)#

Incomplete LU factorization with 0 fill-ins and no pivoting using CSR storage format.

rocsparse_csrilu0_clear deallocates all memory that was allocated by rocsparse_scsrilu0_analysis(), rocsparse_dcsrilu0_analysis(), rocsparse_ccsrilu0_analysis() or rocsparse_zcsrilu0_analysis(). This is especially useful, if memory is an issue and the analysis data is not required for further computation.

Note

Calling rocsparse_csrilu0_clear is optional. All allocated resources will be cleared, when the opaque rocsparse_mat_info struct is destroyed using rocsparse_destroy_mat_info().

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • info[inout] structure that holds the information collected during the analysis step.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerinfo pointer is invalid.

  • rocsparse_status_memory_error – the buffer holding the meta data could not be deallocated.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_gtsv_buffer_size()#

rocsparse_status rocsparse_sgtsv_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const float *dl, const float *d, const float *du, const float *B, rocsparse_int ldb, size_t *buffer_size)#
rocsparse_status rocsparse_dgtsv_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const double *dl, const double *d, const double *du, const double *B, rocsparse_int ldb, size_t *buffer_size)#
rocsparse_status rocsparse_cgtsv_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_float_complex *dl, const rocsparse_float_complex *d, const rocsparse_float_complex *du, const rocsparse_float_complex *B, rocsparse_int ldb, size_t *buffer_size)#
rocsparse_status rocsparse_zgtsv_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_double_complex *dl, const rocsparse_double_complex *d, const rocsparse_double_complex *du, const rocsparse_double_complex *B, rocsparse_int ldb, size_t *buffer_size)#

Tridiagonal solver with pivoting.

rocsparse_gtsv_buffer_size returns the size of the temporary storage buffer that is required by rocsparse_sgtsv(), rocsparse_dgtsv(), rocsparse_cgtsv() and rocsparse_zgtsv(). The temporary storage buffer must be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] size of the tri-diagonal linear system (must be >= 2).

  • n[in] number of columns in the dense matrix B.

  • dl[in] lower diagonal of tri-diagonal system. First entry must be zero.

  • d[in] main diagonal of tri-diagonal system.

  • du[in] upper diagonal of tri-diagonal system. Last entry must be zero.

  • B[in] Dense matrix of size ( ldb, n ).

  • ldb[in] Leading dimension of B. Must satisfy ldb >= max(1, m).

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_sgtsv(), rocsparse_dgtsv(), rocsparse_cgtsv() and rocsparse_zgtsv().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or ldb is invalid.

  • rocsparse_status_invalid_pointerdl, d, du, B or buffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_gtsv()#

rocsparse_status rocsparse_sgtsv(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const float *dl, const float *d, const float *du, float *B, rocsparse_int ldb, void *temp_buffer)#
rocsparse_status rocsparse_dgtsv(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const double *dl, const double *d, const double *du, double *B, rocsparse_int ldb, void *temp_buffer)#
rocsparse_status rocsparse_cgtsv(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_float_complex *dl, const rocsparse_float_complex *d, const rocsparse_float_complex *du, rocsparse_float_complex *B, rocsparse_int ldb, void *temp_buffer)#
rocsparse_status rocsparse_zgtsv(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_double_complex *dl, const rocsparse_double_complex *d, const rocsparse_double_complex *du, rocsparse_double_complex *B, rocsparse_int ldb, void *temp_buffer)#

Tridiagonal solver with pivoting.

rocsparse_gtsv solves a tridiagonal system for multiple right hand sides using pivoting.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] size of the tri-diagonal linear system (must be >= 2).

  • n[in] number of columns in the dense matrix B.

  • dl[in] lower diagonal of tri-diagonal system. First entry must be zero.

  • d[in] main diagonal of tri-diagonal system.

  • du[in] upper diagonal of tri-diagonal system. Last entry must be zero.

  • B[inout] Dense matrix of size ( ldb, n ).

  • ldb[in] Leading dimension of B. Must satisfy ldb >= max(1, m).

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or ldb is invalid.

  • rocsparse_status_invalid_pointerdl, d, du, B or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_gtsv_no_pivot_buffer_size()#

rocsparse_status rocsparse_sgtsv_no_pivot_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const float *dl, const float *d, const float *du, const float *B, rocsparse_int ldb, size_t *buffer_size)#
rocsparse_status rocsparse_dgtsv_no_pivot_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const double *dl, const double *d, const double *du, const double *B, rocsparse_int ldb, size_t *buffer_size)#
rocsparse_status rocsparse_cgtsv_no_pivot_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_float_complex *dl, const rocsparse_float_complex *d, const rocsparse_float_complex *du, const rocsparse_float_complex *B, rocsparse_int ldb, size_t *buffer_size)#
rocsparse_status rocsparse_zgtsv_no_pivot_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_double_complex *dl, const rocsparse_double_complex *d, const rocsparse_double_complex *du, const rocsparse_double_complex *B, rocsparse_int ldb, size_t *buffer_size)#

Tridiagonal solver (no pivoting)

rocsparse_gtsv_no_pivot_buffer_size returns the size of the temporary storage buffer that is required by rocsparse_sgtsv_no_pivot(), rocsparse_dgtsv_no_pivot(), rocsparse_cgtsv_no_pivot() and rocsparse_zgtsv_no_pivot(). The temporary storage buffer must be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] size of the tri-diagonal linear system (must be >= 2).

  • n[in] number of columns in the dense matrix B.

  • dl[in] lower diagonal of tri-diagonal system. First entry must be zero.

  • d[in] main diagonal of tri-diagonal system.

  • du[in] upper diagonal of tri-diagonal system. Last entry must be zero.

  • B[in] Dense matrix of size ( ldb, n ).

  • ldb[in] Leading dimension of B. Must satisfy ldb >= max(1, m).

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_sgtsv_no_pivot(), rocsparse_dgtsv_no_pivot(), rocsparse_cgtsv_no_pivot() and rocsparse_zgtsv_no_pivot().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or ldb is invalid.

  • rocsparse_status_invalid_pointerdl, d, du, B or buffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_gtsv_no_pivot()#

rocsparse_status rocsparse_sgtsv_no_pivot(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const float *dl, const float *d, const float *du, float *B, rocsparse_int ldb, void *temp_buffer)#
rocsparse_status rocsparse_dgtsv_no_pivot(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const double *dl, const double *d, const double *du, double *B, rocsparse_int ldb, void *temp_buffer)#
rocsparse_status rocsparse_cgtsv_no_pivot(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_float_complex *dl, const rocsparse_float_complex *d, const rocsparse_float_complex *du, rocsparse_float_complex *B, rocsparse_int ldb, void *temp_buffer)#
rocsparse_status rocsparse_zgtsv_no_pivot(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_double_complex *dl, const rocsparse_double_complex *d, const rocsparse_double_complex *du, rocsparse_double_complex *B, rocsparse_int ldb, void *temp_buffer)#

Tridiagonal solver (no pivoting)

rocsparse_gtsv_no_pivot solves a tridiagonal linear system for multiple right-hand sides

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] size of the tri-diagonal linear system (must be >= 2).

  • n[in] number of columns in the dense matrix B.

  • dl[in] lower diagonal of tri-diagonal system. First entry must be zero.

  • d[in] main diagonal of tri-diagonal system.

  • du[in] upper diagonal of tri-diagonal system. Last entry must be zero.

  • B[inout] Dense matrix of size ( ldb, n ).

  • ldb[in] Leading dimension of B. Must satisfy ldb >= max(1, m).

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or ldb is invalid.

  • rocsparse_status_invalid_pointerdl, d, du, B or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_gtsv_no_pivot_strided_batch_buffer_size()#

rocsparse_status rocsparse_sgtsv_no_pivot_strided_batch_buffer_size(rocsparse_handle handle, rocsparse_int m, const float *dl, const float *d, const float *du, const float *x, rocsparse_int batch_count, rocsparse_int batch_stride, size_t *buffer_size)#
rocsparse_status rocsparse_dgtsv_no_pivot_strided_batch_buffer_size(rocsparse_handle handle, rocsparse_int m, const double *dl, const double *d, const double *du, const double *x, rocsparse_int batch_count, rocsparse_int batch_stride, size_t *buffer_size)#
rocsparse_status rocsparse_cgtsv_no_pivot_strided_batch_buffer_size(rocsparse_handle handle, rocsparse_int m, const rocsparse_float_complex *dl, const rocsparse_float_complex *d, const rocsparse_float_complex *du, const rocsparse_float_complex *x, rocsparse_int batch_count, rocsparse_int batch_stride, size_t *buffer_size)#
rocsparse_status rocsparse_zgtsv_no_pivot_strided_batch_buffer_size(rocsparse_handle handle, rocsparse_int m, const rocsparse_double_complex *dl, const rocsparse_double_complex *d, const rocsparse_double_complex *du, const rocsparse_double_complex *x, rocsparse_int batch_count, rocsparse_int batch_stride, size_t *buffer_size)#

Strided Batch tridiagonal solver (no pivoting)

rocsparse_gtsv_no_pivot_strided_batch_buffer_size returns the size of the temporary storage buffer that is required by rocsparse_sgtsv_no_pivot_strided_batch(), rocsparse_dgtsv_no_pivot_strided_batch(), rocsparse_cgtsv_no_pivot_strided_batch() and rocsparse_zgtsv_no_pivot_strided_batch(). The temporary storage buffer must be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] size of the tri-diagonal linear system.

  • dl[in] lower diagonal of tri-diagonal system where the ith system lower diagonal starts at dl+batch_stride*i.

  • d[in] main diagonal of tri-diagonal system where the ith system diagonal starts at d+batch_stride*i.

  • du[in] upper diagonal of tri-diagonal system where the ith system upper diagonal starts at du+batch_stride*i.

  • x[inout] Dense array of righthand-sides where the ith righthand-side starts at x+batch_stride*i.

  • batch_count[in] The number of systems to solve.

  • batch_stride[in] The number of elements that separate each system. Must satisfy batch_stride >= m.

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_sgtsv_no_pivot_strided_batch(), rocsparse_dgtsv_no_pivot_strided_batch(), rocsparse_cgtsv_no_pivot_strided_batch() and rocsparse_zgtsv_no_pivot_strided_batch().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, batch_count or batch_stride is invalid.

  • rocsparse_status_invalid_pointerdl, d, du, x or buffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_gtsv_no_pivot_strided_batch()#

rocsparse_status rocsparse_sgtsv_no_pivot_strided_batch(rocsparse_handle handle, rocsparse_int m, const float *dl, const float *d, const float *du, float *x, rocsparse_int batch_count, rocsparse_int batch_stride, void *temp_buffer)#
rocsparse_status rocsparse_dgtsv_no_pivot_strided_batch(rocsparse_handle handle, rocsparse_int m, const double *dl, const double *d, const double *du, double *x, rocsparse_int batch_count, rocsparse_int batch_stride, void *temp_buffer)#
rocsparse_status rocsparse_cgtsv_no_pivot_strided_batch(rocsparse_handle handle, rocsparse_int m, const rocsparse_float_complex *dl, const rocsparse_float_complex *d, const rocsparse_float_complex *du, rocsparse_float_complex *x, rocsparse_int batch_count, rocsparse_int batch_stride, void *temp_buffer)#
rocsparse_status rocsparse_zgtsv_no_pivot_strided_batch(rocsparse_handle handle, rocsparse_int m, const rocsparse_double_complex *dl, const rocsparse_double_complex *d, const rocsparse_double_complex *du, rocsparse_double_complex *x, rocsparse_int batch_count, rocsparse_int batch_stride, void *temp_buffer)#

Strided Batch tridiagonal solver (no pivoting)

rocsparse_gtsv_no_pivot_strided_batch solves a batched tridiagonal linear system

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] size of the tri-diagonal linear system (must be >= 2).

  • dl[in] lower diagonal of tri-diagonal system. First entry must be zero.

  • d[in] main diagonal of tri-diagonal system.

  • du[in] upper diagonal of tri-diagonal system. Last entry must be zero.

  • x[inout] Dense array of righthand-sides where the ith righthand-side starts at x+batch_stride*i.

  • batch_count[in] The number of systems to solve.

  • batch_stride[in] The number of elements that separate each system. Must satisfy batch_stride >= m.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, batch_count or batch_stride is invalid.

  • rocsparse_status_invalid_pointerdl, d, du, x or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_gtsv_interleaved_batch_buffer_size()#

rocsparse_status rocsparse_sgtsv_interleaved_batch_buffer_size(rocsparse_handle handle, rocsparse_gtsv_interleaved_alg alg, rocsparse_int m, const float *dl, const float *d, const float *du, const float *x, rocsparse_int batch_count, rocsparse_int batch_stride, size_t *buffer_size)#
rocsparse_status rocsparse_dgtsv_interleaved_batch_buffer_size(rocsparse_handle handle, rocsparse_gtsv_interleaved_alg alg, rocsparse_int m, const double *dl, const double *d, const double *du, const double *x, rocsparse_int batch_count, rocsparse_int batch_stride, size_t *buffer_size)#
rocsparse_status rocsparse_cgtsv_interleaved_batch_buffer_size(rocsparse_handle handle, rocsparse_gtsv_interleaved_alg alg, rocsparse_int m, const rocsparse_float_complex *dl, const rocsparse_float_complex *d, const rocsparse_float_complex *du, const rocsparse_float_complex *x, rocsparse_int batch_count, rocsparse_int batch_stride, size_t *buffer_size)#
rocsparse_status rocsparse_zgtsv_interleaved_batch_buffer_size(rocsparse_handle handle, rocsparse_gtsv_interleaved_alg alg, rocsparse_int m, const rocsparse_double_complex *dl, const rocsparse_double_complex *d, const rocsparse_double_complex *du, const rocsparse_double_complex *x, rocsparse_int batch_count, rocsparse_int batch_stride, size_t *buffer_size)#

Interleaved Batch tridiagonal solver.

rocsparse_gtsv_interleaved_batch_buffer_size returns the size of the temporary storage buffer that is required by rocsparse_sgtsv_interleaved_batch(), rocsparse_dgtsv_interleaved_batch(), rocsparse_cgtsv_interleaved_batch() and rocsparse_zgtsv_interleaved_batch(). The temporary storage buffer must be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • alg[in] Algorithm to use when solving tridiagonal systems. Options are thomas ( rocsparse_gtsv_interleaved_thomas ), LU ( rocsparse_gtsv_interleaved_lu ), or QR ( rocsparse_gtsv_interleaved_qr ). Passing rocsparse_gtsv_interleaved_default defaults the algorithm to use QR. Thomas algorithm is the fastest but is not stable while LU and QR are slower but are stable.

  • m[in] size of the tri-diagonal linear system.

  • dl[in] lower diagonal of tri-diagonal system. The first element of the lower diagonal must be zero.

  • d[in] main diagonal of tri-diagonal system.

  • du[in] upper diagonal of tri-diagonal system. The last element of the upper diagonal must be zero.

  • x[inout] Dense array of righthand-sides with dimension batch_stride by m.

  • batch_count[in] The number of systems to solve.

  • batch_stride[in] The number of elements that separate consecutive elements in a system. Must satisfy batch_stride >= batch_count.

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_sgtsv_interleaved_batch(), rocsparse_dgtsv_interleaved_batch(), rocsparse_cgtsv_interleaved_batch() and rocsparse_zgtsv_interleaved_batch().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, batch_count, batch_stride is invalid.

  • rocsparse_status_invalid_pointerdl, d, du, x or buffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_gtsv_interleaved_batch()#

rocsparse_status rocsparse_sgtsv_interleaved_batch(rocsparse_handle handle, rocsparse_gtsv_interleaved_alg alg, rocsparse_int m, float *dl, float *d, float *du, float *x, rocsparse_int batch_count, rocsparse_int batch_stride, void *temp_buffer)#
rocsparse_status rocsparse_dgtsv_interleaved_batch(rocsparse_handle handle, rocsparse_gtsv_interleaved_alg alg, rocsparse_int m, double *dl, double *d, double *du, double *x, rocsparse_int batch_count, rocsparse_int batch_stride, void *temp_buffer)#
rocsparse_status rocsparse_cgtsv_interleaved_batch(rocsparse_handle handle, rocsparse_gtsv_interleaved_alg alg, rocsparse_int m, rocsparse_float_complex *dl, rocsparse_float_complex *d, rocsparse_float_complex *du, rocsparse_float_complex *x, rocsparse_int batch_count, rocsparse_int batch_stride, void *temp_buffer)#
rocsparse_status rocsparse_zgtsv_interleaved_batch(rocsparse_handle handle, rocsparse_gtsv_interleaved_alg alg, rocsparse_int m, rocsparse_double_complex *dl, rocsparse_double_complex *d, rocsparse_double_complex *du, rocsparse_double_complex *x, rocsparse_int batch_count, rocsparse_int batch_stride, void *temp_buffer)#

Interleaved Batch tridiagonal solver.

rocsparse_gtsv_interleaved_batch solves a batched tridiagonal linear system. The routine requires a temporary storage buffer that must be allocated by the user. The size of this buffer can be determined by first calling rocsparse_gtsv_interleaved_batch_buffer_size. The user can specify different algorithms for rocsparse_gtsv_interleaved_batch to use. Options are thomas ( rocsparse_gtsv_interleaved_thomas ), LU ( rocsparse_gtsv_interleaved_lu ), or QR ( rocsparse_gtsv_interleaved_qr ). Passing rocsparse_gtsv_interleaved_default defaults the algorithm to use QR.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • alg[in] Algorithm to use when solving tridiagonal systems. Options are thomas ( rocsparse_gtsv_interleaved_thomas ), LU ( rocsparse_gtsv_interleaved_lu ), or QR ( rocsparse_gtsv_interleaved_qr ). Passing rocsparse_gtsv_interleaved_default defaults the algorithm to use QR. Thomas algorithm is the fastest but is not stable while LU and QR are slower but are stable.

  • m[in] size of the tri-diagonal linear system.

  • dl[inout] lower diagonal of tri-diagonal system. The first element of the lower diagonal must be zero.

  • d[inout] main diagonal of tri-diagonal system.

  • du[inout] upper diagonal of tri-diagonal system. The last element of the upper diagonal must be zero.

  • x[inout] Dense array of righthand-sides with dimension batch_stride by m.

  • batch_count[in] The number of systems to solve.

  • batch_stride[in] The number of elements that separate consecutive elements in a system. Must satisfy batch_stride >= batch_count.

  • temp_buffer[in] temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or batch_count or batch_stride is invalid.

  • rocsparse_status_invalid_pointerdl, d, du, x or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_gpsv_interleaved_batch_buffer_size()#

rocsparse_status rocsparse_sgpsv_interleaved_batch_buffer_size(rocsparse_handle handle, rocsparse_gpsv_interleaved_alg alg, rocsparse_int m, const float *ds, const float *dl, const float *d, const float *du, const float *dw, const float *x, rocsparse_int batch_count, rocsparse_int batch_stride, size_t *buffer_size)#
rocsparse_status rocsparse_dgpsv_interleaved_batch_buffer_size(rocsparse_handle handle, rocsparse_gpsv_interleaved_alg alg, rocsparse_int m, const double *ds, const double *dl, const double *d, const double *du, const double *dw, const double *x, rocsparse_int batch_count, rocsparse_int batch_stride, size_t *buffer_size)#
rocsparse_status rocsparse_cgpsv_interleaved_batch_buffer_size(rocsparse_handle handle, rocsparse_gpsv_interleaved_alg alg, rocsparse_int m, const rocsparse_float_complex *ds, const rocsparse_float_complex *dl, const rocsparse_float_complex *d, const rocsparse_float_complex *du, const rocsparse_float_complex *dw, const rocsparse_float_complex *x, rocsparse_int batch_count, rocsparse_int batch_stride, size_t *buffer_size)#
rocsparse_status rocsparse_zgpsv_interleaved_batch_buffer_size(rocsparse_handle handle, rocsparse_gpsv_interleaved_alg alg, rocsparse_int m, const rocsparse_double_complex *ds, const rocsparse_double_complex *dl, const rocsparse_double_complex *d, const rocsparse_double_complex *du, const rocsparse_double_complex *dw, const rocsparse_double_complex *x, rocsparse_int batch_count, rocsparse_int batch_stride, size_t *buffer_size)#

Batched Pentadiagonal solver.

rocsparse_gpsv_interleaved_batch_buffer_size calculates the required buffer size for rocsparse_gpsv_interleaved_batch(). It is the users responsibility to allocate this buffer.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • alg[in] algorithm to solve the linear system.

  • m[in] size of the pentadiagonal linear system.

  • ds[in] lower diagonal (distance 2) of pentadiagonal system. First two entries must be zero.

  • dl[in] lower diagonal of pentadiagonal system. First entry must be zero.

  • d[in] main diagonal of pentadiagonal system.

  • du[in] upper diagonal of pentadiagonal system. Last entry must be zero.

  • dw[in] upper diagonal (distance 2) of pentadiagonal system. Last two entries must be zero.

  • x[in] Dense array of right-hand-sides with dimension batch_stride by m.

  • batch_count[in] The number of systems to solve.

  • batch_stride[in] The number of elements that separate consecutive elements in a system. Must satisfy batch_stride >= batch_count.

  • buffer_size[out] Number of bytes of the temporary storage buffer required.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, alg, batch_count or batch_stride is invalid.

  • rocsparse_status_invalid_pointerds, dl, d, du, dw, x or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_gpsv_interleaved_batch()#

rocsparse_status rocsparse_sgpsv_interleaved_batch(rocsparse_handle handle, rocsparse_gpsv_interleaved_alg alg, rocsparse_int m, float *ds, float *dl, float *d, float *du, float *dw, float *x, rocsparse_int batch_count, rocsparse_int batch_stride, void *temp_buffer)#
rocsparse_status rocsparse_dgpsv_interleaved_batch(rocsparse_handle handle, rocsparse_gpsv_interleaved_alg alg, rocsparse_int m, double *ds, double *dl, double *d, double *du, double *dw, double *x, rocsparse_int batch_count, rocsparse_int batch_stride, void *temp_buffer)#
rocsparse_status rocsparse_cgpsv_interleaved_batch(rocsparse_handle handle, rocsparse_gpsv_interleaved_alg alg, rocsparse_int m, rocsparse_float_complex *ds, rocsparse_float_complex *dl, rocsparse_float_complex *d, rocsparse_float_complex *du, rocsparse_float_complex *dw, rocsparse_float_complex *x, rocsparse_int batch_count, rocsparse_int batch_stride, void *temp_buffer)#
rocsparse_status rocsparse_zgpsv_interleaved_batch(rocsparse_handle handle, rocsparse_gpsv_interleaved_alg alg, rocsparse_int m, rocsparse_double_complex *ds, rocsparse_double_complex *dl, rocsparse_double_complex *d, rocsparse_double_complex *du, rocsparse_double_complex *dw, rocsparse_double_complex *x, rocsparse_int batch_count, rocsparse_int batch_stride, void *temp_buffer)#

Batched Pentadiagonal solver.

rocsparse_gpsv_interleaved_batch solves a batch of pentadiagonal linear systems. The coefficient matrix of each pentadiagonal linear system is defined by five vectors for the lower part (ds, dl), main diagonal (d) and upper part (du, dw).

The function requires a temporary buffer. The size of the required buffer is returned by rocsparse_gpsv_interleaved_batch_buffer_size().

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

The routine is numerically stable because it uses QR to solve the linear systems.

Note

m need to be at least 3, to be a valid pentadiagonal matrix.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • alg[in] algorithm to solve the linear system.

  • m[in] size of the pentadiagonal linear system.

  • ds[inout] lower diagonal (distance 2) of pentadiagonal system. First two entries must be zero.

  • dl[inout] lower diagonal of pentadiagonal system. First entry must be zero.

  • d[inout] main diagonal of pentadiagonal system.

  • du[inout] upper diagonal of pentadiagonal system. Last entry must be zero.

  • dw[inout] upper diagonal (distance 2) of pentadiagonal system. Last two entries must be zero.

  • x[inout] Dense array of right-hand-sides with dimension batch_stride by m.

  • batch_count[in] The number of systems to solve.

  • batch_stride[in] The number of elements that separate consecutive elements in a system. Must satisfy batch_stride >= batch_count.

  • temp_buffer[in] Temporary storage buffer allocated by the user.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, alg, batch_count or batch_stride is invalid.

  • rocsparse_status_invalid_pointerds, dl, d, du, dw, x or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

Sparse Conversion Functions#

This module holds all sparse conversion routines.

The sparse conversion routines describe operations on a matrix in sparse format to obtain a matrix in a different sparse format.

rocsparse_csr2coo()#

rocsparse_status rocsparse_csr2coo(rocsparse_handle handle, const rocsparse_int *csr_row_ptr, rocsparse_int nnz, rocsparse_int m, rocsparse_int *coo_row_ind, rocsparse_index_base idx_base)#

Convert a sparse CSR matrix into a sparse COO matrix.

rocsparse_csr2coo converts the CSR array containing the row offsets, that point to the start of every row, into a COO array of row indices.

Example

This example converts a CSR matrix into a COO matrix.

//     1 2 0 3 0
// A = 0 4 5 0 0
//     6 0 0 7 8

rocsparse_int m   = 3;
rocsparse_int n   = 5;
rocsparse_int nnz = 8;

csr_row_ptr[m+1] = {0, 3, 5, 8};             // device memory
csr_col_ind[nnz] = {0, 1, 3, 1, 2, 0, 3, 4}; // device memory
csr_val[nnz]     = {1, 2, 3, 4, 5, 6, 7, 8}; // device memory

// Allocate COO matrix arrays
rocsparse_int* coo_row_ind;
rocsparse_int* coo_col_ind;
float* coo_val;

hipMalloc((void**)&coo_row_ind, sizeof(rocsparse_int) * nnz);
hipMalloc((void**)&coo_col_ind, sizeof(rocsparse_int) * nnz);
hipMalloc((void**)&coo_val, sizeof(float) * nnz);

// Convert the csr row offsets into coo row indices
rocsparse_csr2coo(handle,
                  csr_row_ptr,
                  nnz,
                  m,
                  coo_row_ind,
                  rocsparse_index_base_zero);

// Copy the column and value arrays
hipMemcpy(coo_col_ind,
          csr_col_ind,
          sizeof(rocsparse_int) * nnz,
          hipMemcpyDeviceToDevice);

hipMemcpy(coo_val,
          csr_val,
          sizeof(float) * nnz,
          hipMemcpyDeviceToDevice);

Note

It can also be used to convert a CSC array containing the column offsets into a COO array of column indices.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • m[in] number of rows of the sparse CSR matrix.

  • coo_row_ind[out] array of nnz elements containing the row indices of the sparse COO matrix.

  • idx_base[in] rocsparse_index_base_zero or rocsparse_index_base_one.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or nnz is invalid.

  • rocsparse_status_invalid_pointercsr_row_ptr or coo_row_ind pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

rocsparse_coo2csr()#

rocsparse_status rocsparse_coo2csr(rocsparse_handle handle, const rocsparse_int *coo_row_ind, rocsparse_int nnz, rocsparse_int m, rocsparse_int *csr_row_ptr, rocsparse_index_base idx_base)#

Convert a sparse COO matrix into a sparse CSR matrix.

rocsparse_coo2csr converts the COO array containing the row indices into a CSR array of row offsets, that point to the start of every row. It is assumed that the COO row index array is sorted.

Example

This example converts a COO matrix into a CSR matrix.

//     1 2 0 3 0
// A = 0 4 5 0 0
//     6 0 0 7 8

rocsparse_int m   = 3;
rocsparse_int n   = 5;
rocsparse_int nnz = 8;

coo_row_ind[nnz] = {0, 0, 0, 1, 1, 2, 2, 2}; // device memory
coo_col_ind[nnz] = {0, 1, 3, 1, 2, 0, 3, 4}; // device memory
coo_val[nnz]     = {1, 2, 3, 4, 5, 6, 7, 8}; // device memory

// Allocate CSR matrix arrays
rocsparse_int* csr_row_ptr;
rocsparse_int* csr_col_ind;
float* csr_val;

hipMalloc((void**)&csr_row_ptr, sizeof(rocsparse_int) * (m + 1));
hipMalloc((void**)&csr_col_ind, sizeof(rocsparse_int) * nnz);
hipMalloc((void**)&csr_val, sizeof(float) * nnz);

// Convert the coo row indices into csr row offsets
rocsparse_coo2csr(handle,
                  coo_row_ind,
                  nnz,
                  m,
                  csr_row_ptr,
                  rocsparse_index_base_zero);

// Copy the column and value arrays
hipMemcpy(csr_col_ind,
          coo_col_ind,
          sizeof(rocsparse_int) * nnz,
          hipMemcpyDeviceToDevice);

hipMemcpy(csr_val,
          coo_val,
          sizeof(float) * nnz,
          hipMemcpyDeviceToDevice);

Note

It can also be used, to convert a COO array containing the column indices into a CSC array of column offsets, that point to the start of every column. Then, it is assumed that the COO column index array is sorted, instead.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • coo_row_ind[in] array of nnz elements containing the row indices of the sparse COO matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • m[in] number of rows of the sparse CSR matrix.

  • csr_row_ptr[out] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • idx_base[in] rocsparse_index_base_zero or rocsparse_index_base_one.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or nnz is invalid.

  • rocsparse_status_invalid_pointercoo_row_ind or csr_row_ptr pointer is invalid.

rocsparse_csr2csc_buffer_size()#

rocsparse_status rocsparse_csr2csc_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_action copy_values, size_t *buffer_size)#

Convert a sparse CSR matrix into a sparse CSC matrix.

rocsparse_csr2csc_buffer_size returns the size of the temporary storage buffer required by rocsparse_scsr2csc(), rocsparse_dcsr2csc(), rocsparse_ccsr2csc() and rocsparse_zcsr2csc(). The temporary storage buffer must be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSR matrix.

  • n[in] number of columns of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • copy_values[in] rocsparse_action_symbolic or rocsparse_action_numeric.

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_scsr2csc(), rocsparse_dcsr2csc(), rocsparse_ccsr2csc() and rocsparse_zcsr2csc().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or nnz is invalid.

  • rocsparse_status_invalid_pointercsr_row_ptr, csr_col_ind or buffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_csr2csc()#

rocsparse_status rocsparse_scsr2csc(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, float *csc_val, rocsparse_int *csc_row_ind, rocsparse_int *csc_col_ptr, rocsparse_action copy_values, rocsparse_index_base idx_base, void *temp_buffer)#
rocsparse_status rocsparse_dcsr2csc(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, double *csc_val, rocsparse_int *csc_row_ind, rocsparse_int *csc_col_ptr, rocsparse_action copy_values, rocsparse_index_base idx_base, void *temp_buffer)#
rocsparse_status rocsparse_ccsr2csc(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_float_complex *csc_val, rocsparse_int *csc_row_ind, rocsparse_int *csc_col_ptr, rocsparse_action copy_values, rocsparse_index_base idx_base, void *temp_buffer)#
rocsparse_status rocsparse_zcsr2csc(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_double_complex *csc_val, rocsparse_int *csc_row_ind, rocsparse_int *csc_col_ptr, rocsparse_action copy_values, rocsparse_index_base idx_base, void *temp_buffer)#

Convert a sparse CSR matrix into a sparse CSC matrix.

rocsparse_csr2csc converts a CSR matrix into a CSC matrix. rocsparse_csr2csc can also be used to convert a CSC matrix into a CSR matrix. copy_values decides whether csc_val is being filled during conversion (rocsparse_action_numeric) or not (rocsparse_action_symbolic).

rocsparse_csr2csc requires extra temporary storage buffer that has to be allocated by the user. Storage buffer size can be determined by rocsparse_csr2csc_buffer_size().

Example

This example computes the transpose of a CSR matrix.

//     1 2 0 3 0
// A = 0 4 5 0 0
//     6 0 0 7 8

rocsparse_int m_A   = 3;
rocsparse_int n_A   = 5;
rocsparse_int nnz_A = 8;

csr_row_ptr_A[m+1] = {0, 3, 5, 8};             // device memory
csr_col_ind_A[nnz] = {0, 1, 3, 1, 2, 0, 3, 4}; // device memory
csr_val_A[nnz]     = {1, 2, 3, 4, 5, 6, 7, 8}; // device memory

// Allocate memory for transposed CSR matrix
rocsparse_int m_T   = n_A;
rocsparse_int n_T   = m_A;
rocsparse_int nnz_T = nnz_A;

rocsparse_int* csr_row_ptr_T;
rocsparse_int* csr_col_ind_T;
float* csr_val_T;

hipMalloc((void**)&csr_row_ptr_T, sizeof(rocsparse_int) * (m_T + 1));
hipMalloc((void**)&csr_col_ind_T, sizeof(rocsparse_int) * nnz_T);
hipMalloc((void**)&csr_val_T, sizeof(float) * nnz_T);

// Obtain the temporary buffer size
size_t buffer_size;
rocsparse_csr2csc_buffer_size(handle,
                              m_A,
                              n_A,
                              nnz_A,
                              csr_row_ptr_A,
                              csr_col_ind_A,
                              rocsparse_action_numeric,
                              &buffer_size);

// Allocate temporary buffer
void* temp_buffer;
hipMalloc(&temp_buffer, buffer_size);

rocsparse_scsr2csc(handle,
                   m_A,
                   n_A,
                   nnz_A,
                   csr_val_A,
                   csr_row_ptr_A,
                   csr_col_ind_A,
                   csr_val_T,
                   csr_col_ind_T,
                   csr_row_ptr_T,
                   rocsparse_action_numeric,
                   rocsparse_index_base_zero,
                   temp_buffer);

Note

The resulting matrix can also be seen as the transpose of the input matrix.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSR matrix.

  • n[in] number of columns of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • csr_val[in] array of nnz elements of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • csc_val[out] array of nnz elements of the sparse CSC matrix.

  • csc_row_ind[out] array of nnz elements containing the row indices of the sparse CSC matrix.

  • csc_col_ptr[out] array of n+1 elements that point to the start of every column of the sparse CSC matrix.

  • copy_values[in] rocsparse_action_symbolic or rocsparse_action_numeric.

  • idx_base[in] rocsparse_index_base_zero or rocsparse_index_base_one.

  • temp_buffer[in] temporary storage buffer allocated by the user, size is returned by rocsparse_csr2csc_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or nnz is invalid.

  • rocsparse_status_invalid_pointercsr_val, csr_row_ptr, csr_col_ind, csc_val, csc_row_ind, csc_col_ptr or temp_buffer pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_gebsr2gebsc_buffer_size()#

rocsparse_status rocsparse_sgebsr2gebsc_buffer_size(rocsparse_handle handle, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, size_t *p_buffer_size)#
rocsparse_status rocsparse_dgebsr2gebsc_buffer_size(rocsparse_handle handle, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, size_t *p_buffer_size)#
rocsparse_status rocsparse_cgebsr2gebsc_buffer_size(rocsparse_handle handle, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, size_t *p_buffer_size)#
rocsparse_status rocsparse_zgebsr2gebsc_buffer_size(rocsparse_handle handle, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, size_t *p_buffer_size)#

Convert a sparse GEneral BSR matrix into a sparse GEneral BSC matrix.

rocsparse_gebsr2gebsc_buffer_size returns the size of the temporary storage buffer required by rocsparse_sgebsr2gebsc(), rocsparse_dgebsr2gebsc(), rocsparse_cgebsr2gebsc() and rocsparse_zgebsr2gebsc(). The temporary storage buffer must be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • mb[in] number of rows of the sparse GEneral BSR matrix.

  • nb[in] number of columns of the sparse GEneral BSR matrix.

  • nnzb[in] number of non-zero entries of the sparse GEneral BSR matrix.

  • bsr_val[in] array of nnzb*row_block_dim*col_block_dim containing the values of the sparse GEneral BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every row of the sparse GEneral BSR matrix.

  • bsr_col_ind[in] array of nnzb elements containing the column indices of the sparse GEneral BSR matrix.

  • row_block_dim[in] row size of the blocks in the sparse general BSR matrix.

  • col_block_dim[in] col size of the blocks in the sparse general BSR matrix.

  • p_buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_sgebsr2gebsc(), rocsparse_dgebsr2gebsc(), rocsparse_cgebsr2gebsc() and rocsparse_zgebsr2gebsc().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, nb or nnzb is invalid.

  • rocsparse_status_invalid_pointerbsr_row_ptr, bsr_col_ind or buffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_gebsr2gebsc()#

rocsparse_status rocsparse_sgebsr2gebsc(rocsparse_handle handle, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, float *bsc_val, rocsparse_int *bsc_row_ind, rocsparse_int *bsc_col_ptr, rocsparse_action copy_values, rocsparse_index_base idx_base, void *temp_buffer)#
rocsparse_status rocsparse_dgebsr2gebsc(rocsparse_handle handle, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, double *bsc_val, rocsparse_int *bsc_row_ind, rocsparse_int *bsc_col_ptr, rocsparse_action copy_values, rocsparse_index_base idx_base, void *temp_buffer)#
rocsparse_status rocsparse_cgebsr2gebsc(rocsparse_handle handle, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, rocsparse_float_complex *bsc_val, rocsparse_int *bsc_row_ind, rocsparse_int *bsc_col_ptr, rocsparse_action copy_values, rocsparse_index_base idx_base, void *temp_buffer)#
rocsparse_status rocsparse_zgebsr2gebsc(rocsparse_handle handle, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, rocsparse_double_complex *bsc_val, rocsparse_int *bsc_row_ind, rocsparse_int *bsc_col_ptr, rocsparse_action copy_values, rocsparse_index_base idx_base, void *temp_buffer)#

Convert a sparse GEneral BSR matrix into a sparse GEneral BSC matrix.

rocsparse_gebsr2gebsc converts a GEneral BSR matrix into a GEneral BSC matrix. rocsparse_gebsr2gebsc can also be used to convert a GEneral BSC matrix into a GEneral BSR matrix. copy_values decides whether bsc_val is being filled during conversion (rocsparse_action_numeric) or not (rocsparse_action_symbolic).

rocsparse_gebsr2gebsc requires extra temporary storage buffer that has to be allocated by the user. Storage buffer size can be determined by rocsparse_gebsr2gebsc_buffer_size().

Example

This example computes the transpose of a GEneral BSR matrix.

//     1 2 0 3
// A = 0 4 5 0
//     6 0 0 7
//     1 2 3 4

rocsparse_int mb_A   = 2;
rocsparse_int row_block_dim = 2;
rocsparse_int col_block_dim = 2;
rocsparse_int nb_A   = 2;
rocsparse_int nnzb_A = 4;

bsr_row_ptr_A[mb_A+1] = {0, 2, 4};               // device memory
bsr_col_ind_A[nnzb_A] = {0, 1, 0, 1}; // device memory
bsr_val_A[nnzb_A]     = {1, 0, 2, 4, 0, 5, 3, 0, 6, 1, 0, 2, 0, 3, 7, 4}; // device memory

// Allocate memory for transposed BSR matrix
rocsparse_int mb_T   = nb_A;
rocsparse_int nb_T   = mb_A;
rocsparse_int nnzb_T = nnzb_A;

rocsparse_int* bsr_row_ptr_T;
rocsparse_int* bsr_col_ind_T;
float* bsr_val_T;

hipMalloc((void**)&bsr_row_ptr_T, sizeof(rocsparse_int) * (mb_T + 1));
hipMalloc((void**)&bsr_col_ind_T, sizeof(rocsparse_int) * nnzb_T);
hipMalloc((void**)&bsr_val_T, sizeof(float) * nnzb_T);

// Obtain the temporary buffer size
size_t buffer_size;
rocsparse_gebsr2gebsc_buffer_size(handle,
                              mb_A,
                              nb_A,
                              nnzb_A,
                              bsr_row_ptr_A,
                              bsr_col_ind_A,
                              rocsparse_action_numeric,
                              &buffer_size);

// Allocate temporary buffer
void* temp_buffer;
hipMalloc(&temp_buffer, buffer_size);

rocsparse_sgebsr2gebsc(handle,
                   mb_A,
                   nb_A,
                   nnzb_A,
                   bsr_val_A,
                   bsr_row_ptr_A,
                   bsr_col_ind_A,
                   row_block_dim,
                   col_block_dim,
                   bsr_val_T,
                   bsr_col_ind_T,
                   bsr_row_ptr_T,
                   rocsparse_action_numeric,
                   rocsparse_index_base_zero,
                   temp_buffer);

Note

The resulting matrix can also be seen as the transpose of the input matrix.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • mb[in] number of rows of the sparse GEneral BSR matrix.

  • nb[in] number of columns of the sparse GEneral BSR matrix.

  • nnzb[in] number of non-zero entries of the sparse GEneral BSR matrix.

  • bsr_val[in] array of nnzb * row_block_dim * col_block_dim elements of the sparse GEneral BSR matrix.

  • bsr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse GEneral BSR matrix.

  • bsr_col_ind[in] array of nnz elements containing the column indices of the sparse GEneral BSR matrix.

  • row_block_dim[in] row size of the blocks in the sparse general BSR matrix.

  • col_block_dim[in] col size of the blocks in the sparse general BSR matrix.

  • bsc_val[out] array of nnz elements of the sparse BSC matrix.

  • bsc_row_ind[out] array of nnz elements containing the row indices of the sparse BSC matrix.

  • bsc_col_ptr[out] array of n+1 elements that point to the start of every column of the sparse BSC matrix.

  • copy_values[in] rocsparse_action_symbolic or rocsparse_action_numeric.

  • idx_base[in] rocsparse_index_base_zero or rocsparse_index_base_one.

  • temp_buffer[in] temporary storage buffer allocated by the user, size is returned by rocsparse_gebsr2gebsc_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb, nb or nnzb is invalid.

  • rocsparse_status_invalid_pointerbsr_val, bsr_row_ptr, bsr_col_ind, bsc_val, bsc_row_ind, bsc_col_ptr or temp_buffer pointer is invalid.

  • rocsparse_status_arch_mismatch – the device is not supported.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_csr2ell_width()#

rocsparse_status rocsparse_csr2ell_width(rocsparse_handle handle, rocsparse_int m, const rocsparse_mat_descr csr_descr, const rocsparse_int *csr_row_ptr, const rocsparse_mat_descr ell_descr, rocsparse_int *ell_width)#

Convert a sparse CSR matrix into a sparse ELL matrix.

rocsparse_csr2ell_width computes the maximum of the per row non-zero elements over all rows, the ELL width, for a given CSR matrix.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSR matrix.

  • csr_descr[in] descriptor of the sparse CSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • ell_descr[in] descriptor of the sparse ELL matrix. Currently, only rocsparse_matrix_type_general is supported.

  • ell_width[out] pointer to the number of non-zero elements per row in ELL storage format.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem is invalid.

  • rocsparse_status_invalid_pointercsr_descr, csr_row_ptr, or ell_width pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csr2ell()#

rocsparse_status rocsparse_scsr2ell(rocsparse_handle handle, rocsparse_int m, const rocsparse_mat_descr csr_descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_mat_descr ell_descr, rocsparse_int ell_width, float *ell_val, rocsparse_int *ell_col_ind)#
rocsparse_status rocsparse_dcsr2ell(rocsparse_handle handle, rocsparse_int m, const rocsparse_mat_descr csr_descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_mat_descr ell_descr, rocsparse_int ell_width, double *ell_val, rocsparse_int *ell_col_ind)#
rocsparse_status rocsparse_ccsr2ell(rocsparse_handle handle, rocsparse_int m, const rocsparse_mat_descr csr_descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_mat_descr ell_descr, rocsparse_int ell_width, rocsparse_float_complex *ell_val, rocsparse_int *ell_col_ind)#
rocsparse_status rocsparse_zcsr2ell(rocsparse_handle handle, rocsparse_int m, const rocsparse_mat_descr csr_descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_mat_descr ell_descr, rocsparse_int ell_width, rocsparse_double_complex *ell_val, rocsparse_int *ell_col_ind)#

Convert a sparse CSR matrix into a sparse ELL matrix.

rocsparse_csr2ell converts a CSR matrix into an ELL matrix. It is assumed, that ell_val and ell_col_ind are allocated. Allocation size is computed by the number of rows times the number of ELL non-zero elements per row, such that \(\text{nnz}_{\text{ELL}} = m \cdot \text{ell_width}\). The number of ELL non-zero elements per row is obtained by rocsparse_csr2ell_width().

Example

This example converts a CSR matrix into an ELL matrix.

//     1 2 0 3 0
// A = 0 4 5 0 0
//     6 0 0 7 8

rocsparse_int m   = 3;
rocsparse_int n   = 5;
rocsparse_int nnz = 8;

csr_row_ptr[m+1] = {0, 3, 5, 8};             // device memory
csr_col_ind[nnz] = {0, 1, 3, 1, 2, 0, 3, 4}; // device memory
csr_val[nnz]     = {1, 2, 3, 4, 5, 6, 7, 8}; // device memory

// Create ELL matrix descriptor
rocsparse_mat_descr ell_descr;
rocsparse_create_mat_descr(&ell_descr);

// Obtain the ELL width
rocsparse_int ell_width;
rocsparse_csr2ell_width(handle,
                        m,
                        csr_descr,
                        csr_row_ptr,
                        ell_descr,
                        &ell_width);

// Compute ELL non-zero entries
rocsparse_int ell_nnz = m * ell_width;

// Allocate ELL column and value arrays
rocsparse_int* ell_col_ind;
hipMalloc((void**)&ell_col_ind, sizeof(rocsparse_int) * ell_nnz);

float* ell_val;
hipMalloc((void**)&ell_val, sizeof(float) * ell_nnz);

// Format conversion
rocsparse_scsr2ell(handle,
                   m,
                   csr_descr,
                   csr_val,
                   csr_row_ptr,
                   csr_col_ind,
                   ell_descr,
                   ell_width,
                   ell_val,
                   ell_col_ind);

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSR matrix.

  • csr_descr[in] descriptor of the sparse CSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val[in] array containing the values of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array containing the column indices of the sparse CSR matrix.

  • ell_descr[in] descriptor of the sparse ELL matrix. Currently, only rocsparse_matrix_type_general is supported.

  • ell_width[in] number of non-zero elements per row in ELL storage format.

  • ell_val[out] array of m times ell_width elements of the sparse ELL matrix.

  • ell_col_ind[out] array of m times ell_width elements containing the column indices of the sparse ELL matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or ell_width is invalid.

  • rocsparse_status_invalid_pointercsr_descr, csr_val, csr_row_ptr, csr_col_ind, ell_descr, ell_val or ell_col_ind pointer is invalid.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_ell2csr_nnz()#

rocsparse_status rocsparse_ell2csr_nnz(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr ell_descr, rocsparse_int ell_width, const rocsparse_int *ell_col_ind, const rocsparse_mat_descr csr_descr, rocsparse_int *csr_row_ptr, rocsparse_int *csr_nnz)#

Convert a sparse ELL matrix into a sparse CSR matrix.

rocsparse_ell2csr_nnz computes the total CSR non-zero elements and the CSR row offsets, that point to the start of every row of the sparse CSR matrix, for a given ELL matrix. It is assumed that csr_row_ptr has been allocated with size m+1.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse ELL matrix.

  • n[in] number of columns of the sparse ELL matrix.

  • ell_descr[in] descriptor of the sparse ELL matrix. Currently, only rocsparse_matrix_type_general is supported.

  • ell_width[in] number of non-zero elements per row in ELL storage format.

  • ell_col_ind[in] array of m times ell_width elements containing the column indices of the sparse ELL matrix.

  • csr_descr[in] descriptor of the sparse CSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • csr_row_ptr[out] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_nnz[out] pointer to the total number of non-zero elements in CSR storage format.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or ell_width is invalid.

  • rocsparse_status_invalid_pointerell_descr, ell_col_ind, csr_descr, csr_row_ptr or csr_nnz pointer is invalid.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_ell2csr()#

rocsparse_status rocsparse_sell2csr(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr ell_descr, rocsparse_int ell_width, const float *ell_val, const rocsparse_int *ell_col_ind, const rocsparse_mat_descr csr_descr, float *csr_val, const rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind)#
rocsparse_status rocsparse_dell2csr(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr ell_descr, rocsparse_int ell_width, const double *ell_val, const rocsparse_int *ell_col_ind, const rocsparse_mat_descr csr_descr, double *csr_val, const rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind)#
rocsparse_status rocsparse_cell2csr(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr ell_descr, rocsparse_int ell_width, const rocsparse_float_complex *ell_val, const rocsparse_int *ell_col_ind, const rocsparse_mat_descr csr_descr, rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind)#
rocsparse_status rocsparse_zell2csr(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr ell_descr, rocsparse_int ell_width, const rocsparse_double_complex *ell_val, const rocsparse_int *ell_col_ind, const rocsparse_mat_descr csr_descr, rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind)#

Convert a sparse ELL matrix into a sparse CSR matrix.

rocsparse_ell2csr converts an ELL matrix into a CSR matrix. It is assumed that csr_row_ptr has already been filled and that csr_val and csr_col_ind are allocated by the user. csr_row_ptr and allocation size of csr_col_ind and csr_val is defined by the number of CSR non-zero elements. Both can be obtained by rocsparse_ell2csr_nnz().

Example

This example converts an ELL matrix into a CSR matrix.

//     1 2 0 3 0
// A = 0 4 5 0 0
//     6 0 0 7 8

rocsparse_int m         = 3;
rocsparse_int n         = 5;
rocsparse_int nnz       = 9;
rocsparse_int ell_width = 3;

ell_col_ind[nnz] = {0, 1, 0, 1, 2, 3, 3, -1, 4}; // device memory
ell_val[nnz]     = {1, 4, 6, 2, 5, 7, 3, 0, 8};  // device memory

// Create CSR matrix descriptor
rocsparse_mat_descr csr_descr;
rocsparse_create_mat_descr(&csr_descr);

// Allocate csr_row_ptr array for row offsets
rocsparse_int* csr_row_ptr;
hipMalloc((void**)&csr_row_ptr, sizeof(rocsparse_int) * (m + 1));

// Obtain the number of CSR non-zero entries
// and fill csr_row_ptr array with row offsets
rocsparse_int csr_nnz;
rocsparse_ell2csr_nnz(handle,
                      m,
                      n,
                      ell_descr,
                      ell_width,
                      ell_col_ind,
                      csr_descr,
                      csr_row_ptr,
                      &csr_nnz);

// Allocate CSR column and value arrays
rocsparse_int* csr_col_ind;
hipMalloc((void**)&csr_col_ind, sizeof(rocsparse_int) * csr_nnz);

float* csr_val;
hipMalloc((void**)&csr_val, sizeof(float) * csr_nnz);

// Format conversion
rocsparse_sell2csr(handle,
                   m,
                   n,
                   ell_descr,
                   ell_width,
                   ell_val,
                   ell_col_ind,
                   csr_descr,
                   csr_val,
                   csr_row_ptr,
                   csr_col_ind);

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse ELL matrix.

  • n[in] number of columns of the sparse ELL matrix.

  • ell_descr[in] descriptor of the sparse ELL matrix. Currently, only rocsparse_matrix_type_general is supported.

  • ell_width[in] number of non-zero elements per row in ELL storage format.

  • ell_val[in] array of m times ell_width elements of the sparse ELL matrix.

  • ell_col_ind[in] array of m times ell_width elements containing the column indices of the sparse ELL matrix.

  • csr_descr[in] descriptor of the sparse CSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val[out] array containing the values of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[out] array containing the column indices of the sparse CSR matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or ell_width is invalid.

  • rocsparse_status_invalid_pointercsr_descr, csr_val, csr_row_ptr, csr_col_ind, ell_descr, ell_val or ell_col_ind pointer is invalid.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_csr2hyb()#

rocsparse_status rocsparse_scsr2hyb(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_hyb_mat hyb, rocsparse_int user_ell_width, rocsparse_hyb_partition partition_type)#
rocsparse_status rocsparse_dcsr2hyb(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_hyb_mat hyb, rocsparse_int user_ell_width, rocsparse_hyb_partition partition_type)#
rocsparse_status rocsparse_ccsr2hyb(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_hyb_mat hyb, rocsparse_int user_ell_width, rocsparse_hyb_partition partition_type)#
rocsparse_status rocsparse_zcsr2hyb(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_hyb_mat hyb, rocsparse_int user_ell_width, rocsparse_hyb_partition partition_type)#

Convert a sparse CSR matrix into a sparse HYB matrix.

rocsparse_csr2hyb converts a CSR matrix into a HYB matrix. It is assumed that hyb has been initialized with rocsparse_create_hyb_mat().

Example

This example converts a CSR matrix into a HYB matrix using user defined partitioning.

// Create HYB matrix structure
rocsparse_hyb_mat hyb;
rocsparse_create_hyb_mat(&hyb);

// User defined ell width
rocsparse_int user_ell_width = 5;

// Perform the conversion
rocsparse_scsr2hyb(handle,
                   m,
                   n,
                   descr,
                   csr_val,
                   csr_row_ptr,
                   csr_col_ind,
                   hyb,
                   user_ell_width,
                   rocsparse_hyb_partition_user);

// Do some work

// Clean up
rocsparse_destroy_hyb_mat(hyb);

Note

This function requires a significant amount of storage for the HYB matrix, depending on the matrix structure.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSR matrix.

  • n[in] number of columns of the sparse CSR matrix.

  • descr[in] descriptor of the sparse CSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val[in] array containing the values of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array containing the column indices of the sparse CSR matrix.

  • hyb[out] sparse matrix in HYB format.

  • user_ell_width[in] width of the ELL part of the HYB matrix (only required if partition_type == rocsparse_hyb_partition_user).

  • partition_type[in] rocsparse_hyb_partition_auto (recommended), rocsparse_hyb_partition_user or rocsparse_hyb_partition_max.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or user_ell_width is invalid.

  • rocsparse_status_invalid_valuepartition_type is invalid.

  • rocsparse_status_invalid_pointerdescr, hyb, csr_val, csr_row_ptr or csr_col_ind pointer is invalid.

  • rocsparse_status_memory_error – the buffer for the HYB matrix could not be allocated.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_hyb2csr_buffer_size()#

rocsparse_status rocsparse_hyb2csr_buffer_size(rocsparse_handle handle, const rocsparse_mat_descr descr, const rocsparse_hyb_mat hyb, const rocsparse_int *csr_row_ptr, size_t *buffer_size)#

Convert a sparse HYB matrix into a sparse CSR matrix.

rocsparse_hyb2csr_buffer_size returns the size of the temporary storage buffer required by rocsparse_shyb2csr(), rocsparse_dhyb2csr(), rocsparse_chyb2csr() and rocsparse_dhyb2csr(). The temporary storage buffer must be allocated by the user.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerdescr, hyb, csr_row_ptr or buffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_hyb2csr()#

rocsparse_status rocsparse_shyb2csr(rocsparse_handle handle, const rocsparse_mat_descr descr, const rocsparse_hyb_mat hyb, float *csr_val, rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind, void *temp_buffer)#
rocsparse_status rocsparse_dhyb2csr(rocsparse_handle handle, const rocsparse_mat_descr descr, const rocsparse_hyb_mat hyb, double *csr_val, rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind, void *temp_buffer)#
rocsparse_status rocsparse_chyb2csr(rocsparse_handle handle, const rocsparse_mat_descr descr, const rocsparse_hyb_mat hyb, rocsparse_float_complex *csr_val, rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind, void *temp_buffer)#
rocsparse_status rocsparse_zhyb2csr(rocsparse_handle handle, const rocsparse_mat_descr descr, const rocsparse_hyb_mat hyb, rocsparse_double_complex *csr_val, rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind, void *temp_buffer)#

Convert a sparse HYB matrix into a sparse CSR matrix.

rocsparse_hyb2csr converts a HYB matrix into a CSR matrix.

rocsparse_hyb2csr requires extra temporary storage buffer that has to be allocated by the user. Storage buffer size can be determined by rocsparse_hyb2csr_buffer_size().

Example

This example converts a HYB matrix into a CSR matrix.

// Create CSR matrix arrays
rocsparse_int* csr_row_ptr;
rocsparse_int* csr_col_ind;
float* csr_val;

hipMalloc((void**)&csr_row_ptr, sizeof(rocsparse_int) * (m + 1));
hipMalloc((void**)&csr_col_ind, sizeof(rocsparse_int) * nnz);
hipMalloc((void**)&csr_val, sizeof(float) * nnz);

// Get required size of temporary buffer
size_t size;
rocsparse_hyb2csr_buffer_size(handle,
                              descr,
                              hyb,
                              csr_row_ptr,
                              &size);

// Allocate temporary buffer
void* buffer;
hipMalloc(&buffer, size);

// Perform the conversion
rocsparse_shyb2csr(handle,
                   descr,
                   hyb,
                   csr_val,
                   csr_row_ptr,
                   csr_col_ind,
                   buffer);

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • descr[in] descriptor of the sparse HYB matrix. Currently, only rocsparse_matrix_type_general is supported.

  • hyb[in] sparse matrix in HYB format.

  • csr_val[out] array containing the values of the sparse CSR matrix.

  • csr_row_ptr[out] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[out] array containing the column indices of the sparse CSR matrix.

  • temp_buffer[in] temporary storage buffer allocated by the user, size is returned by rocsparse_hyb2csr_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerdescr, hyb, csr_val, csr_row_ptr, csr_col_ind or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_bsr2csr()#

rocsparse_status rocsparse_sbsr2csr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, const rocsparse_mat_descr bsr_descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, const rocsparse_mat_descr csr_descr, float *csr_val, rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind)#
rocsparse_status rocsparse_dbsr2csr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, const rocsparse_mat_descr bsr_descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, const rocsparse_mat_descr csr_descr, double *csr_val, rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind)#
rocsparse_status rocsparse_cbsr2csr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, const rocsparse_mat_descr bsr_descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, const rocsparse_mat_descr csr_descr, rocsparse_float_complex *csr_val, rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind)#
rocsparse_status rocsparse_zbsr2csr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, const rocsparse_mat_descr bsr_descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int block_dim, const rocsparse_mat_descr csr_descr, rocsparse_double_complex *csr_val, rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind)#

Convert a sparse BSR matrix into a sparse CSR matrix.

rocsparse_bsr2csr converts a BSR matrix into a CSR matrix. It is assumed, that csr_val, csr_col_ind and csr_row_ptr are allocated. Allocation size for csr_row_ptr is computed by the number of block rows multiplied by the block dimension plus one. Allocation for csr_val and csr_col_ind is computed by the the number of blocks in the BSR matrix multiplied by the block dimension squared.

Example

This example converts a BSR matrix into an CSR matrix.

//     1 4 0 0 0 0
// A = 0 2 3 0 0 0
//     5 0 0 7 8 0
//     0 0 9 0 6 0

rocsparse_int mb   = 2;
rocsparse_int nb   = 3;
rocsparse_int block_dim = 2;
rocsparse_int m = Mb * block_dim;
rocsparse_int n = Nb * block_dim;

bsr_row_ptr[mb+1]                 = {0, 2, 5};                                                    // device memory
bsr_col_ind[nnzb]                 = {0, 1, 0, 1, 2};                                              // device memory
bsr_val[nnzb*block_dim*block_dim] = {1, 0, 4, 2, 0, 3, 0, 0, 5, 0, 0, 0, 0, 9, 7, 0, 8, 6, 0, 0}; // device memory

rocsparse_int nnzb = bsr_row_ptr[mb] - bsr_row_ptr[0];

// Create CSR arrays on device
rocsparse_int* csr_row_ptr;
rocsparse_int* csr_col_ind;
float* csr_val;
hipMalloc((void**)&csr_row_ptr, sizeof(rocsparse_int) * (m + 1));
hipMalloc((void**)&csr_col_ind, sizeof(rocsparse_int) * nnzb * block_dim * block_dim);
hipMalloc((void**)&csr_val, sizeof(float) * nnzb * block_dim * block_dim);

// Create rocsparse handle
rocsparse_local_handle handle;

rocsparse_mat_descr bsr_descr = nullptr;
rocsparse_create_mat_descr(&bsr_descr);

rocsparse_mat_descr csr_descr = nullptr;
rocsparse_create_mat_descr(&csr_descr);

rocsparse_set_mat_index_base(bsr_descr, rocsparse_index_base_zero);
rocsparse_set_mat_index_base(csr_descr, rocsparse_index_base_zero);

// Format conversion
rocsparse_sbsr2csr(handle,
                   rocsparse_direction_column,
                   mb,
                   nb,
                   bsr_descr,
                   bsr_val,
                   bsr_row_ptr,
                   bsr_col_ind,
                   block_dim,
                   csr_descr,
                   csr_val,
                   csr_row_ptr,
                   csr_col_ind);

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] the storage format of the blocks, rocsparse_direction_row or rocsparse_direction_column

  • mb[in] number of block rows in the sparse BSR matrix.

  • nb[in] number of block columns in the sparse BSR matrix.

  • bsr_descr[in] descriptor of the sparse BSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • bsr_val[in] array of nnzb*block_dim*block_dim containing the values of the sparse BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnzb elements containing the block column indices of the sparse BSR matrix.

  • block_dim[in] size of the blocks in the sparse BSR matrix.

  • csr_descr[in] descriptor of the sparse CSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val[out] array of nnzb*block_dim*block_dim elements containing the values of the sparse CSR matrix.

  • csr_row_ptr[out] array of m+1 where m=mb*block_dim elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[out] array of nnzb*block_dim*block_dim elements containing the column indices of the sparse CSR matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb or nb or block_dim is invalid.

  • rocsparse_status_invalid_pointerbsr_val, bsr_row_ptr, bsr_col_ind, csr_val, csr_row_ptr or csr_col_ind pointer is invalid.

rocsparse_gebsr2csr()#

rocsparse_status rocsparse_sgebsr2csr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, const rocsparse_mat_descr bsr_descr, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const rocsparse_mat_descr csr_descr, float *csr_val, rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind)#
rocsparse_status rocsparse_dgebsr2csr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, const rocsparse_mat_descr bsr_descr, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const rocsparse_mat_descr csr_descr, double *csr_val, rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind)#
rocsparse_status rocsparse_cgebsr2csr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, const rocsparse_mat_descr bsr_descr, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const rocsparse_mat_descr csr_descr, rocsparse_float_complex *csr_val, rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind)#
rocsparse_status rocsparse_zgebsr2csr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, const rocsparse_mat_descr bsr_descr, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const rocsparse_mat_descr csr_descr, rocsparse_double_complex *csr_val, rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind)#

Convert a sparse general BSR matrix into a sparse CSR matrix.

rocsparse_gebsr2csr converts a BSR matrix into a CSR matrix. It is assumed, that csr_val, csr_col_ind and csr_row_ptr are allocated. Allocation size for csr_row_ptr is computed by the number of block rows multiplied by the block dimension plus one. Allocation for csr_val and csr_col_ind is computed by the the number of blocks in the BSR matrix multiplied by the product of the block dimensions.

Example

This example converts a general BSR matrix into an CSR matrix.

//     1 4 0 0 0 0
// A = 0 2 3 0 0 0
//     5 0 0 7 8 0
//     0 0 9 0 6 0

rocsparse_int mb   = 2;
rocsparse_int nb   = 2;
rocsparse_int row_block_dim = 2;
rocsparse_int col_block_dim = 3;
rocsparse_int m = Mb * row_block_dim;
rocsparse_int n = Nb * col_block_dim;

bsr_row_ptr[mb+1]                 = {0, 1, 3};                                              // device memory
bsr_col_ind[nnzb]                 = {0, 0, 1};                                              // device memory
bsr_val[nnzb*block_dim*block_dim] = {1, 0, 4, 2, 0, 3, 5, 0, 0, 0, 0, 9, 7, 0, 8, 6, 0, 0}; // device memory

rocsparse_int nnzb = bsr_row_ptr[mb] - bsr_row_ptr[0];

// Create CSR arrays on device
rocsparse_int* csr_row_ptr;
rocsparse_int* csr_col_ind;
float* csr_val;
hipMalloc((void**)&csr_row_ptr, sizeof(rocsparse_int) * (m + 1));
hipMalloc((void**)&csr_col_ind, sizeof(rocsparse_int) * nnzb * row_block_dim * col_block_dim);
hipMalloc((void**)&csr_val, sizeof(float) * nnzb * row_block_dim * col_block_dim);

// Create rocsparse handle
rocsparse_local_handle handle;

rocsparse_mat_descr bsr_descr = nullptr;
rocsparse_create_mat_descr(&bsr_descr);

rocsparse_mat_descr csr_descr = nullptr;
rocsparse_create_mat_descr(&csr_descr);

rocsparse_set_mat_index_base(bsr_descr, rocsparse_index_base_zero);
rocsparse_set_mat_index_base(csr_descr, rocsparse_index_base_zero);

// Format conversion
rocsparse_sgebsr2csr(handle,
                   rocsparse_direction_column,
                   mb,
                   nb,
                   bsr_descr,
                   bsr_val,
                   bsr_row_ptr,
                   bsr_col_ind,
                   row_block_dim,
                   col_block_dim,
                   csr_descr,
                   csr_val,
                   csr_row_ptr,
                   csr_col_ind);

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] the storage format of the blocks, rocsparse_direction_row or rocsparse_direction_column

  • mb[in] number of block rows in the sparse general BSR matrix.

  • nb[in] number of block columns in the sparse general BSR matrix.

  • bsr_descr[in] descriptor of the sparse general BSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • bsr_val[in] array of nnzb*row_block_dim*col_block_dim containing the values of the sparse BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnzb elements containing the block column indices of the sparse BSR matrix.

  • row_block_dim[in] row size of the blocks in the sparse general BSR matrix.

  • col_block_dim[in] column size of the blocks in the sparse general BSR matrix.

  • csr_descr[in] descriptor of the sparse CSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val[out] array of nnzb*row_block_dim*col_block_dim elements containing the values of the sparse CSR matrix.

  • csr_row_ptr[out] array of m+1 where m=mb*row_block_dim elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[out] array of nnzb*block_dim*block_dim elements containing the column indices of the sparse CSR matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb or nb or block_dim is invalid.

  • rocsparse_status_invalid_pointerbsr_val, bsr_row_ptr, bsr_col_ind, csr_val, csr_row_ptr or csr_col_ind pointer is invalid.

rocsparse_gebsr2gebsr_buffer_size()#

rocsparse_status rocsparse_sgebsr2gebsr_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_mat_descr descr_A, const float *bsr_val_A, const rocsparse_int *bsr_row_ptr_A, const rocsparse_int *bsr_col_ind_A, rocsparse_int row_block_dim_A, rocsparse_int col_block_dim_A, rocsparse_int row_block_dim_C, rocsparse_int col_block_dim_C, size_t *buffer_size)#
rocsparse_status rocsparse_dgebsr2gebsr_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_mat_descr descr_A, const double *bsr_val_A, const rocsparse_int *bsr_row_ptr_A, const rocsparse_int *bsr_col_ind_A, rocsparse_int row_block_dim_A, rocsparse_int col_block_dim_A, rocsparse_int row_block_dim_C, rocsparse_int col_block_dim_C, size_t *buffer_size)#
rocsparse_status rocsparse_cgebsr2gebsr_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_mat_descr descr_A, const rocsparse_float_complex *bsr_val_A, const rocsparse_int *bsr_row_ptr_A, const rocsparse_int *bsr_col_ind_A, rocsparse_int row_block_dim_A, rocsparse_int col_block_dim_A, rocsparse_int row_block_dim_C, rocsparse_int col_block_dim_C, size_t *buffer_size)#
rocsparse_status rocsparse_zgebsr2gebsr_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_mat_descr descr_A, const rocsparse_double_complex *bsr_val_A, const rocsparse_int *bsr_row_ptr_A, const rocsparse_int *bsr_col_ind_A, rocsparse_int row_block_dim_A, rocsparse_int col_block_dim_A, rocsparse_int row_block_dim_C, rocsparse_int col_block_dim_C, size_t *buffer_size)#

This function computes the the size of the user allocated temporary storage buffer used when converting a sparse general BSR matrix to another sparse general BSR matrix.

rocsparse_gebsr2gebsr_buffer_size returns the size of the temporary storage buffer that is required by rocsparse_gebsr2gebsr_nnz(), rocsparse_sgebsr2gebsr(), rocsparse_dgebsr2gebsr(), rocsparse_cgebsr2gebsr(), and rocsparse_zgebsr2gebsr(). The temporary storage buffer must be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] the storage format of the blocks, rocsparse_direction_row or rocsparse_direction_column

  • mb[in] number of block rows of the general BSR sparse matrix A.

  • nb[in] number of block columns of the general BSR sparse matrix A.

  • nnzb[in] number of blocks in the general BSR sparse matrix A.

  • descr_A[in] the descriptor of the general BSR sparse matrix A, the supported matrix type is rocsparse_matrix_type_general and also any valid value of the rocsparse_index_base.

  • bsr_val_A[in] array of nnzb*row_block_dim_A*col_block_dim_A containing the values of the sparse general BSR matrix A.

  • bsr_row_ptr_A[in] array of mb+1 elements that point to the start of every block row of the sparse general BSR matrix A.

  • bsr_col_ind_A[in] array of nnzb elements containing the block column indices of the sparse general BSR matrix A.

  • row_block_dim_A[in] row size of the blocks in the sparse general BSR matrix A.

  • col_block_dim_A[in] column size of the blocks in the sparse general BSR matrix A.

  • row_block_dim_C[in] row size of the blocks in the sparse general BSR matrix C.

  • col_block_dim_C[in] column size of the blocks in the sparse general BSR matrix C.

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_gebsr2gebsr_nnz(), rocsparse_sgebsr2gebsr(), rocsparse_dgebsr2gebsr(), rocsparse_cgebsr2gebsr(), and rocsparse_zgebsr2gebsr().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb or nb or nnzb or row_block_dim_A or col_block_dim_A or row_block_dim_C or col_block_dim_C is invalid.

  • rocsparse_status_invalid_pointerbsr_row_ptr_A or bsr_col_ind_A or descr_A or buffer_size pointer is invalid.

rocsparse_gebsr2gebsr_nnz()#

rocsparse_status rocsparse_gebsr2gebsr_nnz(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_mat_descr descr_A, const rocsparse_int *bsr_row_ptr_A, const rocsparse_int *bsr_col_ind_A, rocsparse_int row_block_dim_A, rocsparse_int col_block_dim_A, const rocsparse_mat_descr descr_C, rocsparse_int *bsr_row_ptr_C, rocsparse_int row_block_dim_C, rocsparse_int col_block_dim_C, rocsparse_int *nnz_total_dev_host_ptr, void *temp_buffer)#

This function is used when converting a general BSR sparse matrix A to another general BSR sparse matrix C. Specifically, this function determines the number of non-zero blocks that will exist in C (stored using either a host or device pointer), and computes the row pointer array for C.

The routine does support asynchronous execution.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] the storage format of the blocks, rocsparse_direction_row or rocsparse_direction_column

  • mb[in] number of block rows of the general BSR sparse matrix A.

  • nb[in] number of block columns of the general BSR sparse matrix A.

  • nnzb[in] number of blocks in the general BSR sparse matrix A.

  • descr_A[in] the descriptor of the general BSR sparse matrix A, the supported matrix type is rocsparse_matrix_type_general and also any valid value of the rocsparse_index_base.

  • bsr_row_ptr_A[in] array of mb+1 elements that point to the start of every block row of the sparse general BSR matrix A.

  • bsr_col_ind_A[in] array of nnzb elements containing the block column indices of the sparse general BSR matrix A.

  • row_block_dim_A[in] row size of the blocks in the sparse general BSR matrix A.

  • col_block_dim_A[in] column size of the blocks in the sparse general BSR matrix A.

  • descr_C[in] the descriptor of the general BSR sparse matrix C, the supported matrix type is rocsparse_matrix_type_general and also any valid value of the rocsparse_index_base.

  • bsr_row_ptr_C[in] array of mb_C+1 elements that point to the start of every block row of the sparse general BSR matrix C where mb_C=(m+row_block_dim_C-1)/row_block_dim_C.

  • row_block_dim_C[in] row size of the blocks in the sparse general BSR matrix C.

  • col_block_dim_C[in] column size of the blocks in the sparse general BSR matrix C.

  • nnz_total_dev_host_ptr[out] total number of nonzero blocks in general BSR sparse matrix C stored using device or host memory.

  • temp_buffer[out] buffer allocated by the user whose size is determined by calling rocsparse_xgebsr2gebsr_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb or nb or nnzb or row_block_dim_A or col_block_dim_A or row_block_dim_C or col_block_dim_C is invalid.

  • rocsparse_status_invalid_pointerbsr_row_ptr_A or bsr_col_ind_A or bsr_row_ptr_C or descr_A or descr_C or temp_buffer pointer is invalid.

rocsparse_gebsr2gebsr()#

rocsparse_status rocsparse_sgebsr2gebsr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_mat_descr descr_A, const float *bsr_val_A, const rocsparse_int *bsr_row_ptr_A, const rocsparse_int *bsr_col_ind_A, rocsparse_int row_block_dim_A, rocsparse_int col_block_dim_A, const rocsparse_mat_descr descr_C, float *bsr_val_C, rocsparse_int *bsr_row_ptr_C, rocsparse_int *bsr_col_ind_C, rocsparse_int row_block_dim_C, rocsparse_int col_block_dim_C, void *temp_buffer)#
rocsparse_status rocsparse_dgebsr2gebsr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_mat_descr descr_A, const double *bsr_val_A, const rocsparse_int *bsr_row_ptr_A, const rocsparse_int *bsr_col_ind_A, rocsparse_int row_block_dim_A, rocsparse_int col_block_dim_A, const rocsparse_mat_descr descr_C, double *bsr_val_C, rocsparse_int *bsr_row_ptr_C, rocsparse_int *bsr_col_ind_C, rocsparse_int row_block_dim_C, rocsparse_int col_block_dim_C, void *temp_buffer)#
rocsparse_status rocsparse_cgebsr2gebsr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_mat_descr descr_A, const rocsparse_float_complex *bsr_val_A, const rocsparse_int *bsr_row_ptr_A, const rocsparse_int *bsr_col_ind_A, rocsparse_int row_block_dim_A, rocsparse_int col_block_dim_A, const rocsparse_mat_descr descr_C, rocsparse_float_complex *bsr_val_C, rocsparse_int *bsr_row_ptr_C, rocsparse_int *bsr_col_ind_C, rocsparse_int row_block_dim_C, rocsparse_int col_block_dim_C, void *temp_buffer)#
rocsparse_status rocsparse_zgebsr2gebsr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, const rocsparse_mat_descr descr_A, const rocsparse_double_complex *bsr_val_A, const rocsparse_int *bsr_row_ptr_A, const rocsparse_int *bsr_col_ind_A, rocsparse_int row_block_dim_A, rocsparse_int col_block_dim_A, const rocsparse_mat_descr descr_C, rocsparse_double_complex *bsr_val_C, rocsparse_int *bsr_row_ptr_C, rocsparse_int *bsr_col_ind_C, rocsparse_int row_block_dim_C, rocsparse_int col_block_dim_C, void *temp_buffer)#

This function converts the general BSR sparse matrix A to another general BSR sparse matrix C.

The conversion uses three steps. First, the user calls rocsparse_xgebsr2gebsr_buffer_size() to determine the size of the required temporary storage buffer. The user then allocates this buffer. Secondly, the user then allocates mb_C+1 integers for the row pointer array for C where mb_C=(m+row_block_dim_C-1)/row_block_dim_C. The user then calls rocsparse_xgebsr2gebsr_nnz() to fill in the row pointer array for C ( bsr_row_ptr_C ) and determine the number of non-zero blocks that will exist in C. Finally, the user allocates space for the colimn indices array of C to have nnzb_C elements and space for the values array of C to have nnzb_C*roc_block_dim_C*col_block_dim_C and then calls rocsparse_xgebsr2gebsr() to complete the conversion.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] the storage format of the blocks, rocsparse_direction_row or rocsparse_direction_column

  • mb[in] number of block rows of the general BSR sparse matrix A.

  • nb[in] number of block columns of the general BSR sparse matrix A.

  • nnzb[in] number of blocks in the general BSR sparse matrix A.

  • descr_A[in] the descriptor of the general BSR sparse matrix A, the supported matrix type is rocsparse_matrix_type_general and also any valid value of the rocsparse_index_base.

  • bsr_val_A[in] array of nnzb*row_block_dim_A*col_block_dim_A containing the values of the sparse general BSR matrix A.

  • bsr_row_ptr_A[in] array of mb+1 elements that point to the start of every block row of the sparse general BSR matrix A.

  • bsr_col_ind_A[in] array of nnzb elements containing the block column indices of the sparse general BSR matrix A.

  • row_block_dim_A[in] row size of the blocks in the sparse general BSR matrix A.

  • col_block_dim_A[in] column size of the blocks in the sparse general BSR matrix A.

  • descr_C[in] the descriptor of the general BSR sparse matrix C, the supported matrix type is rocsparse_matrix_type_general and also any valid value of the rocsparse_index_base.

  • bsr_val_C[in] array of nnzb_C*row_block_dim_C*col_block_dim_C containing the values of the sparse general BSR matrix C.

  • bsr_row_ptr_C[in] array of mb_C+1 elements that point to the start of every block row of the sparse general BSR matrix C.

  • bsr_col_ind_C[in] array of nnzb_C elements containing the block column indices of the sparse general BSR matrix C.

  • row_block_dim_C[in] row size of the blocks in the sparse general BSR matrix C.

  • col_block_dim_C[in] column size of the blocks in the sparse general BSR matrix C.

  • temp_buffer[out] buffer allocated by the user whose size is determined by calling rocsparse_xgebsr2gebsr_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizemb or nb or nnzb or row_block_dim_A or col_block_dim_A or row_block_dim_C or col_block_dim_C is invalid.

  • rocsparse_status_invalid_pointerbsr_row_ptr_A or bsr_col_ind_A or bsr_val_A or bsr_row_ptr_C or bsr_col_ind_C or bsr_val_C or descr_A or descr_C or temp_buffer pointer is invalid.

rocsparse_csr2bsr_nnz()#

rocsparse_status rocsparse_csr2bsr_nnz(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr csr_descr, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_int block_dim, const rocsparse_mat_descr bsr_descr, rocsparse_int *bsr_row_ptr, rocsparse_int *bsr_nnz)#

This function computes the number of nonzero block columns per row and the total number of nonzero blocks in a sparse BSR matrix given a sparse CSR matrix as input.

The routine does support asynchronous execution if the pointer mode is set to device.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] direction that specified whether to count nonzero elements by rocsparse_direction_row or by rocsparse_direction_row.

  • m[in] number of rows of the sparse CSR matrix.

  • n[in] number of columns of the sparse CSR matrix.

  • csr_descr[in] descriptor of the sparse CSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • csr_row_ptr[in] integer array containing m+1 elements that point to the start of each row of the CSR matrix

  • csr_col_ind[in] integer array of the column indices for each non-zero element in the CSR matrix

  • block_dim[in] the block dimension of the BSR matrix. Between 1 and min(m, n)

  • bsr_descr[in] descriptor of the sparse BSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • bsr_row_ptr[out] integer array containing mb+1 elements that point to the start of each block row of the BSR matrix

  • bsr_nnz[out] total number of nonzero elements in device or host memory.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or block_dim is invalid.

  • rocsparse_status_invalid_pointercsr_row_ptr or csr_col_ind or bsr_row_ptr or bsr_nnz pointer is invalid.

rocsparse_csr2bsr()#

rocsparse_status rocsparse_scsr2bsr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr csr_descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_int block_dim, const rocsparse_mat_descr bsr_descr, float *bsr_val, rocsparse_int *bsr_row_ptr, rocsparse_int *bsr_col_ind)#
rocsparse_status rocsparse_dcsr2bsr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr csr_descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_int block_dim, const rocsparse_mat_descr bsr_descr, double *bsr_val, rocsparse_int *bsr_row_ptr, rocsparse_int *bsr_col_ind)#
rocsparse_status rocsparse_ccsr2bsr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr csr_descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_int block_dim, const rocsparse_mat_descr bsr_descr, rocsparse_float_complex *bsr_val, rocsparse_int *bsr_row_ptr, rocsparse_int *bsr_col_ind)#
rocsparse_status rocsparse_zcsr2bsr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr csr_descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_int block_dim, const rocsparse_mat_descr bsr_descr, rocsparse_double_complex *bsr_val, rocsparse_int *bsr_row_ptr, rocsparse_int *bsr_col_ind)#

Convert a sparse CSR matrix into a sparse BSR matrix.

rocsparse_csr2bsr converts a CSR matrix into a BSR matrix. It is assumed, that bsr_val, bsr_col_ind and bsr_row_ptr are allocated. Allocation size for bsr_row_ptr is computed as mb+1 where mb is the number of block rows in the BSR matrix. Allocation size for bsr_val and bsr_col_ind is computed using csr2bsr_nnz() which also fills in bsr_row_ptr.

rocsparse_csr2bsr requires extra temporary storage that is allocated internally if block_dim>16

Example

This example converts a CSR matrix into an BSR matrix.

//     1 4 0 0 0 0
// A = 0 2 3 0 0 0
//     5 0 0 7 8 0
//     0 0 9 0 6 0

rocsparse_int m   = 4;
rocsparse_int n   = 6;
rocsparse_int block_dim = 2;
rocsparse_int nnz = 9;
rocsparse_int mb = (m + block_dim - 1) / block_dim;
rocsparse_int nb = (n + block_dim - 1) / block_dim;

csr_row_ptr[m+1]  = {0, 2, 4, 7, 9};             // device memory
csr_col_ind[nnz]  = {0, 1, 1, 2, 0, 3, 4, 2, 4}; // device memory
csr_val[nnz]      = {1, 4, 2, 3, 5, 7, 8, 9, 6}; // device memory

hipMalloc(&bsr_row_ptr, sizeof(rocsparse_int) *(mb + 1));
rocsparse_int nnzb;
rocsparse_int* nnzTotalHostPtr = &nnzb;
csr2bsr_nnz(handle,
            rocsparse_direction_row,
            m,
            n,
            csr_descr,
            csr_row_ptr,
            csr_col_ind,
            block_dim,
            bsr_descr,
            bsr_row_ptr,
            nnzTotalHostPtr);
nnzb = *nnzTotalDevHostPtr;
hipMalloc(&bsr_col_ind, sizeof(int)*nnzb);
hipMalloc(&bsr_val, sizeof(float)*(block_dim * block_dim) * nnzb);
scsr2bsr(handle,
         rocsparse_direction_row,
         m,
         n,
         csr_descr,
         csr_val,
         csr_row_ptr,
         csr_col_ind,
         block_dim,
         bsr_descr,
         bsr_val,
         bsr_row_ptr,
         bsr_col_ind);

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] the storage format of the blocks, rocsparse_direction_row or rocsparse_direction_column

  • m[in] number of rows in the sparse CSR matrix.

  • n[in] number of columns in the sparse CSR matrix.

  • csr_descr[in] descriptor of the sparse CSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val[in] array of nnz elements containing the values of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • block_dim[in] size of the blocks in the sparse BSR matrix.

  • bsr_descr[in] descriptor of the sparse BSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • bsr_val[out] array of nnzb*block_dim*block_dim containing the values of the sparse BSR matrix.

  • bsr_row_ptr[out] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[out] array of nnzb elements containing the block column indices of the sparse BSR matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or block_dim is invalid.

  • rocsparse_status_invalid_pointerbsr_val, bsr_row_ptr, bsr_col_ind, csr_val, csr_row_ptr or csr_col_ind pointer is invalid.

rocsparse_csr2gebsr_nnz()#

rocsparse_status rocsparse_csr2gebsr_nnz(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr csr_descr, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_mat_descr bsr_descr, rocsparse_int *bsr_row_ptr, rocsparse_int row_block_dim, rocsparse_int col_block_dim, rocsparse_int *bsr_nnz_devhost, void *p_buffer)#

This function computes the number of nonzero block columns per row and the total number of nonzero blocks in a sparse GEneral BSR matrix given a sparse CSR matrix as input.

The routine does support asynchronous execution if the pointer mode is set to device.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] direction that specified whether to count nonzero elements by rocsparse_direction_row or by rocsparse_direction_row.

  • m[in] number of rows of the sparse CSR matrix.

  • n[in] number of columns of the sparse CSR matrix.

  • csr_descr[in] descriptor of the sparse CSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • csr_row_ptr[in] integer array containing m+1 elements that point to the start of each row of the CSR matrix

  • csr_col_ind[in] integer array of the column indices for each non-zero element in the CSR matrix

  • bsr_descr[in] descriptor of the sparse GEneral BSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • bsr_row_ptr[out] integer array containing mb+1 elements that point to the start of each block row of the General BSR matrix

  • row_block_dim[in] the row block dimension of the GEneral BSR matrix. Between 1 and min(m, n)

  • col_block_dim[in] the col block dimension of the GEneral BSR matrix. Between 1 and min(m, n)

  • bsr_nnz_devhost[out] total number of nonzero elements in device or host memory.

  • p_buffer[in] buffer allocated by the user whose size is determined by calling rocsparse_xcsr2gebsr_buffer_size.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or row_block_dim col_block_dim is invalid.

  • rocsparse_status_invalid_pointercsr_row_ptr or csr_col_ind or bsr_row_ptr or bsr_nnz_devhost pointer is invalid.

rocsparse_csr2gebsr_buffer_size()#

rocsparse_status rocsparse_scsr2gebsr_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr csr_descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, size_t *p_buffer_size)#
rocsparse_status rocsparse_dcsr2gebsr_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr csr_descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, size_t *p_buffer_size)#
rocsparse_status rocsparse_ccsr2gebsr_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr csr_descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, size_t *p_buffer_size)#
rocsparse_status rocsparse_zcsr2gebsr_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr csr_descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, size_t *p_buffer_size)#

rocsparse_csr2gebsr_buffer_size returns the size of the temporary buffer that is required by rocsparse_csr2gebcsr_nnz, rocsparse_scsr2gebcsr, rocsparse_dcsr2gebsr, rocsparse_ccsr2gebsr and rocsparse_zcsr2gebsr. The temporary storage buffer must be allocated by the user.

This function computes the number of nonzero block columns per row and the total number of nonzero blocks in a sparse GEneral BSR matrix given a sparse CSR matrix as input.

The routine does support asynchronous execution if the pointer mode is set to device.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] direction that specified whether to count nonzero elements by rocsparse_direction_row or by rocsparse_direction_row.

  • m[in] number of rows of the sparse CSR matrix.

  • n[in] number of columns of the sparse CSR matrix.

  • csr_descr[in] descriptor of the sparse CSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val[in] array of nnz elements containing the values of the sparse CSR matrix.

  • csr_row_ptr[in] integer array containing m+1 elements that point to the start of each row of the CSR matrix

  • csr_col_ind[in] integer array of the column indices for each non-zero element in the CSR matrix

  • row_block_dim[in] the row block dimension of the GEneral BSR matrix. Between 1 and m

  • col_block_dim[in] the col block dimension of the GEneral BSR matrix. Between 1 and n

  • p_buffer_size[out] (host/device) number of bytes of the temporary storage buffer required by rocsparse_csr2gebsr_nnz and rocsparse_scsr2gebsr.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or row_block_dim col_block_dim is invalid.

  • rocsparse_status_invalid_pointercsr_val or csr_row_ptr or csr_col_ind or bsr_row_ptr or p_buffer_size pointer is invalid.

rocsparse_csr2gebsr()#

rocsparse_status rocsparse_scsr2gebsr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr csr_descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_mat_descr bsr_descr, float *bsr_val, rocsparse_int *bsr_row_ptr, rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, void *p_buffer)#
rocsparse_status rocsparse_dcsr2gebsr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr csr_descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_mat_descr bsr_descr, double *bsr_val, rocsparse_int *bsr_row_ptr, rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, void *p_buffer)#
rocsparse_status rocsparse_ccsr2gebsr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr csr_descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_mat_descr bsr_descr, rocsparse_float_complex *bsr_val, rocsparse_int *bsr_row_ptr, rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, void *p_buffer)#
rocsparse_status rocsparse_zcsr2gebsr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr csr_descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const rocsparse_mat_descr bsr_descr, rocsparse_double_complex *bsr_val, rocsparse_int *bsr_row_ptr, rocsparse_int *bsr_col_ind, rocsparse_int row_block_dim, rocsparse_int col_block_dim, void *p_buffer)#

Convert a sparse CSR matrix into a sparse GEneral BSR matrix.

rocsparse_csr2gebsr converts a CSR matrix into a GEneral BSR matrix. It is assumed, that bsr_val, bsr_col_ind and bsr_row_ptr are allocated. Allocation size for bsr_row_ptr is computed as mb+1 where mb is the number of block rows in the GEneral BSR matrix. Allocation size for bsr_val and bsr_col_ind is computed using csr2gebsr_nnz() which also fills in bsr_row_ptr.

Example

This example converts a CSR matrix into an BSR matrix.

//     1 4 0 0 0 0
// A = 0 2 3 0 0 0
//     5 0 0 7 8 0
//     0 0 9 0 6 0

rocsparse_int m   = 4;
rocsparse_int n   = 6;
rocsparse_int row_block_dim = 2;
rocsparse_int col_block_dim = 3;
rocsparse_int nnz = 9;
rocsparse_int mb = (m + row_block_dim - 1) / row_block_dim;
rocsparse_int nb = (n + col_block_dim - 1) / col_block_dim;

csr_row_ptr[m+1]  = {0, 2, 4, 7, 9};             // device memory
csr_col_ind[nnz]  = {0, 1, 1, 2, 0, 3, 4, 2, 4}; // device memory
csr_val[nnz]      = {1, 4, 2, 3, 5, 7, 8, 9, 6}; // device memory

hipMalloc(&bsr_row_ptr, sizeof(rocsparse_int) *(mb + 1));
rocsparse_int nnzb;
rocsparse_int* nnzTotalHostPtr = &nnzb;
csr2gebsr_nnz(handle,
            rocsparse_direction_row,
            m,
            n,
            csr_descr,
            csr_row_ptr,
            csr_col_ind,
            row_block_dim,
            col_block_dim,
            bsr_descr,
            bsr_row_ptr,
            nnzTotalHostPtr);
nnzb = *nnzTotalDevHostPtr;
hipMalloc(&bsr_col_ind, sizeof(int)*nnzb);
hipMalloc(&bsr_val, sizeof(float)*(row_block_dim * col_block_dim) * nnzb);
scsr2gebsr(handle,
         rocsparse_direction_row,
         m,
         n,
         csr_descr,
         csr_val,
         csr_row_ptr,
         csr_col_ind,
         row_block_dim,
         col_block_dim,
         bsr_descr,
         bsr_val,
         bsr_row_ptr,
         bsr_col_ind);

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] the storage format of the blocks, rocsparse_direction_row or rocsparse_direction_column

  • m[in] number of rows in the sparse CSR matrix.

  • n[in] number of columns in the sparse CSR matrix.

  • csr_descr[in] descriptor of the sparse CSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val[in] array of nnz elements containing the values of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • bsr_descr[in] descriptor of the sparse BSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • bsr_val[out] array of nnzb* row_block_dim* containing the values of the sparse BSR matrix.

  • bsr_row_ptr[out] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[out] array of nnzb elements containing the block column indices of the sparse BSR matrix.

  • row_block_dim[in] row size of the blocks in the sparse GEneral BSR matrix.

  • col_block_dim[in] col size of the blocks in the sparse GEneral BSR matrix.

  • p_buffer[in] buffer allocated by the user whose size is determined by calling rocsparse_xcsr2gebsr_buffer_size.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or row_block_dim or col_block_dim is invalid.

  • rocsparse_status_invalid_pointerbsr_val, bsr_row_ptr, bsr_col_ind, csr_val, csr_row_ptr or csr_col_ind pointer is invalid.

rocsparse_csr2csr_compress()#

rocsparse_status rocsparse_scsr2csr_compress(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr_A, const float *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, rocsparse_int nnz_A, const rocsparse_int *nnz_per_row, float *csr_val_C, rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C, float tol)#
rocsparse_status rocsparse_dcsr2csr_compress(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr_A, const double *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, rocsparse_int nnz_A, const rocsparse_int *nnz_per_row, double *csr_val_C, rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C, double tol)#
rocsparse_status rocsparse_ccsr2csr_compress(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr_A, const rocsparse_float_complex *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, rocsparse_int nnz_A, const rocsparse_int *nnz_per_row, rocsparse_float_complex *csr_val_C, rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C, rocsparse_float_complex tol)#
rocsparse_status rocsparse_zcsr2csr_compress(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr_A, const rocsparse_double_complex *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, rocsparse_int nnz_A, const rocsparse_int *nnz_per_row, rocsparse_double_complex *csr_val_C, rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C, rocsparse_double_complex tol)#

Convert a sparse CSR matrix into a compressed sparse CSR matrix.

rocsparse_csr2csr_compress converts a CSR matrix into a compressed CSR matrix by removing entries in the input CSR matrix that are below a non-negative threshold tol

Example

This example demonstrates how to compress a CSR matrix. Compressing a CSR matrix involves two steps. First we use nnz_compress() to determine how many entries will be in the final compressed CSR matrix. Then we call csr2csr_compress() to finish the compression and fill in the column indices and values arrays of the compressed CSR matrix.

//     1 2 0 3 0
// A = 0 4 5 0 0
//     6 0 0 7 8

float tol = 0.0f;

rocsparse_int m     = 3;
rocsparse_int n     = 5;
rocsparse_int nnz_A = 8;

csr_row_ptr_A[m+1]   = {0, 3, 5, 8};             // device memory
csr_col_ind_A[nnz_A] = {0, 1, 3, 1, 2, 0, 3, 4}; // device memory
csr_val_A[nnz_A]     = {1, 0, 3, 4, 0, 6, 7, 0}; // device memory

// Allocate memory for the row pointer array of the compressed CSR matrix
rocsparse_int* csr_row_ptr_C;
hipMalloc(csr_row_ptr_C, sizeof(rocsparse_int) * (m + 1));

// Allocate memory for the nnz_per_row array
rocsparse_int* nnz_per_row;
hipMalloc(nnz_per_row, sizeof(rocsparse_int) * m);

// Call nnz_compress() which fills in nnz_per_row array and finds the number
// of entries that will be in the compressed CSR matrix
rocsparse_int nnz_C;
nnz_compress(handle,
             m,
             descr_A,
             csr_val_A,
             csr_row_ptr_A,
             nnz_per_row,
             &nnz_C,
             tol);

// Allocate column indices and values array for the compressed CSR matrix
rocsparse_int* csr_col_ind_C;
rocsparse_int* csr_val_C;
hipMalloc(csr_col_ind_C, sizeof(rocsparse_int) * nnz_C;
hipMalloc(csr_val_C, sizeof(rocsparse_int) * nnz_C;

// Finish compression by calling csr2csr_compress()
csr2csr_compress(handle,
                 m,
                 n,
                 descr_A,
                 csr_val_A,
                 csr_row_ptr_A,
                 csr_col_ind_A,
                 nnz_A,
                 nnz_per_row,
                 csr_val_C,
                 csr_row_ptr_C,
                 csr_col_ind_C,
                 tol);

Note

In the case of complex matrices only the magnitude of the real part of tol is used.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSR matrix.

  • n[in] number of columns of the sparse CSR matrix.

  • descr_A[in] matrix descriptor for the CSR matrix

  • csr_val_A[in] array of nnz_A elements of the sparse CSR matrix.

  • csr_row_ptr_A[in] array of m+1 elements that point to the start of every row of the uncompressed sparse CSR matrix.

  • csr_col_ind_A[in] array of nnz_A elements containing the column indices of the uncompressed sparse CSR matrix.

  • nnz_A[in] number of elements in the column indices and values arrays of the uncompressed sparse CSR matrix.

  • nnz_per_row[in] array of length m containing the number of entries that will be kept per row in the final compressed CSR matrix.

  • csr_val_C[out] array of nnz_C elements of the compressed sparse CSC matrix.

  • csr_row_ptr_C[out] array of m+1 elements that point to the start of every column of the compressed sparse CSR matrix.

  • csr_col_ind_C[out] array of nnz_C elements containing the row indices of the compressed sparse CSR matrix.

  • tol[in] the non-negative tolerance used for compression. If tol is complex then only the magnitude of the real part is used. Entries in the input uncompressed CSR array that are below the tolerance are removed in output compressed CSR matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or nnz_A is invalid.

  • rocsparse_status_invalid_valuetol is invalid.

  • rocsparse_status_invalid_pointercsr_val_A, csr_row_ptr_A, csr_col_ind_A, csr_val_C, csr_row_ptr_C, csr_col_ind_C or nnz_per_row pointer is invalid.

rocsparse_create_identity_permutation()#

rocsparse_status rocsparse_create_identity_permutation(rocsparse_handle handle, rocsparse_int n, rocsparse_int *p)#

Create the identity map.

rocsparse_create_identity_permutation stores the identity map in p, such that \(p = 0:1:(n-1)\).

for(i = 0; i < n; ++i)
{
    p[i] = i;
}

Example

The following example creates an identity permutation.

rocsparse_int size = 200;

// Allocate memory to hold the identity map
rocsparse_int* perm;
hipMalloc((void**)&perm, sizeof(rocsparse_int) * size);

// Fill perm with the identity permutation
rocsparse_create_identity_permutation(handle, size, perm);

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • n[in] size of the map p.

  • p[out] array of n integers containing the map.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizen is invalid.

  • rocsparse_status_invalid_pointerp pointer is invalid.

rocsparse_csrsort_buffer_size()#

rocsparse_status rocsparse_csrsort_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, size_t *buffer_size)#

Sort a sparse CSR matrix.

rocsparse_csrsort_buffer_size returns the size of the temporary storage buffer required by rocsparse_csrsort(). The temporary storage buffer must be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSR matrix.

  • n[in] number of columns of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_csrsort().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or nnz is invalid.

  • rocsparse_status_invalid_pointercsr_row_ptr, csr_col_ind or buffer_size pointer is invalid.

rocsparse_csrsort()#

rocsparse_status rocsparse_csrsort(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind, rocsparse_int *perm, void *temp_buffer)#

Sort a sparse CSR matrix.

rocsparse_csrsort sorts a matrix in CSR format. The sorted permutation vector perm can be used to obtain sorted csr_val array. In this case, perm must be initialized as the identity permutation, see rocsparse_create_identity_permutation().

rocsparse_csrsort requires extra temporary storage buffer that has to be allocated by the user. Storage buffer size can be determined by rocsparse_csrsort_buffer_size().

Example

The following example sorts a \(3 \times 3\) CSR matrix.

//     1 2 3
// A = 4 5 6
//     7 8 9
rocsparse_int m   = 3;
rocsparse_int n   = 3;
rocsparse_int nnz = 9;

csr_row_ptr[m + 1] = {0, 3, 6, 9};                // device memory
csr_col_ind[nnz]   = {2, 0, 1, 0, 1, 2, 0, 2, 1}; // device memory
csr_val[nnz]       = {3, 1, 2, 4, 5, 6, 7, 9, 8}; // device memory

// Create permutation vector perm as the identity map
rocsparse_int* perm;
hipMalloc((void**)&perm, sizeof(rocsparse_int) * nnz);
rocsparse_create_identity_permutation(handle, nnz, perm);

// Allocate temporary buffer
size_t buffer_size;
void* temp_buffer;
rocsparse_csrsort_buffer_size(handle, m, n, nnz, csr_row_ptr, csr_col_ind, &buffer_size);
hipMalloc(&temp_buffer, buffer_size);

// Sort the CSR matrix
rocsparse_csrsort(handle, m, n, nnz, descr, csr_row_ptr, csr_col_ind, perm, temp_buffer);

// Gather sorted csr_val array
float* csr_val_sorted;
hipMalloc((void**)&csr_val_sorted, sizeof(float) * nnz);
rocsparse_sgthr(handle, nnz, csr_val, csr_val_sorted, perm, rocsparse_index_base_zero);

// Clean up
hipFree(temp_buffer);
hipFree(perm);
hipFree(csr_val);

Note

perm can be NULL if a sorted permutation vector is not required.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSR matrix.

  • n[in] number of columns of the sparse CSR matrix.

  • nnz[in] number of non-zero entries of the sparse CSR matrix.

  • descr[in] descriptor of the sparse CSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[inout] array of nnz elements containing the column indices of the sparse CSR matrix.

  • perm[inout] array of nnz integers containing the unsorted map indices, can be NULL.

  • temp_buffer[in] temporary storage buffer allocated by the user, size is returned by rocsparse_csrsort_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or nnz is invalid.

  • rocsparse_status_invalid_pointerdescr, csr_row_ptr, csr_col_ind or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_cscsort_buffer_size()#

rocsparse_status rocsparse_cscsort_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_int *csc_col_ptr, const rocsparse_int *csc_row_ind, size_t *buffer_size)#

Sort a sparse CSC matrix.

rocsparse_cscsort_buffer_size returns the size of the temporary storage buffer required by rocsparse_cscsort(). The temporary storage buffer must be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSC matrix.

  • n[in] number of columns of the sparse CSC matrix.

  • nnz[in] number of non-zero entries of the sparse CSC matrix.

  • csc_col_ptr[in] array of n+1 elements that point to the start of every column of the sparse CSC matrix.

  • csc_row_ind[in] array of nnz elements containing the row indices of the sparse CSC matrix.

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_cscsort().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or nnz is invalid.

  • rocsparse_status_invalid_pointercsc_col_ptr, csc_row_ind or buffer_size pointer is invalid.

rocsparse_cscsort()#

rocsparse_status rocsparse_cscsort(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_int *csc_col_ptr, rocsparse_int *csc_row_ind, rocsparse_int *perm, void *temp_buffer)#

Sort a sparse CSC matrix.

rocsparse_cscsort sorts a matrix in CSC format. The sorted permutation vector perm can be used to obtain sorted csc_val array. In this case, perm must be initialized as the identity permutation, see rocsparse_create_identity_permutation().

rocsparse_cscsort requires extra temporary storage buffer that has to be allocated by the user. Storage buffer size can be determined by rocsparse_cscsort_buffer_size().

Example

The following example sorts a \(3 \times 3\) CSC matrix.

//     1 2 3
// A = 4 5 6
//     7 8 9
rocsparse_int m   = 3;
rocsparse_int n   = 3;
rocsparse_int nnz = 9;

csc_col_ptr[m + 1] = {0, 3, 6, 9};                // device memory
csc_row_ind[nnz]   = {2, 0, 1, 0, 1, 2, 0, 2, 1}; // device memory
csc_val[nnz]       = {7, 1, 4, 2, 5, 8, 3, 9, 6}; // device memory

// Create permutation vector perm as the identity map
rocsparse_int* perm;
hipMalloc((void**)&perm, sizeof(rocsparse_int) * nnz);
rocsparse_create_identity_permutation(handle, nnz, perm);

// Allocate temporary buffer
size_t buffer_size;
void* temp_buffer;
rocsparse_cscsort_buffer_size(handle, m, n, nnz, csc_col_ptr, csc_row_ind, &buffer_size);
hipMalloc(&temp_buffer, buffer_size);

// Sort the CSC matrix
rocsparse_cscsort(handle, m, n, nnz, descr, csc_col_ptr, csc_row_ind, perm, temp_buffer);

// Gather sorted csc_val array
float* csc_val_sorted;
hipMalloc((void**)&csc_val_sorted, sizeof(float) * nnz);
rocsparse_sgthr(handle, nnz, csc_val, csc_val_sorted, perm, rocsparse_index_base_zero);

// Clean up
hipFree(temp_buffer);
hipFree(perm);
hipFree(csc_val);

Note

perm can be NULL if a sorted permutation vector is not required.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSC matrix.

  • n[in] number of columns of the sparse CSC matrix.

  • nnz[in] number of non-zero entries of the sparse CSC matrix.

  • descr[in] descriptor of the sparse CSC matrix. Currently, only rocsparse_matrix_type_general is supported.

  • csc_col_ptr[in] array of n+1 elements that point to the start of every column of the sparse CSC matrix.

  • csc_row_ind[inout] array of nnz elements containing the row indices of the sparse CSC matrix.

  • perm[inout] array of nnz integers containing the unsorted map indices, can be NULL.

  • temp_buffer[in] temporary storage buffer allocated by the user, size is returned by rocsparse_cscsort_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or nnz is invalid.

  • rocsparse_status_invalid_pointerdescr, csc_col_ptr, csc_row_ind or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

  • rocsparse_status_not_implementedrocsparse_matrix_type != rocsparse_matrix_type_general.

rocsparse_coosort_buffer_size()#

rocsparse_status rocsparse_coosort_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, size_t *buffer_size)#

Sort a sparse COO matrix.

coosort_buffer_size returns the size of the temporary storage buffer that is required by rocsparse_coosort_by_row() and rocsparse_coosort_by_column(). The temporary storage buffer has to be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse COO matrix.

  • n[in] number of columns of the sparse COO matrix.

  • nnz[in] number of non-zero entries of the sparse COO matrix.

  • coo_row_ind[in] array of nnz elements containing the row indices of the sparse COO matrix.

  • coo_col_ind[in] array of nnz elements containing the column indices of the sparse COO matrix.

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_coosort_by_row() and rocsparse_coosort_by_column().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or nnz is invalid.

  • rocsparse_status_invalid_pointercoo_row_ind, coo_col_ind or buffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_coosort_by_row()#

rocsparse_status rocsparse_coosort_by_row(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, rocsparse_int *coo_row_ind, rocsparse_int *coo_col_ind, rocsparse_int *perm, void *temp_buffer)#

Sort a sparse COO matrix by row.

rocsparse_coosort_by_row sorts a matrix in COO format by row. The sorted permutation vector perm can be used to obtain sorted coo_val array. In this case, perm must be initialized as the identity permutation, see rocsparse_create_identity_permutation().

rocsparse_coosort_by_row requires extra temporary storage buffer that has to be allocated by the user. Storage buffer size can be determined by rocsparse_coosort_buffer_size().

Example

The following example sorts a \(3 \times 3\) COO matrix by row indices.

//     1 2 3
// A = 4 5 6
//     7 8 9
rocsparse_int m   = 3;
rocsparse_int n   = 3;
rocsparse_int nnz = 9;

coo_row_ind[nnz] = {0, 1, 2, 0, 1, 2, 0, 1, 2}; // device memory
coo_col_ind[nnz] = {0, 0, 0, 1, 1, 1, 2, 2, 2}; // device memory
coo_val[nnz]     = {1, 4, 7, 2, 5, 8, 3, 6, 9}; // device memory

// Create permutation vector perm as the identity map
rocsparse_int* perm;
hipMalloc((void**)&perm, sizeof(rocsparse_int) * nnz);
rocsparse_create_identity_permutation(handle, nnz, perm);

// Allocate temporary buffer
size_t buffer_size;
void* temp_buffer;
rocsparse_coosort_buffer_size(handle,
                              m,
                              n,
                              nnz,
                              coo_row_ind,
                              coo_col_ind,
                              &buffer_size);
hipMalloc(&temp_buffer, buffer_size);

// Sort the COO matrix
rocsparse_coosort_by_row(handle,
                         m,
                         n,
                         nnz,
                         coo_row_ind,
                         coo_col_ind,
                         perm,
                         temp_buffer);

// Gather sorted coo_val array
float* coo_val_sorted;
hipMalloc((void**)&coo_val_sorted, sizeof(float) * nnz);
rocsparse_sgthr(handle, nnz, coo_val, coo_val_sorted, perm, rocsparse_index_base_zero);

// Clean up
hipFree(temp_buffer);
hipFree(perm);
hipFree(coo_val);

Note

perm can be NULL if a sorted permutation vector is not required.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse COO matrix.

  • n[in] number of columns of the sparse COO matrix.

  • nnz[in] number of non-zero entries of the sparse COO matrix.

  • coo_row_ind[inout] array of nnz elements containing the row indices of the sparse COO matrix.

  • coo_col_ind[inout] array of nnz elements containing the column indices of the sparse COO matrix.

  • perm[inout] array of nnz integers containing the unsorted map indices, can be NULL.

  • temp_buffer[in] temporary storage buffer allocated by the user, size is returned by rocsparse_coosort_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or nnz is invalid.

  • rocsparse_status_invalid_pointercoo_row_ind, coo_col_ind or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_coosort_by_column()#

rocsparse_status rocsparse_coosort_by_column(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, rocsparse_int *coo_row_ind, rocsparse_int *coo_col_ind, rocsparse_int *perm, void *temp_buffer)#

Sort a sparse COO matrix by column.

rocsparse_coosort_by_column sorts a matrix in COO format by column. The sorted permutation vector perm can be used to obtain sorted coo_val array. In this case, perm must be initialized as the identity permutation, see rocsparse_create_identity_permutation().

rocsparse_coosort_by_column requires extra temporary storage buffer that has to be allocated by the user. Storage buffer size can be determined by rocsparse_coosort_buffer_size().

Example

The following example sorts a \(3 \times 3\) COO matrix by column indices.

//     1 2 3
// A = 4 5 6
//     7 8 9
rocsparse_int m   = 3;
rocsparse_int n   = 3;
rocsparse_int nnz = 9;

coo_row_ind[nnz] = {0, 0, 0, 1, 1, 1, 2, 2, 2}; // device memory
coo_col_ind[nnz] = {0, 1, 2, 0, 1, 2, 0, 1, 2}; // device memory
coo_val[nnz]     = {1, 2, 3, 4, 5, 6, 7, 8, 9}; // device memory

// Create permutation vector perm as the identity map
rocsparse_int* perm;
hipMalloc((void**)&perm, sizeof(rocsparse_int) * nnz);
rocsparse_create_identity_permutation(handle, nnz, perm);

// Allocate temporary buffer
size_t buffer_size;
void* temp_buffer;
rocsparse_coosort_buffer_size(handle,
                              m,
                              n,
                              nnz,
                              coo_row_ind,
                              coo_col_ind,
                              &buffer_size);
hipMalloc(&temp_buffer, buffer_size);

// Sort the COO matrix
rocsparse_coosort_by_column(handle,
                            m,
                            n,
                            nnz,
                            coo_row_ind,
                            coo_col_ind,
                            perm,
                            temp_buffer);

// Gather sorted coo_val array
float* coo_val_sorted;
hipMalloc((void**)&coo_val_sorted, sizeof(float) * nnz);
rocsparse_sgthr(handle, nnz, coo_val, coo_val_sorted, perm, rocsparse_index_base_zero);

// Clean up
hipFree(temp_buffer);
hipFree(perm);
hipFree(coo_val);

Note

perm can be NULL if a sorted permutation vector is not required.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse COO matrix.

  • n[in] number of columns of the sparse COO matrix.

  • nnz[in] number of non-zero entries of the sparse COO matrix.

  • coo_row_ind[inout] array of nnz elements containing the row indices of the sparse COO matrix.

  • coo_col_ind[inout] array of nnz elements containing the column indices of the sparse COO matrix.

  • perm[inout] array of nnz integers containing the unsorted map indices, can be NULL.

  • temp_buffer[in] temporary storage buffer allocated by the user, size is returned by rocsparse_coosort_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, n or nnz is invalid.

  • rocsparse_status_invalid_pointercoo_row_ind, coo_col_ind or temp_buffer pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_nnz_compress()#

rocsparse_status rocsparse_snnz_compress(rocsparse_handle handle, rocsparse_int m, const rocsparse_mat_descr descr_A, const float *csr_val_A, const rocsparse_int *csr_row_ptr_A, rocsparse_int *nnz_per_row, rocsparse_int *nnz_C, float tol)#
rocsparse_status rocsparse_dnnz_compress(rocsparse_handle handle, rocsparse_int m, const rocsparse_mat_descr descr_A, const double *csr_val_A, const rocsparse_int *csr_row_ptr_A, rocsparse_int *nnz_per_row, rocsparse_int *nnz_C, double tol)#
rocsparse_status rocsparse_cnnz_compress(rocsparse_handle handle, rocsparse_int m, const rocsparse_mat_descr descr_A, const rocsparse_float_complex *csr_val_A, const rocsparse_int *csr_row_ptr_A, rocsparse_int *nnz_per_row, rocsparse_int *nnz_C, rocsparse_float_complex tol)#
rocsparse_status rocsparse_znnz_compress(rocsparse_handle handle, rocsparse_int m, const rocsparse_mat_descr descr_A, const rocsparse_double_complex *csr_val_A, const rocsparse_int *csr_row_ptr_A, rocsparse_int *nnz_per_row, rocsparse_int *nnz_C, rocsparse_double_complex tol)#

Given a sparse CSR matrix and a non-negative tolerance, this function computes how many entries would be left in each row of the matrix if elements less than the tolerance were removed. It also computes the total number of remaining elements in the matrix.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse CSR matrix.

  • descr_A[in] the descriptor of the sparse CSR matrix.

  • csr_val_A[in] array of nnz_A elements of the sparse CSR matrix.

  • csr_row_ptr_A[in] array of m+1 elements that point to the start of every row of the uncompressed sparse CSR matrix.

  • nnz_per_row[out] array of length m containing the number of entries that will be kept per row in the final compressed CSR matrix.

  • nnz_C[out] number of elements in the column indices and values arrays of the compressed sparse CSR matrix. Can be either host or device pointer.

  • tol[in] the non-negative tolerance used for compression. If tol is complex then only the magnitude of the real part is used. Entries in the input uncompressed CSR array that are below the tolerance are removed in output compressed CSR matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n is invalid.

  • rocsparse_status_invalid_valuetol is invalid.

  • rocsparse_status_invalid_pointercsr_val_A or csr_row_ptr_A or nnz_per_row or nnz_C pointer is invalid.

rocsparse_nnz()#

rocsparse_status rocsparse_snnz(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const float *A, rocsparse_int ld, rocsparse_int *nnz_per_row_columns, rocsparse_int *nnz_total_dev_host_ptr)#
rocsparse_status rocsparse_dnnz(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const double *A, rocsparse_int ld, rocsparse_int *nnz_per_row_columns, rocsparse_int *nnz_total_dev_host_ptr)#
rocsparse_status rocsparse_cnnz(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const rocsparse_float_complex *A, rocsparse_int ld, rocsparse_int *nnz_per_row_columns, rocsparse_int *nnz_total_dev_host_ptr)#
rocsparse_status rocsparse_znnz(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const rocsparse_double_complex *A, rocsparse_int ld, rocsparse_int *nnz_per_row_columns, rocsparse_int *nnz_total_dev_host_ptr)#

This function computes the number of nonzero elements per row or column and the total number of nonzero elements in a dense matrix.

The routine does support asynchronous execution if the pointer mode is set to device.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • dir[in] direction that specified whether to count nonzero elements by rocsparse_direction_row or by rocsparse_direction_row.

  • m[in] number of rows of the dense matrix A.

  • n[in] number of columns of the dense matrix A.

  • descr[in] the descriptor of the dense matrix A.

  • A[in] array of dimensions (ld, n)

  • ld[in] leading dimension of dense array A.

  • nnz_per_row_columns[out] array of size m or n containing the number of nonzero elements per row or column, respectively.

  • nnz_total_dev_host_ptr[out] total number of nonzero elements in device or host memory.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or ld is invalid.

  • rocsparse_status_invalid_pointerA or nnz_per_row_columns or nnz_total_dev_host_ptr pointer is invalid.

rocsparse_dense2csr()#

rocsparse_status rocsparse_sdense2csr(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const float *A, rocsparse_int ld, const rocsparse_int *nnz_per_rows, float *csr_val, rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind)#
rocsparse_status rocsparse_ddense2csr(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const double *A, rocsparse_int ld, const rocsparse_int *nnz_per_rows, double *csr_val, rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind)#
rocsparse_status rocsparse_cdense2csr(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const rocsparse_float_complex *A, rocsparse_int ld, const rocsparse_int *nnz_per_rows, rocsparse_float_complex *csr_val, rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind)#
rocsparse_status rocsparse_zdense2csr(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const rocsparse_double_complex *A, rocsparse_int ld, const rocsparse_int *nnz_per_rows, rocsparse_double_complex *csr_val, rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind)#

This function converts the matrix A in dense format into a sparse matrix in CSR format. All the parameters are assumed to have been pre-allocated by the user and the arrays are filled in based on nnz_per_row, which can be pre-computed with rocsparse_xnnz(). It is executed asynchronously with respect to the host and may return control to the application on the host before the entire result is ready.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the dense matrix A.

  • n[in] number of columns of the dense matrix A.

  • descr[in] the descriptor of the dense matrix A, the supported matrix type is rocsparse_matrix_type_general and also any valid value of the rocsparse_index_base.

  • A[in] array of dimensions (ld, n)

  • ld[in] leading dimension of dense array A.

  • nnz_per_rows[in] array of size n containing the number of non-zero elements per row.

  • csr_val[out] array of nnz ( = csr_row_ptr[m] - csr_row_ptr[0] ) nonzero elements of matrix A.

  • csr_row_ptr[out] integer array of m+1 elements that contains the start of every row and the end of the last row plus one.

  • csr_col_ind[out] integer array of nnz ( = csr_row_ptr[m] - csr_row_ptr[0] ) column indices of the non-zero elements of matrix A.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or ld is invalid.

  • rocsparse_status_invalid_pointerA or nnz_per_rows or csr_val csr_row_ptr or csr_col_ind pointer is invalid.

rocsparse_dense2csc()#

rocsparse_status rocsparse_sdense2csc(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const float *A, rocsparse_int ld, const rocsparse_int *nnz_per_columns, float *csc_val, rocsparse_int *csc_col_ptr, rocsparse_int *csc_row_ind)#
rocsparse_status rocsparse_ddense2csc(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const double *A, rocsparse_int ld, const rocsparse_int *nnz_per_columns, double *csc_val, rocsparse_int *csc_col_ptr, rocsparse_int *csc_row_ind)#
rocsparse_status rocsparse_cdense2csc(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const rocsparse_float_complex *A, rocsparse_int ld, const rocsparse_int *nnz_per_columns, rocsparse_float_complex *csc_val, rocsparse_int *csc_col_ptr, rocsparse_int *csc_row_ind)#
rocsparse_status rocsparse_zdense2csc(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const rocsparse_double_complex *A, rocsparse_int ld, const rocsparse_int *nnz_per_columns, rocsparse_double_complex *csc_val, rocsparse_int *csc_col_ptr, rocsparse_int *csc_row_ind)#

This function converts the matrix A in dense format into a sparse matrix in CSC format. All the parameters are assumed to have been pre-allocated by the user and the arrays are filled in based on nnz_per_columns, which can be pre-computed with rocsparse_xnnz(). It is executed asynchronously with respect to the host and may return control to the application on the host before the entire result is ready.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the dense matrix A.

  • n[in] number of columns of the dense matrix A.

  • descr[in] the descriptor of the dense matrix A, the supported matrix type is rocsparse_matrix_type_general and also any valid value of the rocsparse_index_base.

  • A[in] array of dimensions (ld, n)

  • ld[in] leading dimension of dense array A.

  • nnz_per_columns[in] array of size n containing the number of non-zero elements per column.

  • csc_val[out] array of nnz ( = csc_col_ptr[m] - csc_col_ptr[0] ) nonzero elements of matrix A.

  • csc_col_ptr[out] integer array of m+1 elements that contains the start of every column and the end of the last column plus one.

  • csc_row_ind[out] integer array of nnz ( = csc_col_ptr[m] - csc_col_ptr[0] ) column indices of the non-zero elements of matrix A.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or ld is invalid.

  • rocsparse_status_invalid_pointerA or nnz_per_columns or csc_val csc_col_ptr or csc_row_ind pointer is invalid.

rocsparse_dense2coo()#

rocsparse_status rocsparse_sdense2coo(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const float *A, rocsparse_int ld, const rocsparse_int *nnz_per_rows, float *coo_val, rocsparse_int *coo_row_ind, rocsparse_int *coo_col_ind)#
rocsparse_status rocsparse_ddense2coo(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const double *A, rocsparse_int ld, const rocsparse_int *nnz_per_rows, double *coo_val, rocsparse_int *coo_row_ind, rocsparse_int *coo_col_ind)#
rocsparse_status rocsparse_cdense2coo(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const rocsparse_float_complex *A, rocsparse_int ld, const rocsparse_int *nnz_per_rows, rocsparse_float_complex *coo_val, rocsparse_int *coo_row_ind, rocsparse_int *coo_col_ind)#
rocsparse_status rocsparse_zdense2coo(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const rocsparse_double_complex *A, rocsparse_int ld, const rocsparse_int *nnz_per_rows, rocsparse_double_complex *coo_val, rocsparse_int *coo_row_ind, rocsparse_int *coo_col_ind)#

This function converts the matrix A in dense format into a sparse matrix in COO format. All the parameters are assumed to have been pre-allocated by the user and the arrays are filled in based on nnz_per_rows, which can be pre-computed with rocsparse_xnnz().

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the dense matrix A.

  • n[in] number of columns of the dense matrix A.

  • descr[in] the descriptor of the dense matrix A, the supported matrix type is rocsparse_matrix_type_general and also any valid value of the rocsparse_index_base.

  • A[in] array of dimensions (ld, n)

  • ld[in] leading dimension of dense array A.

  • nnz_per_rows[in] array of size n containing the number of non-zero elements per row.

  • coo_val[out] array of nnz nonzero elements of matrix A.

  • coo_row_ind[out] integer array of nnz row indices of the non-zero elements of matrix A.

  • coo_col_ind[out] integer array of nnz column indices of the non-zero elements of matrix A.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or ld is invalid.

  • rocsparse_status_invalid_pointerA or nnz_per_rows or coo_val coo_col_ind or coo_row_ind pointer is invalid.

rocsparse_csr2dense()#

rocsparse_status rocsparse_scsr2dense(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, float *A, rocsparse_int ld)#
rocsparse_status rocsparse_dcsr2dense(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, double *A, rocsparse_int ld)#
rocsparse_status rocsparse_ccsr2dense(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_float_complex *A, rocsparse_int ld)#
rocsparse_status rocsparse_zcsr2dense(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_double_complex *A, rocsparse_int ld)#

This function converts the sparse matrix in CSR format into a dense matrix. It is executed asynchronously with respect to the host and may return control to the application on the host before the entire result is ready.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the dense matrix A.

  • n[in] number of columns of the dense matrix A.

  • descr[in] the descriptor of the dense matrix A, the supported matrix type is rocsparse_matrix_type_general and also any valid value of the rocsparse_index_base.

  • csr_val[in] array of nnz ( = csr_row_ptr[m] - csr_row_ptr[0] ) nonzero elements of matrix A.

  • csr_row_ptr[in] integer array of m+1 elements that contains the start of every row and the end of the last row plus one.

  • csr_col_ind[in] integer array of nnz ( = csr_row_ptr[m] - csr_row_ptr[0] ) column indices of the non-zero elements of matrix A.

  • A[out] array of dimensions (ld, n)

  • ld[out] leading dimension of dense array A.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or ld is invalid.

  • rocsparse_status_invalid_pointerA or csr_val csr_row_ptr or csr_col_ind pointer is invalid.

rocsparse_csc2dense()#

rocsparse_status rocsparse_scsc2dense(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const float *csc_val, const rocsparse_int *csc_col_ptr, const rocsparse_int *csc_row_ind, float *A, rocsparse_int ld)#
rocsparse_status rocsparse_dcsc2dense(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const double *csc_val, const rocsparse_int *csc_col_ptr, const rocsparse_int *csc_row_ind, double *A, rocsparse_int ld)#
rocsparse_status rocsparse_ccsc2dense(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const rocsparse_float_complex *csc_val, const rocsparse_int *csc_col_ptr, const rocsparse_int *csc_row_ind, rocsparse_float_complex *A, rocsparse_int ld)#
rocsparse_status rocsparse_zcsc2dense(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const rocsparse_mat_descr descr, const rocsparse_double_complex *csc_val, const rocsparse_int *csc_col_ptr, const rocsparse_int *csc_row_ind, rocsparse_double_complex *A, rocsparse_int ld)#

This function converts the sparse matrix in CSC format into a dense matrix. It is executed asynchronously with respect to the host and may return control to the application on the host before the entire result is ready.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the dense matrix A.

  • n[in] number of columns of the dense matrix A.

  • descr[in] the descriptor of the dense matrix A, the supported matrix type is rocsparse_matrix_type_general and also any valid value of the rocsparse_index_base.

  • csc_val[in] array of nnz ( = csc_col_ptr[m] - csc_col_ptr[0] ) nonzero elements of matrix A.

  • csc_col_ptr[in] integer array of m+1 elements that contains the start of every row and the end of the last row plus one.

  • csc_row_ind[in] integer array of nnz ( = csc_col_ptr[m] - csc_col_ptr[0] ) column indices of the non-zero elements of matrix A.

  • A[out] array of dimensions (ld, n)

  • ld[out] leading dimension of dense array A.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or ld is invalid.

  • rocsparse_status_invalid_pointerA or csc_val csc_col_ptr or csc_row_ind pointer is invalid.

rocsparse_coo2dense()#

rocsparse_status rocsparse_scoo2dense(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_mat_descr descr, const float *coo_val, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, float *A, rocsparse_int ld)#
rocsparse_status rocsparse_dcoo2dense(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_mat_descr descr, const double *coo_val, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, double *A, rocsparse_int ld)#
rocsparse_status rocsparse_ccoo2dense(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_float_complex *coo_val, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, rocsparse_float_complex *A, rocsparse_int ld)#
rocsparse_status rocsparse_zcoo2dense(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_double_complex *coo_val, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, rocsparse_double_complex *A, rocsparse_int ld)#

This function converts the sparse matrix in COO format into a dense matrix. It is executed asynchronously with respect to the host and may return control to the application on the host before the entire result is ready.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the dense matrix A.

  • n[in] number of columns of the dense matrix A.

  • nnz[in] number of non-zero entries of the sparse COO matrix.

  • descr[in] the descriptor of the dense matrix A, the supported matrix type is rocsparse_matrix_type_general and also any valid value of the rocsparse_index_base.

  • coo_val[in] array of nnz nonzero elements of matrix A.

  • coo_row_ind[in] integer array of nnz row indices of the non-zero elements of matrix A.

  • coo_col_ind[in] integer array of nnz column indices of the non-zero elements of matrix A.

  • A[out] array of dimensions (ld, n)

  • ld[out] leading dimension of dense array A.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or nnz or ld is invalid.

  • rocsparse_status_invalid_pointerA or coo_val coo_col_ind or coo_row_ind pointer is invalid.

rocsparse_prune_dense2csr_buffer_size()#

rocsparse_status rocsparse_sprune_dense2csr_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const float *A, rocsparse_int lda, const float *threshold, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, size_t *buffer_size)#
rocsparse_status rocsparse_dprune_dense2csr_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const double *A, rocsparse_int lda, const double *threshold, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, size_t *buffer_size)#

This function computes the the size of the user allocated temporary storage buffer used when converting and pruning a dense matrix to a CSR matrix.

rocsparse_prune_dense2csr_buffer_size returns the size of the temporary storage buffer that is required by rocsparse_sprune_dense2csr_nnz(), rocsparse_dprune_dense2csr_nnz(), rocsparse_sprune_dense2csr(), and rocsparse_dprune_dense2csr(). The temporary storage buffer must be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the dense matrix A.

  • n[in] number of columns of the dense matrix A.

  • A[in] array of dimensions (lda, n)

  • lda[in] leading dimension of dense array A.

  • threshold[in] pointer to the pruning non-negative threshold which can exist in either host or device memory.

  • descr[in] the descriptor of the dense matrix A, the supported matrix type is rocsparse_matrix_type_general and also any valid value of the rocsparse_index_base.

  • csr_val[in] array of nnz ( = csr_row_ptr[m] - csr_row_ptr[0] ) nonzero elements of matrix A.

  • csr_row_ptr[in] integer array of m+1 elements that contains the start of every row and the end of the last row plus one.

  • csr_col_ind[in] integer array of nnz ( = csr_row_ptr[m] - csr_row_ptr[0] ) column indices of the non-zero elements of matrix A.

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_sprune_dense2csr_nnz(), rocsparse_dprune_dense2csr_nnz(), rocsparse_sprune_dense2csr() and rocsparse_dprune_dense2csr().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerbuffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_prune_dense2csr_nnz()#

rocsparse_status rocsparse_sprune_dense2csr_nnz(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const float *A, rocsparse_int lda, const float *threshold, const rocsparse_mat_descr descr, rocsparse_int *csr_row_ptr, rocsparse_int *nnz_total_dev_host_ptr, void *temp_buffer)#
rocsparse_status rocsparse_dprune_dense2csr_nnz(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const double *A, rocsparse_int lda, const double *threshold, const rocsparse_mat_descr descr, rocsparse_int *csr_row_ptr, rocsparse_int *nnz_total_dev_host_ptr, void *temp_buffer)#

This function computes the number of nonzero elements per row and the total number of nonzero elements in a dense matrix once elements less than the threshold are pruned from the matrix.

The routine does support asynchronous execution if the pointer mode is set to device.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the dense matrix A.

  • n[in] number of columns of the dense matrix A.

  • A[in] array of dimensions (lda, n)

  • lda[in] leading dimension of dense array A.

  • threshold[in] pointer to the pruning non-negative threshold which can exist in either host or device memory.

  • descr[in] the descriptor of the dense matrix A.

  • csr_row_ptr[out] integer array of m+1 elements that contains the start of every row and the end of the last row plus one.

  • nnz_total_dev_host_ptr[out] total number of nonzero elements in device or host memory.

  • temp_buffer[out] buffer allocated by the user whose size is determined by calling rocsparse_xprune_dense2csr_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or lda is invalid.

  • rocsparse_status_invalid_pointerA or threshold or descr or csr_row_ptr or nnz_total_dev_host_ptr or temp_buffer pointer is invalid.

rocsparse_prune_dense2csr()#

rocsparse_status rocsparse_sprune_dense2csr(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const float *A, rocsparse_int lda, const float *threshold, const rocsparse_mat_descr descr, float *csr_val, const rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind, void *temp_buffer)#
rocsparse_status rocsparse_dprune_dense2csr(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const double *A, rocsparse_int lda, const double *threshold, const rocsparse_mat_descr descr, double *csr_val, const rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind, void *temp_buffer)#

This function converts the matrix A in dense format into a sparse matrix in CSR format while pruning values that are less than the threshold. All the parameters are assumed to have been pre-allocated by the user.

The user first allocates csr_row_ptr to have m+1 elements and then calls rocsparse_xprune_dense2csr_nnz() which fills in the csr_row_ptr array and stores the number of elements that are larger than the pruning threshold in nnz_total_dev_host_ptr. The user then allocates csr_col_ind and csr_val to have size nnz_total_dev_host_ptr and completes the conversion by calling rocsparse_xprune_dense2csr(). A temporary storage buffer is used by both rocsparse_xprune_dense2csr_nnz() and rocsparse_xprune_dense2csr() and must be allocated by the user and whose size is determined by rocsparse_xprune_dense2csr_buffer_size(). The routine rocsparse_xprune_dense2csr() is executed asynchronously with respect to the host and may return control to the application on the host before the entire result is ready.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the dense matrix A.

  • n[in] number of columns of the dense matrix A.

  • A[in] array of dimensions (lda, n)

  • lda[in] leading dimension of dense array A.

  • threshold[in] pointer to the non-negative pruning threshold which can exist in either host or device memory.

  • descr[in] the descriptor of the dense matrix A, the supported matrix type is rocsparse_matrix_type_general and also any valid value of the rocsparse_index_base.

  • csr_val[out] array of nnz ( = csr_row_ptr[m] - csr_row_ptr[0] ) nonzero elements of matrix A.

  • csr_row_ptr[in] integer array of m+1 elements that contains the start of every row and the end of the last row plus one.

  • csr_col_ind[out] integer array of nnz ( = csr_row_ptr[m] - csr_row_ptr[0] ) column indices of the non-zero elements of matrix A.

  • temp_buffer[in] temporary storage buffer allocated by the user, size is returned by rocsparse_xprune_dense2csr_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or lda is invalid.

  • rocsparse_status_invalid_pointerA or descr or threshold or csr_val or csr_row_ptr or csr_col_ind or temp_buffer pointer is invalid.

rocsparse_prune_csr2csr_buffer_size()#

rocsparse_status rocsparse_sprune_csr2csr_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz_A, const rocsparse_mat_descr csr_descr_A, const float *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const float *threshold, const rocsparse_mat_descr csr_descr_C, const float *csr_val_C, const rocsparse_int *csr_row_ptr_C, const rocsparse_int *csr_col_ind_C, size_t *buffer_size)#
rocsparse_status rocsparse_dprune_csr2csr_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz_A, const rocsparse_mat_descr csr_descr_A, const double *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const double *threshold, const rocsparse_mat_descr csr_descr_C, const double *csr_val_C, const rocsparse_int *csr_row_ptr_C, const rocsparse_int *csr_col_ind_C, size_t *buffer_size)#

Convert and prune sparse CSR matrix into a sparse CSR matrix.

rocsparse_prune_csr2csr_buffer_size returns the size of the temporary buffer that is required by rocsparse_sprune_csr2csr_nnz, rocsparse_dprune_csr2csr_nnz, rocsparse_sprune_csr2csr, and rocsparse_dprune_csr2csr. The temporary storage buffer must be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows in the sparse CSR matrix.

  • n[in] number of columns in the sparse CSR matrix.

  • nnz_A[in] number of non-zeros in the sparse CSR matrix A.

  • csr_descr_A[in] descriptor of the sparse CSR matrix A. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val_A[in] array of nnz_A elements containing the values of the sparse CSR matrix A.

  • csr_row_ptr_A[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix A.

  • csr_col_ind_A[in] array of nnz_A elements containing the column indices of the sparse CSR matrix A.

  • threshold[in] pointer to the non-negative pruning threshold which can exist in either host or device memory.

  • csr_descr_C[in] descriptor of the sparse CSR matrix C. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val_C[in] array of nnz_C elements containing the values of the sparse CSR matrix C.

  • csr_row_ptr_C[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix C.

  • csr_col_ind_C[in] array of nnz_C elements containing the column indices of the sparse CSR matrix C.

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_sprune_csr2csr_nnz, rocsparse_dprune_csr2csr_nnz, rocsparse_sprune_csr2csr, and rocsparse_dprune_csr2csr.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerbuffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_prune_csr2csr_nnz()#

rocsparse_status rocsparse_sprune_csr2csr_nnz(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz_A, const rocsparse_mat_descr csr_descr_A, const float *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const float *threshold, const rocsparse_mat_descr csr_descr_C, rocsparse_int *csr_row_ptr_C, rocsparse_int *nnz_total_dev_host_ptr, void *temp_buffer)#
rocsparse_status rocsparse_dprune_csr2csr_nnz(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz_A, const rocsparse_mat_descr csr_descr_A, const double *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const double *threshold, const rocsparse_mat_descr csr_descr_C, rocsparse_int *csr_row_ptr_C, rocsparse_int *nnz_total_dev_host_ptr, void *temp_buffer)#

Convert and prune sparse CSR matrix into a sparse CSR matrix.

rocsparse_prune_csr2csr_nnz computes the number of nonzero elements per row and the total number of nonzero elements in a sparse CSR matrix once elements less than the threshold are pruned from the matrix.

Note

The routine does support asynchronous execution if the pointer mode is set to device.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows in the sparse CSR matrix.

  • n[in] number of columns in the sparse CSR matrix.

  • nnz_A[in] number of non-zeros in the sparse CSR matrix A.

  • csr_descr_A[in] descriptor of the sparse CSR matrix A. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val_A[in] array of nnz_A elements containing the values of the sparse CSR matrix A.

  • csr_row_ptr_A[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix A.

  • csr_col_ind_A[in] array of nnz_A elements containing the column indices of the sparse CSR matrix A.

  • threshold[in] pointer to the non-negative pruning threshold which can exist in either host or device memory.

  • csr_descr_C[in] descriptor of the sparse CSR matrix C. Currently, only rocsparse_matrix_type_general is supported.

  • csr_row_ptr_C[out] array of m+1 elements that point to the start of every row of the sparse CSR matrix C.

  • nnz_total_dev_host_ptr[out] total number of nonzero elements in device or host memory.

  • temp_buffer[out] buffer allocated by the user whose size is determined by calling rocsparse_xprune_csr2csr_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or nnz_A is invalid.

  • rocsparse_status_invalid_pointerthreshold or csr_descr_A or csr_descr_C or csr_val_A or csr_row_ptr_A or csr_col_ind_A or csr_row_ptr_C or nnz_total_dev_host_ptr or temp_buffer pointer is invalid.

rocsparse_prune_csr2csr()#

rocsparse_status rocsparse_sprune_csr2csr(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz_A, const rocsparse_mat_descr csr_descr_A, const float *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const float *threshold, const rocsparse_mat_descr csr_descr_C, float *csr_val_C, const rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C, void *temp_buffer)#
rocsparse_status rocsparse_dprune_csr2csr(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz_A, const rocsparse_mat_descr csr_descr_A, const double *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, const double *threshold, const rocsparse_mat_descr csr_descr_C, double *csr_val_C, const rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C, void *temp_buffer)#

Convert and prune sparse CSR matrix into a sparse CSR matrix.

This function converts the sparse CSR matrix A into a sparse CSR matrix C by pruning values in A that are less than the threshold. All the parameters are assumed to have been pre-allocated by the user. The user first calls rocsparse_xprune_csr2csr_buffer_size() to determine the size of the buffer used by rocsparse_xprune_csr2csr_nnz() and rocsparse_xprune_csr2csr() which the user then allocates. The user then allocates csr_row_ptr_C to have m+1 elements and then calls rocsparse_xprune_csr2csr_nnz() which fills in the csr_row_ptr_C array stores then number of elements that are larger than the pruning threshold in nnz_total_dev_host_ptr. The user then calls rocsparse_xprune_csr2csr() to complete the conversion. It is executed asynchronously with respect to the host and may return control to the application on the host before the entire result is ready.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows in the sparse CSR matrix.

  • n[in] number of columns in the sparse CSR matrix.

  • nnz_A[in] number of non-zeros in the sparse CSR matrix A.

  • csr_descr_A[in] descriptor of the sparse CSR matrix A. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val_A[in] array of nnz_A elements containing the values of the sparse CSR matrix A.

  • csr_row_ptr_A[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix A.

  • csr_col_ind_A[in] array of nnz_A elements containing the column indices of the sparse CSR matrix A.

  • threshold[in] pointer to the non-negative pruning threshold which can exist in either host or device memory.

  • csr_descr_C[in] descriptor of the sparse CSR matrix C. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val_C[out] array of nnz_C elements containing the values of the sparse CSR matrix C.

  • csr_row_ptr_C[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix C.

  • csr_col_ind_C[out] array of nnz_C elements containing the column indices of the sparse CSR matrix C.

  • temp_buffer[in] buffer allocated by the user whose size is determined by calling rocsparse_xprune_csr2csr_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or nnz_A is invalid.

  • rocsparse_status_invalid_pointerthreshold or csr_descr_A or csr_descr_C or csr_val_A or csr_row_ptr_A or csr_col_ind_A or csr_val_C or csr_row_ptr_C or csr_col_ind_C or temp_buffer pointer is invalid.

rocsparse_prune_dense2csr_by_percentage_buffer_size()#

rocsparse_status rocsparse_sprune_dense2csr_by_percentage_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const float *A, rocsparse_int lda, float percentage, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_dprune_dense2csr_by_percentage_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const double *A, rocsparse_int lda, double percentage, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_mat_info info, size_t *buffer_size)#

This function computes the size of the user allocated temporary storage buffer used when converting and pruning by percentage a dense matrix to a CSR matrix.

When converting and pruning a dense matrix A to a CSR matrix by percentage the following steps are performed. First the user calls rocsparse_prune_dense2csr_by_percentage_buffer_size which determines the size of the temporary storage buffer. Once determined, this buffer must be allocated by the user. Next the user allocates the csr_row_ptr array to have m+1 elements and calls rocsparse_prune_dense2csr_nnz_by_percentage. Finally the user finishes the conversion by allocating the csr_col_ind and csr_val arrays (whos size is determined by the value at nnz_total_dev_host_ptr) and calling rocsparse_prune_dense2csr_by_percentage.

The pruning by percentage works by first sorting the absolute values of the dense matrix A. We then determine a position in this sorted array by

\[ pos = ceil(m*n*(percentage/100)) - 1 pos = min(pos, m*n-1) pos = max(pos, 0) threshold = sorted_A[pos] \]
Once we have this threshold we prune values in the dense matrix A as in rocsparse_prune_dense2csr. It is executed asynchronously with respect to the host and may return control to the application on the host before the entire result is ready.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the dense matrix A.

  • n[in] number of columns of the dense matrix A.

  • A[in] array of dimensions (lda, n)

  • lda[in] leading dimension of dense array A.

  • percentage[in] percentage >= 0 and percentage <= 100.

  • descr[in] the descriptor of the dense matrix A, the supported matrix type is rocsparse_matrix_type_general and also any valid value of the rocsparse_index_base.

  • csr_val[in] array of nnz ( = csr_row_ptr[m] - csr_row_ptr[0] ) nonzero elements of matrix A.

  • csr_row_ptr[in] integer array of m+1 elements that contains the start of every row and the end of the last row plus one.

  • csr_col_ind[in] integer array of nnz ( = csr_row_ptr[m] - csr_row_ptr[0] ) column indices of the non-zero elements of matrix A.

  • info[in] prune information structure

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_sprune_dense2csr_nnz_by_percentage(), rocsparse_dprune_dense2csr_nnz_by_percentage(), rocsparse_sprune_dense2csr_by_percentage() and rocsparse_dprune_dense2csr_by_percentage().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerbuffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_prune_dense2csr_nnz_by_percentage()#

rocsparse_status rocsparse_sprune_dense2csr_nnz_by_percentage(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const float *A, rocsparse_int lda, float percentage, const rocsparse_mat_descr descr, rocsparse_int *csr_row_ptr, rocsparse_int *nnz_total_dev_host_ptr, rocsparse_mat_info info, void *temp_buffer)#
rocsparse_status rocsparse_dprune_dense2csr_nnz_by_percentage(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const double *A, rocsparse_int lda, double percentage, const rocsparse_mat_descr descr, rocsparse_int *csr_row_ptr, rocsparse_int *nnz_total_dev_host_ptr, rocsparse_mat_info info, void *temp_buffer)#

This function computes the number of nonzero elements per row and the total number of nonzero elements in a dense matrix when converting and pruning by percentage a dense matrix to a CSR matrix.

When converting and pruning a dense matrix A to a CSR matrix by percentage the following steps are performed. First the user calls rocsparse_prune_dense2csr_by_percentage_buffer_size which determines the size of the temporary storage buffer. Once determined, this buffer must be allocated by the user. Next the user allocates the csr_row_ptr array to have m+1 elements and calls rocsparse_prune_dense2csr_nnz_by_percentage. Finally the user finishes the conversion by allocating the csr_col_ind and csr_val arrays (whos size is determined by the value at nnz_total_dev_host_ptr) and calling rocsparse_prune_dense2csr_by_percentage.

The pruning by percentage works by first sorting the absolute values of the dense matrix A. We then determine a position in this sorted array by

\[ pos = ceil(m*n*(percentage/100)) - 1 pos = min(pos, m*n-1) pos = max(pos, 0) threshold = sorted_A[pos] \]
Once we have this threshold we prune values in the dense matrix A as in rocsparse_prune_dense2csr. The routine does support asynchronous execution if the pointer mode is set to device.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the dense matrix A.

  • n[in] number of columns of the dense matrix A.

  • A[in] array of dimensions (lda, n)

  • lda[in] leading dimension of dense array A.

  • percentage[in] percentage >= 0 and percentage <= 100.

  • descr[in] the descriptor of the dense matrix A.

  • csr_row_ptr[out] integer array of m+1 elements that contains the start of every row and the end of the last row plus one.

  • nnz_total_dev_host_ptr[out] total number of nonzero elements in device or host memory.

  • info[in] prune information structure

  • temp_buffer[out] buffer allocated by the user whose size is determined by calling rocsparse_xprune_dense2csr_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or lda or percentage is invalid.

  • rocsparse_status_invalid_pointerA or descr or info or csr_row_ptr or nnz_total_dev_host_ptr or temp_buffer pointer is invalid.

rocsparse_prune_dense2csr_by_percentage()#

rocsparse_status rocsparse_sprune_dense2csr_by_percentage(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const float *A, rocsparse_int lda, float percentage, const rocsparse_mat_descr descr, float *csr_val, const rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind, rocsparse_mat_info info, void *temp_buffer)#
rocsparse_status rocsparse_dprune_dense2csr_by_percentage(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, const double *A, rocsparse_int lda, double percentage, const rocsparse_mat_descr descr, double *csr_val, const rocsparse_int *csr_row_ptr, rocsparse_int *csr_col_ind, rocsparse_mat_info info, void *temp_buffer)#

This function converts the matrix A in dense format into a sparse matrix in CSR format while pruning values based on percentage.

When converting and pruning a dense matrix A to a CSR matrix by percentage the following steps are performed. First the user calls rocsparse_prune_dense2csr_by_percentage_buffer_size which determines the size of the temporary storage buffer. Once determined, this buffer must be allocated by the user. Next the user allocates the csr_row_ptr array to have m+1 elements and calls rocsparse_prune_dense2csr_nnz_by_percentage. Finally the user finishes the conversion by allocating the csr_col_ind and csr_val arrays (whos size is determined by the value at nnz_total_dev_host_ptr) and calling rocsparse_prune_dense2csr_by_percentage.

The pruning by percentage works by first sorting the absolute values of the dense matrix A. We then determine a position in this sorted array by

\[ pos = ceil(m*n*(percentage/100)) - 1 pos = min(pos, m*n-1) pos = max(pos, 0) threshold = sorted_A[pos] \]
Once we have this threshold we prune values in the dense matrix A as in rocsparse_prune_dense2csr. It is executed asynchronously with respect to the host and may return control to the application on the host before the entire result is ready.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the dense matrix A.

  • n[in] number of columns of the dense matrix A.

  • A[in] array of dimensions (lda, n)

  • lda[in] leading dimension of dense array A.

  • percentage[in] percentage >= 0 and percentage <= 100.

  • descr[in] the descriptor of the dense matrix A, the supported matrix type is rocsparse_matrix_type_general and also any valid value of the rocsparse_index_base.

  • csr_val[out] array of nnz ( = csr_row_ptr[m] - csr_row_ptr[0] ) nonzero elements of matrix A.

  • csr_row_ptr[in] integer array of m+1 elements that contains the start of every row and the end of the last row plus one.

  • csr_col_ind[out] integer array of nnz ( = csr_row_ptr[m] - csr_row_ptr[0] ) column indices of the non-zero elements of matrix A.

  • info[in] prune information structure

  • temp_buffer[in] temporary storage buffer allocated by the user, size is returned by rocsparse_xprune_dense2csr_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or lda or percentage is invalid.

  • rocsparse_status_invalid_pointerA or descr or info or csr_val or csr_row_ptr or csr_col_ind or temp_buffer pointer is invalid.

rocsparse_prune_csr2csr_by_percentage_buffer_size()#

rocsparse_status rocsparse_sprune_csr2csr_by_percentage_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz_A, const rocsparse_mat_descr csr_descr_A, const float *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, float percentage, const rocsparse_mat_descr csr_descr_C, const float *csr_val_C, const rocsparse_int *csr_row_ptr_C, const rocsparse_int *csr_col_ind_C, rocsparse_mat_info info, size_t *buffer_size)#
rocsparse_status rocsparse_dprune_csr2csr_by_percentage_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz_A, const rocsparse_mat_descr csr_descr_A, const double *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, double percentage, const rocsparse_mat_descr csr_descr_C, const double *csr_val_C, const rocsparse_int *csr_row_ptr_C, const rocsparse_int *csr_col_ind_C, rocsparse_mat_info info, size_t *buffer_size)#

Convert and prune by percentage a sparse CSR matrix into a sparse CSR matrix.

rocsparse_prune_csr2csr__by_percentage_buffer_size returns the size of the temporary buffer that is required by rocsparse_sprune_csr2csr_nnz_by_percentage, rocsparse_dprune_csr2csr_nnz_by_percentage, rocsparse_sprune_csr2csr_by_percentage, and rocsparse_dprune_csr2csr_by_percentage. The temporary storage buffer must be allocated by the user.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows in the sparse CSR matrix.

  • n[in] number of columns in the sparse CSR matrix.

  • nnz_A[in] number of non-zeros in the sparse CSR matrix A.

  • csr_descr_A[in] descriptor of the sparse CSR matrix A. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val_A[in] array of nnz_A elements containing the values of the sparse CSR matrix A.

  • csr_row_ptr_A[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix A.

  • csr_col_ind_A[in] array of nnz_A elements containing the column indices of the sparse CSR matrix A.

  • percentage[in] percentage >= 0 and percentage <= 100.

  • csr_descr_C[in] descriptor of the sparse CSR matrix C. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val_C[in] array of nnz_C elements containing the values of the sparse CSR matrix C.

  • csr_row_ptr_C[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix C.

  • csr_col_ind_C[in] array of nnz_C elements containing the column indices of the sparse CSR matrix C.

  • info[in] prune info structure.

  • buffer_size[out] number of bytes of the temporary storage buffer required by rocsparse_sprune_csr2csr_nnz_by_percentage, rocsparse_dprune_csr2csr_nnz_by_percentage, rocsparse_sprune_csr2csr_by_percentage, and rocsparse_dprune_csr2csr_by_percentage.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerbuffer_size pointer is invalid.

  • rocsparse_status_internal_error – an internal error occurred.

rocsparse_prune_csr2csr_nnz_by_percentage()#

rocsparse_status rocsparse_sprune_csr2csr_nnz_by_percentage(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz_A, const rocsparse_mat_descr csr_descr_A, const float *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, float percentage, const rocsparse_mat_descr csr_descr_C, rocsparse_int *csr_row_ptr_C, rocsparse_int *nnz_total_dev_host_ptr, rocsparse_mat_info info, void *temp_buffer)#
rocsparse_status rocsparse_dprune_csr2csr_nnz_by_percentage(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz_A, const rocsparse_mat_descr csr_descr_A, const double *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, double percentage, const rocsparse_mat_descr csr_descr_C, rocsparse_int *csr_row_ptr_C, rocsparse_int *nnz_total_dev_host_ptr, rocsparse_mat_info info, void *temp_buffer)#

Convert and prune by percentage a sparse CSR matrix into a sparse CSR matrix.

rocsparse_prune_csr2csr_nnz_by_percentage computes the number of nonzero elements per row and the total number of nonzero elements in a sparse CSR matrix once elements less than the threshold are pruned from the matrix.

Note

The routine does support asynchronous execution if the pointer mode is set to device.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows in the sparse CSR matrix.

  • n[in] number of columns in the sparse CSR matrix.

  • nnz_A[in] number of non-zeros in the sparse CSR matrix A.

  • csr_descr_A[in] descriptor of the sparse CSR matrix A. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val_A[in] array of nnz_A elements containing the values of the sparse CSR matrix A.

  • csr_row_ptr_A[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix A.

  • csr_col_ind_A[in] array of nnz_A elements containing the column indices of the sparse CSR matrix A.

  • percentage[in] percentage >= 0 and percentage <= 100.

  • csr_descr_C[in] descriptor of the sparse CSR matrix C. Currently, only rocsparse_matrix_type_general is supported.

  • csr_row_ptr_C[out] array of m+1 elements that point to the start of every row of the sparse CSR matrix C.

  • nnz_total_dev_host_ptr[out] total number of nonzero elements in device or host memory.

  • info[in] prune info structure.

  • temp_buffer[out] buffer allocated by the user whose size is determined by calling rocsparse_xprune_csr2csr_by_percentage_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or nnz_A or percentage is invalid.

  • rocsparse_status_invalid_pointercsr_descr_A or csr_descr_C or info or csr_val_A or csr_row_ptr_A or csr_col_ind_A or csr_row_ptr_C or nnz_total_dev_host_ptr or temp_buffer pointer is invalid.

rocsparse_prune_csr2csr_by_percentage()#

rocsparse_status rocsparse_sprune_csr2csr_by_percentage(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz_A, const rocsparse_mat_descr csr_descr_A, const float *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, float percentage, const rocsparse_mat_descr csr_descr_C, float *csr_val_C, const rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C, rocsparse_mat_info info, void *temp_buffer)#
rocsparse_status rocsparse_dprune_csr2csr_by_percentage(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz_A, const rocsparse_mat_descr csr_descr_A, const double *csr_val_A, const rocsparse_int *csr_row_ptr_A, const rocsparse_int *csr_col_ind_A, double percentage, const rocsparse_mat_descr csr_descr_C, double *csr_val_C, const rocsparse_int *csr_row_ptr_C, rocsparse_int *csr_col_ind_C, rocsparse_mat_info info, void *temp_buffer)#

Convert and prune by percentage a sparse CSR matrix into a sparse CSR matrix.

This function converts the sparse CSR matrix A into a sparse CSR matrix C by pruning values in A that are less than the threshold. All the parameters are assumed to have been pre-allocated by the user. The user first calls rocsparse_xprune_csr2csr_buffer_size() to determine the size of the buffer used by rocsparse_xprune_csr2csr_nnz() and rocsparse_xprune_csr2csr() which the user then allocates. The user then allocates csr_row_ptr_C to have m+1 elements and then calls rocsparse_xprune_csr2csr_nnz() which fills in the csr_row_ptr_C array stores then number of elements that are larger than the pruning threshold in nnz_total_dev_host_ptr. The user then calls rocsparse_xprune_csr2csr() to complete the conversion. It is executed asynchronously with respect to the host and may return control to the application on the host before the entire result is ready.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows in the sparse CSR matrix.

  • n[in] number of columns in the sparse CSR matrix.

  • nnz_A[in] number of non-zeros in the sparse CSR matrix A.

  • csr_descr_A[in] descriptor of the sparse CSR matrix A. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val_A[in] array of nnz_A elements containing the values of the sparse CSR matrix A.

  • csr_row_ptr_A[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix A.

  • csr_col_ind_A[in] array of nnz_A elements containing the column indices of the sparse CSR matrix A.

  • percentage[in] percentage >= 0 and percentage <= 100.

  • csr_descr_C[in] descriptor of the sparse CSR matrix C. Currently, only rocsparse_matrix_type_general is supported.

  • csr_val_C[out] array of nnz_C elements containing the values of the sparse CSR matrix C.

  • csr_row_ptr_C[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix C.

  • csr_col_ind_C[out] array of nnz_C elements containing the column indices of the sparse CSR matrix C.

  • info[in] prune info structure.

  • temp_buffer[in] buffer allocated by the user whose size is determined by calling rocsparse_xprune_csr2csr_buffer_size().

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or n or nnz_A or percentage is invalid.

  • rocsparse_status_invalid_pointercsr_descr_A or csr_descr_C or info or csr_val_A or csr_row_ptr_A or csr_col_ind_A or csr_val_C or csr_row_ptr_C or csr_col_ind_C or temp_buffer pointer is invalid.

rocsparse_rocsparse_bsrpad_value()#

rocsparse_status rocsparse_sbsrpad_value(rocsparse_handle handle, rocsparse_int m, rocsparse_int mb, rocsparse_int nnzb, rocsparse_int block_dim, float value, const rocsparse_mat_descr bsr_descr, float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind)#
rocsparse_status rocsparse_dbsrpad_value(rocsparse_handle handle, rocsparse_int m, rocsparse_int mb, rocsparse_int nnzb, rocsparse_int block_dim, double value, const rocsparse_mat_descr bsr_descr, double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind)#
rocsparse_status rocsparse_cbsrpad_value(rocsparse_handle handle, rocsparse_int m, rocsparse_int mb, rocsparse_int nnzb, rocsparse_int block_dim, rocsparse_float_complex value, const rocsparse_mat_descr bsr_descr, rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind)#
rocsparse_status rocsparse_zbsrpad_value(rocsparse_handle handle, rocsparse_int m, rocsparse_int mb, rocsparse_int nnzb, rocsparse_int block_dim, rocsparse_double_complex value, const rocsparse_mat_descr bsr_descr, rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind)#

Pads a value to the diagonal of the last block (if the last block is a diagonal block) in the sparse BSR matrix when the matrix expands outside m x m.

When converting from a CSR matrix to a BSR matrix the resulting BSR matrix will be larger when m < mb * block_dim. In these situations, the CSR to BSR conversion will expand the BSR matrix to have zeros when outside m x m. This routine converts the resulting BSR matrix to one that has a value on the last diagonal blocks diagonal if this last block is a diagonal block in the BSR matrix.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of the sparse BSR matrix.

  • mb[in] number of block rows of the sparse BSR matrix.

  • nnzb[in] number of non-zero blocks of the sparse BSR matrix.

  • block_dim[in] block dimension of the sparse BSR matrix.

  • value[in] scalar value that is set on the diagonal of the last block when the matrix expands outside of m x m

  • bsr_descr[in] descriptor of the sparse BSR matrix. Currently, only rocsparse_matrix_type_general is supported.

  • bsr_val[inout] array of nnzb blocks of the sparse BSR matrix.

  • bsr_row_ptr[in] array of mb+1 elements that point to the start of every block row of the sparse BSR matrix.

  • bsr_col_ind[in] array of nnzb elements containing the block column indices of the sparse BSR matrix.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem, mb, nnzb or block_dim is invalid.

  • rocsparse_status_invalid_pointerbsr_descr, bsr_val, bsr_row_ind, bsr_col_ind, pointer is invalid.

Reordering Functions#

This module holds all sparse reordering routines.

The sparse reordering routines describe algorithm for reordering sparse matrices.

rocsparse_csrcolor()#

rocsparse_status rocsparse_scsrcolor(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const float *fraction_to_color, rocsparse_int *ncolors, rocsparse_int *coloring, rocsparse_int *reordering, rocsparse_mat_info info)#
rocsparse_status rocsparse_dcsrcolor(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const double *fraction_to_color, rocsparse_int *ncolors, rocsparse_int *coloring, rocsparse_int *reordering, rocsparse_mat_info info)#
rocsparse_status rocsparse_ccsrcolor(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const float *fraction_to_color, rocsparse_int *ncolors, rocsparse_int *coloring, rocsparse_int *reordering, rocsparse_mat_info info)#
rocsparse_status rocsparse_zcsrcolor(rocsparse_handle handle, rocsparse_int m, rocsparse_int nnz, const rocsparse_mat_descr descr, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, const double *fraction_to_color, rocsparse_int *ncolors, rocsparse_int *coloring, rocsparse_int *reordering, rocsparse_mat_info info)#

Coloring of the adjacency graph of the matrix \(A\) stored in the CSR format.

rocsparse_csrcolor performs the coloring of the undirected graph represented by the (symmetric) sparsity pattern of the matrix \(A\) stored in CSR format. Graph coloring is a way of coloring the nodes of a graph such that no two adjacent nodes are of the same color. The fraction_to_color is a parameter to only color a given percentage of the graph nodes, the remaining uncolored nodes receive distinct new colors. The optional reordering array is a permutation array such that unknowns of the same color are grouped. The matrix \(A\) must be stored as a general matrix with a symmetric sparsity pattern, and if the matrix \(A\) is non-symmetric then the user is responsible to provide the symmetric part \(\frac{A+A^T}{2}\).

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • m[in] number of rows of sparse matrix \(A\).

  • nnz[in] number of non-zero entries of sparse matrix \(A\).

  • descr[in] sparse matrix descriptor.

  • csr_val[in] array of nnz elements of the sparse CSR matrix.

  • csr_row_ptr[in] array of m+1 elements that point to the start of every row of the sparse CSR matrix.

  • csr_col_ind[in] array of nnz elements containing the column indices of the sparse CSR matrix.

  • fraction_to_color[in] fraction of nodes to be colored, which should be in the interval [0.0,1.0], for example 0.8 implies that 80 percent of nodes will be colored.

  • ncolors[out] resulting number of distinct colors.

  • coloring[out] resulting mapping of colors.

  • reordering[out] optional resulting reordering permutation if reordering is a non-null pointer.

  • info[inout] structure that holds the information collected during the coloring algorithm.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_sizem or nnz is invalid.

  • rocsparse_status_invalid_pointerdescr, csr_val, csr_row_ptr, csr_col_ind, fraction_to_color, ncolors, coloring or info pointer is invalid.

Utility Functions#

This module holds all sparse utility routines.

The sparse utility routines allow for testing whether matrix data is valid for different matrix formats

rocsparse_check_matrix_csr_buffer_size()#

rocsparse_status rocsparse_scheck_matrix_csr_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_dcheck_matrix_csr_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_ccheck_matrix_csr_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_zcheck_matrix_csr_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#

Check matrix to see if it is valid.

rocsparse_check_matrix_csr_buffer_size computes the required buffer size needed when calling rocsparse_check_matrix_csr

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base or matrix_type or uplo or storage is invalid.

  • rocsparse_status_invalid_sizem n or nnz is invalid.

  • rocsparse_status_invalid_pointercsr_val, csr_row_ptr, csr_col_ind or buffer_size pointer is invalid.

rocsparse_check_matrix_csr()#

rocsparse_status rocsparse_scheck_matrix_csr(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const float *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_dcheck_matrix_csr(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const double *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_ccheck_matrix_csr(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_float_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_zcheck_matrix_csr(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_double_complex *csr_val, const rocsparse_int *csr_row_ptr, const rocsparse_int *csr_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#

Check matrix to see if it is valid.

rocsparse_check_matrix_csr checks if the input CSR matrix is valid.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base or matrix_type or uplo or storage is invalid.

  • rocsparse_status_invalid_sizem n or nnz is invalid.

  • rocsparse_status_invalid_pointercsr_val, csr_row_ptr, csr_col_ind, temp_buffer or data_status pointer is invalid.

rocsparse_check_matrix_csc_buffer_size()#

rocsparse_status rocsparse_scheck_matrix_csc_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const float *csc_val, const rocsparse_int *csc_col_ptr, const rocsparse_int *csc_row_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_dcheck_matrix_csc_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const double *csc_val, const rocsparse_int *csc_col_ptr, const rocsparse_int *csc_row_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_ccheck_matrix_csc_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_float_complex *csc_val, const rocsparse_int *csc_col_ptr, const rocsparse_int *csc_row_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_zcheck_matrix_csc_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_double_complex *csc_val, const rocsparse_int *csc_col_ptr, const rocsparse_int *csc_row_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#

Check matrix to see if it is valid.

rocsparse_check_matrix_csc_buffer_size computes the required buffer size needed when calling rocsparse_check_matrix_csc

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base or matrix_type or uplo or storage is invalid.

  • rocsparse_status_invalid_sizem n or nnz is invalid.

  • rocsparse_status_invalid_pointercsc_val, csc_col_ptr, csc_row_ind or buffer_size pointer is invalid.

rocsparse_check_matrix_csc()#

rocsparse_status rocsparse_scheck_matrix_csc(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const float *csc_val, const rocsparse_int *csc_col_ptr, const rocsparse_int *csc_row_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_dcheck_matrix_csc(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const double *csc_val, const rocsparse_int *csc_col_ptr, const rocsparse_int *csc_row_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_ccheck_matrix_csc(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_float_complex *csc_val, const rocsparse_int *csc_col_ptr, const rocsparse_int *csc_row_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_zcheck_matrix_csc(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_double_complex *csc_val, const rocsparse_int *csc_col_ptr, const rocsparse_int *csc_row_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#

Check matrix to see if it is valid.

rocsparse_check_matrix_csc checks if the input CSC matrix is valid.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base or matrix_type or uplo or storage is invalid.

  • rocsparse_status_invalid_sizem n or nnz is invalid.

  • rocsparse_status_invalid_pointercsc_val, csc_col_ptr, csc_row_ind, temp_buffer or data_status pointer is invalid.

rocsparse_check_matrix_coo_buffer_size()#

rocsparse_status rocsparse_scheck_matrix_coo_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const float *coo_val, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_dcheck_matrix_coo_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const double *coo_val, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_ccheck_matrix_coo_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_float_complex *coo_val, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_zcheck_matrix_coo_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_double_complex *coo_val, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#

Check matrix to see if it is valid.

rocsparse_check_matrix_coo_buffer_size computes the required buffer size needed when calling rocsparse_check_matrix_coo

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base or matrix_type or uplo or storage is invalid.

  • rocsparse_status_invalid_sizem n or nnz is invalid.

  • rocsparse_status_invalid_pointercoo_val, coo_row_ind, coo_col_ind or buffer_size pointer is invalid.

rocsparse_check_matrix_coo()#

rocsparse_status rocsparse_scheck_matrix_coo(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const float *coo_val, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_dcheck_matrix_coo(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const double *coo_val, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_ccheck_matrix_coo(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_float_complex *coo_val, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_zcheck_matrix_coo(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int nnz, const rocsparse_double_complex *coo_val, const rocsparse_int *coo_row_ind, const rocsparse_int *coo_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#

Check matrix to see if it is valid.

rocsparse_check_matrix_coo checks if the input COO matrix is valid.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base or matrix_type or uplo or storage is invalid.

  • rocsparse_status_invalid_sizem n or nnz is invalid.

  • rocsparse_status_invalid_pointercoo_val, coo_row_ind, coo_col_ind, temp_buffer or data_status pointer is invalid.

rocsparse_check_matrix_gebsr_buffer_size()#

rocsparse_status rocsparse_scheck_matrix_gebsr_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_dcheck_matrix_gebsr_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_ccheck_matrix_gebsr_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_zcheck_matrix_gebsr_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#

Check matrix to see if it is valid.

rocsparse_check_matrix_gebsr_buffer_size computes the required buffer size needed when calling rocsparse_check_matrix_gebsr

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valuedir or idx_base or matrix_type or uplo or storage is invalid.

  • rocsparse_status_invalid_sizemb nb nnzb row_block_dim or col_block_dim is invalid.

  • rocsparse_status_invalid_pointerbsr_val, bsr_row_ptr, bsr_col_ind or buffer_size pointer is invalid.

rocsparse_check_matrix_gebsr()#

rocsparse_status rocsparse_scheck_matrix_gebsr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const float *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_dcheck_matrix_gebsr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const double *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_ccheck_matrix_gebsr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const rocsparse_float_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_zcheck_matrix_gebsr(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const rocsparse_double_complex *bsr_val, const rocsparse_int *bsr_row_ptr, const rocsparse_int *bsr_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#

Check matrix to see if it is valid.

rocsparse_check_matrix_gebsr checks if the input GEBSR matrix is valid.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valuedir or idx_base or matrix_type or uplo or storage is invalid.

  • rocsparse_status_invalid_sizemb nb nnzb row_block_dim or col_block_dim is invalid.

  • rocsparse_status_invalid_pointerbsr_val, bsr_row_ptr, bsr_col_ind, temp_buffer or data_status pointer is invalid.

rocsparse_check_matrix_gebsc_buffer_size()#

rocsparse_status rocsparse_scheck_matrix_gebsc_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const float *bsc_val, const rocsparse_int *bsc_col_ptr, const rocsparse_int *bsc_row_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_dcheck_matrix_gebsc_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const double *bsc_val, const rocsparse_int *bsc_col_ptr, const rocsparse_int *bsc_row_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_ccheck_matrix_gebsc_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const rocsparse_float_complex *bsc_val, const rocsparse_int *bsc_col_ptr, const rocsparse_int *bsc_row_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_zcheck_matrix_gebsc_buffer_size(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const rocsparse_double_complex *bsc_val, const rocsparse_int *bsc_col_ptr, const rocsparse_int *bsc_row_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#

Check matrix to see if it is valid.

rocsparse_check_matrix_gebsc_buffer_size computes the required buffer size needed when calling rocsparse_check_matrix_gebsc

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valuedir or idx_base or matrix_type or uplo or storage is invalid.

  • rocsparse_status_invalid_sizemb nb nnzb row_block_dim or col_block_dim is invalid.

  • rocsparse_status_invalid_pointerbsc_val, bsc_col_ptr, bsc_row_ind or buffer_size pointer is invalid.

rocsparse_check_matrix_gebsc()#

rocsparse_status rocsparse_scheck_matrix_gebsc(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const float *bsc_val, const rocsparse_int *bsc_col_ptr, const rocsparse_int *bsc_row_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_dcheck_matrix_gebsc(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const double *bsc_val, const rocsparse_int *bsc_col_ptr, const rocsparse_int *bsc_row_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_ccheck_matrix_gebsc(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const rocsparse_float_complex *bsc_val, const rocsparse_int *bsc_col_ptr, const rocsparse_int *bsc_row_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_zcheck_matrix_gebsc(rocsparse_handle handle, rocsparse_direction dir, rocsparse_int mb, rocsparse_int nb, rocsparse_int nnzb, rocsparse_int row_block_dim, rocsparse_int col_block_dim, const rocsparse_double_complex *bsc_val, const rocsparse_int *bsc_col_ptr, const rocsparse_int *bsc_row_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#

Check matrix to see if it is valid.

rocsparse_check_matrix_gebsc checks if the input GEBSC matrix is valid.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valuedir or idx_base or matrix_type or uplo or storage is invalid.

  • rocsparse_status_invalid_sizemb nb nnzb row_block_dim or col_block_dim is invalid.

  • rocsparse_status_invalid_pointerbsc_val, bsc_col_ptr, bsc_row_ind, temp_buffer or data_status pointer is invalid.

rocsparse_check_matrix_ell_buffer_size()#

rocsparse_status rocsparse_scheck_matrix_ell_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int ell_width, const float *ell_val, const rocsparse_int *ell_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_dcheck_matrix_ell_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int ell_width, const double *ell_val, const rocsparse_int *ell_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_ccheck_matrix_ell_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int ell_width, const rocsparse_float_complex *ell_val, const rocsparse_int *ell_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#
rocsparse_status rocsparse_zcheck_matrix_ell_buffer_size(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int ell_width, const rocsparse_double_complex *ell_val, const rocsparse_int *ell_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#

Check matrix to see if it is valid.

rocsparse_check_matrix_ell_buffer_size computes the required buffer size needed when calling rocsparse_check_matrix_ell

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base or matrix_type or uplo or storage is invalid.

  • rocsparse_status_invalid_sizem n or ell_width is invalid.

  • rocsparse_status_invalid_pointerell_val, ell_col_ind or buffer_size pointer is invalid.

rocsparse_check_matrix_ell()#

rocsparse_status rocsparse_scheck_matrix_ell(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int ell_width, const float *ell_val, const rocsparse_int *ell_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_dcheck_matrix_ell(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int ell_width, const double *ell_val, const rocsparse_int *ell_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_ccheck_matrix_ell(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int ell_width, const rocsparse_float_complex *ell_val, const rocsparse_int *ell_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#
rocsparse_status rocsparse_zcheck_matrix_ell(rocsparse_handle handle, rocsparse_int m, rocsparse_int n, rocsparse_int ell_width, const rocsparse_double_complex *ell_val, const rocsparse_int *ell_col_ind, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#

Check matrix to see if it is valid.

rocsparse_check_matrix_ell checks if the input ELL matrix is valid.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base or matrix_type or uplo or storage is invalid.

  • rocsparse_status_invalid_sizem n or ell_width is invalid.

  • rocsparse_status_invalid_pointerell_val, ell_col_ind, temp_buffer or data_status pointer is invalid.

rocsparse_check_matrix_hyb_buffer_size()#

rocsparse_status rocsparse_check_matrix_hyb_buffer_size(rocsparse_handle handle, const rocsparse_hyb_mat hyb, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, size_t *buffer_size)#

Check matrix to see if it is valid.

rocsparse_check_matrix_hyb_buffer_size computes the required buffer size needed when calling rocsparse_check_matrix_hyb

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base or matrix_type or uplo or storage is invalid.

  • rocsparse_status_invalid_pointerhyb or buffer_size pointer is invalid.

rocsparse_check_matrix_hyb()#

rocsparse_status rocsparse_check_matrix_hyb(rocsparse_handle handle, const rocsparse_hyb_mat hyb, rocsparse_index_base idx_base, rocsparse_matrix_type matrix_type, rocsparse_fill_mode uplo, rocsparse_storage_mode storage, rocsparse_data_status *data_status, void *temp_buffer)#

Check matrix to see if it is valid.

rocsparse_check_matrix_hyb checks if the input HYB matrix is valid.

Parameters:
Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_valueidx_base or matrix_type or uplo or storage is invalid.

  • rocsparse_status_invalid_pointerhyb or data_status pointer is invalid.

Sparse Generic Functions#

This module holds all sparse generic routines.

The sparse generic routines describe operations that manipulate sparse matrices.

rocsparse_axpby()#

rocsparse_status rocsparse_axpby(rocsparse_handle handle, const void *alpha, const rocsparse_spvec_descr x, const void *beta, rocsparse_dnvec_descr y)#

Scale a sparse vector and add it to a scaled dense vector.

rocsparse_axpby multiplies the sparse vector \(x\) with scalar \(\alpha\) and adds the result to the dense vector \(y\) that is multiplied with scalar \(\beta\), such that

\[ y := \alpha \cdot x + \beta \cdot y \]

for(i = 0; i < nnz; ++i)
{
    y[x_ind[i]] = alpha * x_val[i] + beta * y[x_ind[i]]
}

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • alpha[in] scalar \(\alpha\).

  • x[in] sparse matrix descriptor.

  • beta[in] scalar \(\beta\).

  • y[inout] dense matrix descriptor.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointeralpha, x, beta or y pointer is invalid.

rocsparse_gather()#

rocsparse_status rocsparse_gather(rocsparse_handle handle, const rocsparse_dnvec_descr y, rocsparse_spvec_descr x)#

Gather elements from a dense vector and store them into a sparse vector.

rocsparse_gather gathers the elements from the dense vector \(y\) and stores them in the sparse vector \(x\).

for(i = 0; i < nnz; ++i)
{
    x_val[i] = y[x_ind[i]];
}

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • y[in] dense vector \(y\).

  • x[out] sparse vector \(x\).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerx or y pointer is invalid.

rocsparse_scatter()#

rocsparse_status rocsparse_scatter(rocsparse_handle handle, const rocsparse_spvec_descr x, rocsparse_dnvec_descr y)#

Scatter elements from a sparse vector into a dense vector.

rocsparse_scatter scatters the elements from the sparse vector \(x\) in the dense vector \(y\).

for(i = 0; i < nnz; ++i)
{
    y[x_ind[i]] = x_val[i];
}

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • x[in] sparse vector \(x\).

  • y[out] dense vector \(y\).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerx or y pointer is invalid.

rocsparse_rot()#

rocsparse_status rocsparse_rot(rocsparse_handle handle, const void *c, const void *s, rocsparse_spvec_descr x, rocsparse_dnvec_descr y)#

Apply Givens rotation to a dense and a sparse vector.

rocsparse_rot applies the Givens rotation matrix \(G\) to the sparse vector \(x\) and the dense vector \(y\), where

\[\begin{split} G = \begin{pmatrix} c & s \\ -s & c \end{pmatrix} \end{split}\]

for(i = 0; i < nnz; ++i)
{
    x_tmp = x_val[i];
    y_tmp = y[x_ind[i]];

    x_val[i]    = c * x_tmp + s * y_tmp;
    y[x_ind[i]] = c * y_tmp - s * x_tmp;
}

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • c[in] pointer to the cosine element of \(G\), can be on host or device.

  • s[in] pointer to the sine element of \(G\), can be on host or device.

  • x[inout] sparse vector \(x\).

  • y[inout] dense vector \(y\).

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerc, s, x or y pointer is invalid.

rocsparse_spvv()#

rocsparse_status rocsparse_spvv(rocsparse_handle handle, rocsparse_operation trans, const rocsparse_spvec_descr x, const rocsparse_dnvec_descr y, void *result, rocsparse_datatype compute_type, size_t *buffer_size, void *temp_buffer)#

Sparse vector inner dot product.

rocsparse_spvv computes the inner dot product of the sparse vecotr \(x\) with the dense vector \(y\), such that

\[ \text{result} := x^{'} \cdot y, \]
with
\[\begin{split} op(x) = \left\{ \begin{array}{ll} x, & \text{if trans == rocsparse_operation_none} \\ \bar{x}, & \text{if trans == rocsparse_operation_conjugate_transpose} \\ \end{array} \right. \end{split}\]

result = 0;
for(i = 0; i < nnz; ++i)
{
    result += x_val[i] * y[x_ind[i]];
}

Note

This function writes the required allocation size (in bytes) to buffer_size and returns without performing the SpVV operation, when a nullptr is passed for temp_buffer.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans[in] sparse vector operation type.

  • x[in] sparse vector descriptor.

  • y[in] dense vector descriptor.

  • result[out] pointer to the result, can be host or device memory

  • compute_type[in] floating point precision for the SpVV computation.

  • buffer_size[out] number of bytes of the temporary storage buffer. buffer_size is set when temp_buffer is nullptr.

  • temp_buffer[in] temporary storage buffer allocated by the user. When a nullptr is passed, the required allocation size (in bytes) is written to buffer_size and function returns without performing the SpVV operation.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointerx, y, result or buffer_size pointer is invalid.

  • rocsparse_status_not_implementedcompute_type is currently not supported.

rocsparse_spmv()#

rocsparse_status rocsparse_spmv(rocsparse_handle handle, rocsparse_operation trans, const void *alpha, const rocsparse_spmat_descr mat, const rocsparse_dnvec_descr x, const void *beta, const rocsparse_dnvec_descr y, rocsparse_datatype compute_type, rocsparse_spmv_alg alg, size_t *buffer_size, void *temp_buffer)#

Sparse matrix vector multiplication.

rocsparse_spmv multiplies the scalar \(\alpha\) with a sparse \(m \times n\) matrix and the dense vector \(x\) and adds the result to the dense vector \(y\) that is multiplied by the scalar \(\beta\), such that

\[ y := \alpha \cdot op(A) \cdot x + \beta \cdot y, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans == rocsparse_operation_none} \\ A^T, & \text{if trans == rocsparse_operation_transpose} \\ A^H, & \text{if trans == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

Note

This function writes the required allocation size (in bytes) to buffer_size and returns without performing the SpMV operation, when a nullptr is passed for temp_buffer.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

The sparse matrix formats currently supported are: rocsparse_format_bsr, rocsparse_format_coo, rocsparse_format_coo_aos, rocsparse_format_csr, rocsparse_format_csc and rocsparse_format_ell.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans[in] matrix operation type.

  • alpha[in] scalar \(\alpha\).

  • mat[in] matrix descriptor.

  • x[in] vector descriptor.

  • beta[in] scalar \(\beta\).

  • y[inout] vector descriptor.

  • compute_type[in] floating point precision for the SpMV computation.

  • alg[in] SpMV algorithm for the SpMV computation.

  • buffer_size[out] number of bytes of the temporary storage buffer. buffer_size is set when temp_buffer is nullptr.

  • temp_buffer[in] temporary storage buffer allocated by the user. When a nullptr is passed, the required allocation size (in bytes) is written to buffer_size and function returns without performing the SpMV operation.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointeralpha, mat, x, beta, y or buffer_size pointer is invalid.

  • rocsparse_status_invalid_value – the value of trans, trans_B, compute_type, alg is incorrect.

  • rocsparse_status_not_implementedcompute_type or alg is currently not supported.

rocsparse_spmv_ex()#

rocsparse_status rocsparse_spmv_ex(rocsparse_handle handle, rocsparse_operation trans, const void *alpha, const rocsparse_spmat_descr mat, const rocsparse_dnvec_descr x, const void *beta, const rocsparse_dnvec_descr y, rocsparse_datatype compute_type, rocsparse_spmv_alg alg, rocsparse_spmv_stage stage, size_t *buffer_size, void *temp_buffer)#

Sparse matrix vector multiplication.

rocsparse_spmv_ex multiplies the scalar \(\alpha\) with a sparse \(m \times n\) matrix and the dense vector \(x\) and adds the result to the dense vector \(y\) that is multiplied by the scalar \(\beta\), such that

\[ y := \alpha \cdot op(A) \cdot x + \beta \cdot y, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans == rocsparse_operation_none} \\ A^T, & \text{if trans == rocsparse_operation_transpose} \\ A^H, & \text{if trans == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

Note

This function writes the required allocation size (in bytes) to buffer_size and returns without performing the SpMV operation, when a nullptr is passed for temp_buffer.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

SpMV_ex requires three stages to complete. The first stage rocsparse_spmv_stage_buffer_size will return the size of the temporary storage buffer that is required for subsequent calls to rocsparse_spmv_ex. The second stage rocsparse_spmv_stage_preprocess will preprocess data that would be saved in the temporary storage buffer. In the final stage rocsparse_spmv_stage_compute, the actual computation is performed.

Note

If rocsparse_spmv_stage_auto is selected, rocSPARSE will automatically detect which stage is required based on the following indicators: If temp_buffer is equal to nullptr, the required buffer size will be returned. Else, the SpMV_ex preprocess and the SpMV algorithm will be executed.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans[in] matrix operation type.

  • alpha[in] scalar \(\alpha\).

  • mat[in] matrix descriptor.

  • x[in] vector descriptor.

  • beta[in] scalar \(\beta\).

  • y[inout] vector descriptor.

  • compute_type[in] floating point precision for the SpMV computation.

  • alg[in] SpMV algorithm for the SpMV computation.

  • stage[in] SpMV stage for the SpMV computation.

  • buffer_size[out] number of bytes of the temporary storage buffer. buffer_size is set when temp_buffer is nullptr.

  • temp_buffer[in] temporary storage buffer allocated by the user. When a nullptr is passed, the required allocation size (in bytes) is written to buffer_size and function returns without performing the SpMV operation.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointeralpha, mat, x, beta, y or buffer_size pointer is invalid.

  • rocsparse_status_invalid_value – the value of trans, trans_B, compute_type, alg or stage is incorrect.

  • rocsparse_status_not_implementedcompute_type or alg is currently not supported.

rocsparse_spsv()#

rocsparse_status rocsparse_spsv(rocsparse_handle handle, rocsparse_operation trans, const void *alpha, const rocsparse_spmat_descr mat, const rocsparse_dnvec_descr x, const rocsparse_dnvec_descr y, rocsparse_datatype compute_type, rocsparse_spsv_alg alg, rocsparse_spsv_stage stage, size_t *buffer_size, void *temp_buffer)#

Sparse triangular solve.

rocsparse_spsv_solve solves a sparse triangular linear system of a sparse \(m \times m\) matrix, defined in CSR or COO storage format, a dense solution vector \(y\) and the right-hand side \(x\) that is multiplied by \(\alpha\), such that

\[ op(A) \cdot y = \alpha \cdot x, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans == rocsparse_operation_none} \\ A^T, & \text{if trans == rocsparse_operation_transpose} \\ A^H, & \text{if trans == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

Note

SpSV requires three stages to complete. The first stage rocsparse_spsv_stage_buffer_size will return the size of the temporary storage buffer that is required for subsequent calls. The second stage rocsparse_spsv_stage_preprocess will preprocess data that would be saved in the temporary storage buffer. In the final stage rocsparse_spsv_stage_compute, the actual computation is performed.

Note

If rocsparse_spsv_stage_auto is selected, rocSPARSE will automatically detect which stage is required based on the following indicators: If temp_buffer is equal to nullptr, the required buffer size will be returned. If buffer_size is equal to nullptr, analysis will be performed. Otherwise, the SpSV preprocess and the SpSV algorithm will be executed.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Currently, only trans == rocsparse_operation_none and trans == rocsparse_operation_transpose is supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans[in] matrix operation type.

  • alpha[in] scalar \(\alpha\).

  • mat[in] matrix descriptor.

  • x[in] vector descriptor.

  • y[inout] vector descriptor.

  • compute_type[in] floating point precision for the SpSV computation.

  • alg[in] SpSV algorithm for the SpSV computation.

  • stage[in] SpSV stage for the SpSV computation.

  • buffer_size[out] number of bytes of the temporary storage buffer.

  • temp_buffer[in] temporary storage buffer allocated by the user. When a nullptr is passed, the required allocation size (in bytes) is written to buffer_size and function returns without performing the SpSV operation.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointeralpha, mat, x, y, descr or buffer_size pointer is invalid.

  • rocsparse_status_not_implementedtrans, compute_type, stage or alg is currently not supported.

rocsparse_spsm()#

rocsparse_status rocsparse_spsm(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, const void *alpha, const rocsparse_spmat_descr matA, const rocsparse_dnmat_descr matB, const rocsparse_dnmat_descr matC, rocsparse_datatype compute_type, rocsparse_spsm_alg alg, rocsparse_spsm_stage stage, size_t *buffer_size, void *temp_buffer)#

Sparse triangular system solve.

rocsparse_spsm_solve solves a sparse triangular linear system of a sparse \(m \times m\) matrix, defined in CSR or COO storage format, a dense solution matrix \(C\) and the right-hand side \(B\) that is multiplied by \(\alpha\), such that

\[ op(A) \cdot C = \alpha \cdot op(B), \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans == rocsparse_operation_none} \\ A^T, & \text{if trans == rocsparse_operation_transpose} \\ A^H, & \text{if trans == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]
and
\[\begin{split} op(B) = \left\{ \begin{array}{ll} B, & \text{if trans_B == rocsparse_operation_none} \\ B^T, & \text{if trans_B == rocsparse_operation_transpose} \\ B^H, & \text{if trans_B == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

Note

SpSM requires three stages to complete. The first stage rocsparse_spsm_stage_buffer_size will return the size of the temporary storage buffer that is required for subsequent calls. The second stage rocsparse_spsm_stage_preprocess will preprocess data that would be saved in the temporary storage buffer. In the final stage rocsparse_spsm_stage_compute, the actual computation is performed.

Note

If rocsparse_spsm_stage_auto is selected, rocSPARSE will automatically detect which stage is required based on the following indicators: If temp_buffer is equal to nullptr, the required buffer size will be returned. If buffer_size is equal to nullptr, analysis will be performed. Otherwise, the SpSM preprocess and the SpSM algorithm will be executed.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Currently, only trans_A == rocsparse_operation_none and trans_A == rocsparse_operation_transpose is supported. Currently, only trans_B == rocsparse_operation_none and trans_B == rocsparse_operation_transpose is supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans_A[in] matrix operation type for the sparse matrix A.

  • trans_B[in] matrix operation type for the dense matrix B.

  • alpha[in] scalar \(\alpha\).

  • matA[in] sparse matrix descriptor.

  • matB[in] dense matrix descriptor.

  • matC[inout] dense matrix descriptor.

  • compute_type[in] floating point precision for the SpSM computation.

  • alg[in] SpSM algorithm for the SpSM computation.

  • stage[in] SpSM stage for the SpSM computation.

  • buffer_size[out] number of bytes of the temporary storage buffer.

  • temp_buffer[in] temporary storage buffer allocated by the user. When a nullptr is passed, the required allocation size (in bytes) is written to buffer_size and function returns without performing the SpSM operation.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointeralpha, matA, matB, matC, descr or buffer_size pointer is invalid.

  • rocsparse_status_not_implementedtrans_A, trans_B, compute_type, stage or alg is currently not supported.

rocsparse_spmm()#

rocsparse_status rocsparse_spmm(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, const void *alpha, const rocsparse_spmat_descr mat_A, const rocsparse_dnmat_descr mat_B, const void *beta, const rocsparse_dnmat_descr mat_C, rocsparse_datatype compute_type, rocsparse_spmm_alg alg, rocsparse_spmm_stage stage, size_t *buffer_size, void *temp_buffer)#

Sparse matrix dense matrix multiplication, extension routine.

rocsparse_spmm (or rocsparse_spmm_ex ) multiplies the scalar \(\alpha\) with a sparse \(m \times k\) matrix \(A\), defined in CSR or COO or Blocked ELL storage format, and the dense \(k \times n\) matrix \(B\) and adds the result to the dense \(m \times n\) matrix \(C\) that is multiplied by the scalar \(\beta\), such that

\[ C := \alpha \cdot op(A) \cdot op(B) + \beta \cdot C, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans_A == rocsparse_operation_none} \\ A^T, & \text{if trans_A == rocsparse_operation_transpose} \\ A^H, & \text{if trans_A == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]
and
\[\begin{split} op(B) = \left\{ \begin{array}{ll} B, & \text{if trans_B == rocsparse_operation_none} \\ B^T, & \text{if trans_B == rocsparse_operation_transpose} \\ B^H, & \text{if trans_B == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

Example

This example performs sparse matrix-dense matrix multiplication, C = alpha * A * B + beta * C

//     1 4 0 0 0 0
// A = 0 2 3 0 0 0
//     5 0 0 7 8 0
//     0 0 9 0 6 0

//     1 4 2
//     1 2 3
// B = 5 4 0
//     3 1 9
//     1 2 2
//     0 3 0

//     1 1 5
// C = 1 2 1
//     1 3 1
//     6 2 4

rocsparse_int m   = 4;
rocsparse_int k   = 6;
rocsparse_int n   = 3;

csr_row_ptr[m + 1] = {0, 1, 3};                                              // device memory
csr_col_ind[nnz]   = {0, 0, 1};                                              // device memory
csr_val[nnz]       = {1, 0, 4, 2, 0, 3, 5, 0, 0, 0, 0, 9, 7, 0, 8, 6, 0, 0}; // device memory

B[k * n]       = {1, 1, 5, 3, 1, 0, 4, 2, 4, 1, 2, 3, 2, 3, 0, 9, 2, 0};     // device memory
C[m * n]       = {1, 1, 1, 6, 1, 2, 3, 2, 5, 1, 1, 4};                       // device memory

rocsparse_int nnz = csr_row_ptr[m] - csr_row_ptr[0];

float alpha = 1.0f;
float beta  = 0.0f;

// Create CSR arrays on device
rocsparse_int* csr_row_ptr;
rocsparse_int* csr_col_ind;
float* csr_val;
float* B;
float* C;
hipMalloc((void**)&csr_row_ptr, sizeof(rocsparse_int) * (m + 1));
hipMalloc((void**)&csr_col_ind, sizeof(rocsparse_int) * nnz);
hipMalloc((void**)&csr_val, sizeof(float) * nnz);
hipMalloc((void**)&B, sizeof(float) * k * n);
hipMalloc((void**)&C, sizeof(float) * m * n);

// Create rocsparse handle
rocsparse_local_handle handle;

// Types
rocsparse_indextype itype = rocsparse_indextype_i32;
rocsparse_indextype jtype = rocsparse_indextype_i32;
rocsparse_datatype  ttype = rocsparse_datatype_f32_r;

// Create descriptors
rocsparse_spmat_descr mat_A;
rocsparse_dnmat_descr mat_B;
rocsparse_dnmat_descr mat_C;

rocsparse_create_csr_descr(&mat_A, m, k, nnz, csr_row_ptr, csr_col_ind, csr_val, itype, jtype, rocsparse_index_base_zero, ttype);
rocsparse_create_dnmat_descr(&mat_B, k, n, k, B, ttype, rocsparse_order_column);
rocsparse_create_dnmat_descr(&mat_C, m, n, m, C, ttype, rocsparse_order_column);

// Query SpMM buffer
size_t buffer_size;
rocsparse_spmm(handle,
               rocsparse_operation_none,
               rocsparse_operation_none,
               &alpha,
               mat_A,
               mat_B,
               &beta,
               mat_C,
               ttype,
               rocsparse_spmm_alg_default,
               rocsparse_spmm_stage_buffer_size,
               &buffer_size,
               nullptr));

// Allocate buffer
void* buffer;
hipMalloc(&buffer, buffer_size);

rocsparse_spmm(handle,
               rocsparse_operation_none,
               rocsparse_operation_none,
               &alpha,
               mat_A,
               mat_B,
               &beta,
               mat_C,
               ttype,
               rocsparse_spmm_alg_default,
               rocsparse_spmm_stage_preprocess,
               &buffer_size,
               buffer));

// Pointer mode host
rocsparse_spmm(handle,
               rocsparse_operation_none,
               rocsparse_operation_none,
               &alpha,
               mat_A,
               mat_B,
               &beta,
               mat_C,
               ttype,
               rocsparse_spmm_alg_default,
               rocsparse_spmm_stage_compute,
               &buffer_size,
               buffer));

// Clear up on device
hipFree(csr_row_ptr);
hipFree(csr_col_ind);
hipFree(csr_val);
hipFree(B);
hipFree(C);
hipFree(temp_buffer);

rocsparse_destroy_spmat_descr(mat_A);
rocsparse_destroy_dnmat_descr(mat_B);
rocsparse_destroy_dnmat_descr(mat_C);

Example

SpMM also supports batched computation for CSR and COO matrices. There are three supported batch modes: C_i = A * B_i C_i = A_i * B C_i = A_i * B_i The batch mode is determined by the batch count and stride passed for each matrix. For example to use the first batch mode (C_i = A * B_i) with 100 batches for non-transposed A, B, and C, one passes: batch_count_A = 1 batch_count_B = 100 batch_count_C = 100 offsets_batch_stride_A = 0 columns_values_batch_stride_A = 0 batch_stride_B = k * n batch_stride_C = m * n To use the second batch mode (C_i = A_i * B) one could use: batch_count_A = 100 batch_count_B = 1 batch_count_C = 100 offsets_batch_stride_A = m + 1 columns_values_batch_stride_A = nnz batch_stride_B = 0 batch_stride_C = m * n And to use the third batch mode (C_i = A_i * B_i) one could use: batch_count_A = 100 batch_count_B = 100 batch_count_C = 100 offsets_batch_stride_A = m + 1 columns_values_batch_stride_A = nnz batch_stride_B = k * n batch_stride_C = m * n An example of the first batch mode (C_i = A * B_i) is provided below.

//     1 4 0 0 0 0
// A = 0 2 3 0 0 0
//     5 0 0 7 8 0
//     0 0 9 0 6 0

rocsparse_int m   = 4;
rocsparse_int k   = 6;
rocsparse_int n   = 3;

csr_row_ptr[m + 1] = {0, 1, 3};                                              // device memory
csr_col_ind[nnz]   = {0, 0, 1};                                              // device memory
csr_val[nnz]       = {1, 0, 4, 2, 0, 3, 5, 0, 0, 0, 0, 9, 7, 0, 8, 6, 0, 0}; // device memory

B[batch_count_B * k * n]       = {...};     // device memory
C[batch_count_C * m * n]       = {...};     // device memory

rocsparse_int nnz = csr_row_ptr[m] - csr_row_ptr[0];

rocsparse_int batch_count_A = 1;
rocsparse_int batch_count_B = 100;
rocsparse_int batch_count_C = 100;

rocsparse_int offsets_batch_stride_A        = 0;
rocsparse_int columns_values_batch_stride_A = 0;
rocsparse_int batch_stride_B                = k * n;
rocsparse_int batch_stride_C                = m * n;

float alpha = 1.0f;
float beta  = 0.0f;

// Create CSR arrays on device
rocsparse_int* csr_row_ptr;
rocsparse_int* csr_col_ind;
float* csr_val;
float* B;
float* C;
hipMalloc((void**)&csr_row_ptr, sizeof(rocsparse_int) * (m + 1));
hipMalloc((void**)&csr_col_ind, sizeof(rocsparse_int) * nnz);
hipMalloc((void**)&csr_val, sizeof(float) * nnz);
hipMalloc((void**)&B, sizeof(float) * batch_count_B * k * n);
hipMalloc((void**)&C, sizeof(float) * batch_count_C * m * n);

// Create rocsparse handle
rocsparse_local_handle handle;

// Types
rocsparse_indextype itype = rocsparse_indextype_i32;
rocsparse_indextype jtype = rocsparse_indextype_i32;
rocsparse_datatype  ttype = rocsparse_datatype_f32_r;

// Create descriptors
rocsparse_spmat_descr mat_A;
rocsparse_dnmat_descr mat_B;
rocsparse_dnmat_descr mat_C;

rocsparse_create_csr_descr(&mat_A, m, k, nnz, csr_row_ptr, csr_col_ind, csr_val, itype, jtype, rocsparse_index_base_zero, ttype);
rocsparse_create_dnmat_descr(&mat_B, k, n, k, B, ttype, rocsparse_order_column);
rocsparse_create_dnmat_descr(&mat_C, m, n, m, C, ttype, rocsparse_order_column);

rocsparse_csr_set_strided_batch(mat_A, batch_count_A, offsets_batch_stride_A, columns_values_batch_stride_A);
rocsparse_dnmat_set_strided_batch(B, batch_count_B, batch_stride_B);
rocsparse_dnmat_set_strided_batch(C, batch_count_C, batch_stride_C);

// Query SpMM buffer
size_t buffer_size;
rocsparse_spmm(handle,
               rocsparse_operation_none,
               rocsparse_operation_none,
               &alpha,
               mat_A,
               mat_B,
               &beta,
               mat_C,
               ttype,
               rocsparse_spmm_alg_default,
               rocsparse_spmm_stage_buffer_size,
               &buffer_size,
               nullptr));

// Allocate buffer
void* buffer;
hipMalloc(&buffer, buffer_size);

rocsparse_spmm(handle,
               rocsparse_operation_none,
               rocsparse_operation_none,
               &alpha,
               mat_A,
               mat_B,
               &beta,
               mat_C,
               ttype,
               rocsparse_spmm_alg_default,
               rocsparse_spmm_stage_preprocess,
               &buffer_size,
               buffer));

// Pointer mode host
rocsparse_spmm(handle,
               rocsparse_operation_none,
               rocsparse_operation_none,
               &alpha,
               mat_A,
               mat_B,
               &beta,
               mat_C,
               ttype,
               rocsparse_spmm_alg_default,
               rocsparse_spmm_stage_compute,
               &buffer_size,
               buffer));

// Clear up on device
hipFree(csr_row_ptr);
hipFree(csr_col_ind);
hipFree(csr_val);
hipFree(B);
hipFree(C);
hipFree(temp_buffer);

rocsparse_destroy_spmat_descr(mat_A);
rocsparse_destroy_dnmat_descr(mat_B);
rocsparse_destroy_dnmat_descr(mat_C);

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Currently, only trans_A == rocsparse_operation_none is supported for COO and Blocked ELL formats.

Note

Currently, only CSR, COO and Blocked ELL sparse formats are supported.

Note

Different algorithms are available which can provide better performance for different matrices. Currently, the available algorithms are rocsparse_spmm_alg_csr, rocsparse_spmm_alg_csr_row_split or rocsparse_spmm_alg_csr_merge for CSR matrices, rocsparse_spmm_alg_bell for Blocked ELL matrices and rocsparse_spmm_alg_coo_segmented or rocsparse_spmm_alg_coo_atomic for COO matrices. Additionally, one can specify the algorithm to be rocsparse_spmm_alg_default. In the case of CSR matrices this will set the algorithm to be rocsparse_spmm_alg_csr, in the case of Blocked ELL matrices this will set the algorithm to be rocsparse_spmm_alg_bell and for COO matrices it will set the algorithm to be rocsparse_spmm_alg_coo_atomic. When A is transposed, rocsparse_spmm will revert to using rocsparse_spmm_alg_csr for CSR format and rocsparse_spmm_alg_coo_atomic for COO format regardless of algorithm selected.

Note

This function writes the required allocation size (in bytes) to buffer_size and returns without performing the SpMM operation, when a nullptr is passed for temp_buffer.

Note

SpMM requires three stages to complete. The first stage rocsparse_spmm_stage_buffer_size will return the size of the temporary storage buffer that is required for subsequent calls to rocsparse_spmm (or rocsparse_spmm_ex). The second stage rocsparse_spmm_stage_preprocess will preprocess data that would be saved in the temporary storage buffer. In the final stage rocsparse_spmm_stage_compute, the actual computation is performed.

Note

If rocsparse_spmm_stage_auto is selected, rocSPARSE will automatically detect which stage is required based on the following indicators: If temp_buffer is equal to nullptr, the required buffer size will be returned. Else, the SpMM preprocess and the SpMM algorithm will be executed.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans_A[in] matrix operation type.

  • trans_B[in] matrix operation type.

  • alpha[in] scalar \(\alpha\).

  • mat_A[in] matrix descriptor.

  • mat_B[in] matrix descriptor.

  • beta[in] scalar \(\beta\).

  • mat_C[in] matrix descriptor.

  • compute_type[in] floating point precision for the SpMM computation.

  • alg[in] SpMM algorithm for the SpMM computation.

  • stage[in] SpMM stage for the SpMM computation.

  • buffer_size[out] number of bytes of the temporary storage buffer. buffer_size is set when temp_buffer is nullptr.

  • temp_buffer[in] temporary storage buffer allocated by the user. When a nullptr is passed, the required allocation size (in bytes) is written to buffer_size and function returns without performing the SpMM operation.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointeralpha, mat_A, mat_B, mat_C, beta, or buffer_size pointer is invalid.

  • rocsparse_status_not_implementedtrans_A, trans_B, compute_type or alg is currently not supported.

rocsparse_spmm_ex()#

rocsparse_status rocsparse_spmm_ex(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, const void *alpha, const rocsparse_spmat_descr mat_A, const rocsparse_dnmat_descr mat_B, const void *beta, const rocsparse_dnmat_descr mat_C, rocsparse_datatype compute_type, rocsparse_spmm_alg alg, rocsparse_spmm_stage stage, size_t *buffer_size, void *temp_buffer)#

Sparse matrix dense matrix multiplication, extension routine.

rocsparse_spmm (or rocsparse_spmm_ex ) multiplies the scalar \(\alpha\) with a sparse \(m \times k\) matrix \(A\), defined in CSR or COO or Blocked ELL storage format, and the dense \(k \times n\) matrix \(B\) and adds the result to the dense \(m \times n\) matrix \(C\) that is multiplied by the scalar \(\beta\), such that

\[ C := \alpha \cdot op(A) \cdot op(B) + \beta \cdot C, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans_A == rocsparse_operation_none} \\ A^T, & \text{if trans_A == rocsparse_operation_transpose} \\ A^H, & \text{if trans_A == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]
and
\[\begin{split} op(B) = \left\{ \begin{array}{ll} B, & \text{if trans_B == rocsparse_operation_none} \\ B^T, & \text{if trans_B == rocsparse_operation_transpose} \\ B^H, & \text{if trans_B == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

Example

This example performs sparse matrix-dense matrix multiplication, C = alpha * A * B + beta * C

//     1 4 0 0 0 0
// A = 0 2 3 0 0 0
//     5 0 0 7 8 0
//     0 0 9 0 6 0

//     1 4 2
//     1 2 3
// B = 5 4 0
//     3 1 9
//     1 2 2
//     0 3 0

//     1 1 5
// C = 1 2 1
//     1 3 1
//     6 2 4

rocsparse_int m   = 4;
rocsparse_int k   = 6;
rocsparse_int n   = 3;

csr_row_ptr[m + 1] = {0, 1, 3};                                              // device memory
csr_col_ind[nnz]   = {0, 0, 1};                                              // device memory
csr_val[nnz]       = {1, 0, 4, 2, 0, 3, 5, 0, 0, 0, 0, 9, 7, 0, 8, 6, 0, 0}; // device memory

B[k * n]       = {1, 1, 5, 3, 1, 0, 4, 2, 4, 1, 2, 3, 2, 3, 0, 9, 2, 0};     // device memory
C[m * n]       = {1, 1, 1, 6, 1, 2, 3, 2, 5, 1, 1, 4};                       // device memory

rocsparse_int nnz = csr_row_ptr[m] - csr_row_ptr[0];

float alpha = 1.0f;
float beta  = 0.0f;

// Create CSR arrays on device
rocsparse_int* csr_row_ptr;
rocsparse_int* csr_col_ind;
float* csr_val;
float* B;
float* C;
hipMalloc((void**)&csr_row_ptr, sizeof(rocsparse_int) * (m + 1));
hipMalloc((void**)&csr_col_ind, sizeof(rocsparse_int) * nnz);
hipMalloc((void**)&csr_val, sizeof(float) * nnz);
hipMalloc((void**)&B, sizeof(float) * k * n);
hipMalloc((void**)&C, sizeof(float) * m * n);

// Create rocsparse handle
rocsparse_local_handle handle;

// Types
rocsparse_indextype itype = rocsparse_indextype_i32;
rocsparse_indextype jtype = rocsparse_indextype_i32;
rocsparse_datatype  ttype = rocsparse_datatype_f32_r;

// Create descriptors
rocsparse_spmat_descr mat_A;
rocsparse_dnmat_descr mat_B;
rocsparse_dnmat_descr mat_C;

rocsparse_create_csr_descr(&mat_A, m, k, nnz, csr_row_ptr, csr_col_ind, csr_val, itype, jtype, rocsparse_index_base_zero, ttype);
rocsparse_create_dnmat_descr(&mat_B, k, n, k, B, ttype, rocsparse_order_column);
rocsparse_create_dnmat_descr(&mat_C, m, n, m, C, ttype, rocsparse_order_column);

// Query SpMM buffer
size_t buffer_size;
rocsparse_spmm(handle,
               rocsparse_operation_none,
               rocsparse_operation_none,
               &alpha,
               mat_A,
               mat_B,
               &beta,
               mat_C,
               ttype,
               rocsparse_spmm_alg_default,
               rocsparse_spmm_stage_buffer_size,
               &buffer_size,
               nullptr));

// Allocate buffer
void* buffer;
hipMalloc(&buffer, buffer_size);

rocsparse_spmm(handle,
               rocsparse_operation_none,
               rocsparse_operation_none,
               &alpha,
               mat_A,
               mat_B,
               &beta,
               mat_C,
               ttype,
               rocsparse_spmm_alg_default,
               rocsparse_spmm_stage_preprocess,
               &buffer_size,
               buffer));

// Pointer mode host
rocsparse_spmm(handle,
               rocsparse_operation_none,
               rocsparse_operation_none,
               &alpha,
               mat_A,
               mat_B,
               &beta,
               mat_C,
               ttype,
               rocsparse_spmm_alg_default,
               rocsparse_spmm_stage_compute,
               &buffer_size,
               buffer));

// Clear up on device
hipFree(csr_row_ptr);
hipFree(csr_col_ind);
hipFree(csr_val);
hipFree(B);
hipFree(C);
hipFree(temp_buffer);

rocsparse_destroy_spmat_descr(mat_A);
rocsparse_destroy_dnmat_descr(mat_B);
rocsparse_destroy_dnmat_descr(mat_C);

Example

SpMM also supports batched computation for CSR and COO matrices. There are three supported batch modes: C_i = A * B_i C_i = A_i * B C_i = A_i * B_i The batch mode is determined by the batch count and stride passed for each matrix. For example to use the first batch mode (C_i = A * B_i) with 100 batches for non-transposed A, B, and C, one passes: batch_count_A = 1 batch_count_B = 100 batch_count_C = 100 offsets_batch_stride_A = 0 columns_values_batch_stride_A = 0 batch_stride_B = k * n batch_stride_C = m * n To use the second batch mode (C_i = A_i * B) one could use: batch_count_A = 100 batch_count_B = 1 batch_count_C = 100 offsets_batch_stride_A = m + 1 columns_values_batch_stride_A = nnz batch_stride_B = 0 batch_stride_C = m * n And to use the third batch mode (C_i = A_i * B_i) one could use: batch_count_A = 100 batch_count_B = 100 batch_count_C = 100 offsets_batch_stride_A = m + 1 columns_values_batch_stride_A = nnz batch_stride_B = k * n batch_stride_C = m * n An example of the first batch mode (C_i = A * B_i) is provided below.

//     1 4 0 0 0 0
// A = 0 2 3 0 0 0
//     5 0 0 7 8 0
//     0 0 9 0 6 0

rocsparse_int m   = 4;
rocsparse_int k   = 6;
rocsparse_int n   = 3;

csr_row_ptr[m + 1] = {0, 1, 3};                                              // device memory
csr_col_ind[nnz]   = {0, 0, 1};                                              // device memory
csr_val[nnz]       = {1, 0, 4, 2, 0, 3, 5, 0, 0, 0, 0, 9, 7, 0, 8, 6, 0, 0}; // device memory

B[batch_count_B * k * n]       = {...};     // device memory
C[batch_count_C * m * n]       = {...};     // device memory

rocsparse_int nnz = csr_row_ptr[m] - csr_row_ptr[0];

rocsparse_int batch_count_A = 1;
rocsparse_int batch_count_B = 100;
rocsparse_int batch_count_C = 100;

rocsparse_int offsets_batch_stride_A        = 0;
rocsparse_int columns_values_batch_stride_A = 0;
rocsparse_int batch_stride_B                = k * n;
rocsparse_int batch_stride_C                = m * n;

float alpha = 1.0f;
float beta  = 0.0f;

// Create CSR arrays on device
rocsparse_int* csr_row_ptr;
rocsparse_int* csr_col_ind;
float* csr_val;
float* B;
float* C;
hipMalloc((void**)&csr_row_ptr, sizeof(rocsparse_int) * (m + 1));
hipMalloc((void**)&csr_col_ind, sizeof(rocsparse_int) * nnz);
hipMalloc((void**)&csr_val, sizeof(float) * nnz);
hipMalloc((void**)&B, sizeof(float) * batch_count_B * k * n);
hipMalloc((void**)&C, sizeof(float) * batch_count_C * m * n);

// Create rocsparse handle
rocsparse_local_handle handle;

// Types
rocsparse_indextype itype = rocsparse_indextype_i32;
rocsparse_indextype jtype = rocsparse_indextype_i32;
rocsparse_datatype  ttype = rocsparse_datatype_f32_r;

// Create descriptors
rocsparse_spmat_descr mat_A;
rocsparse_dnmat_descr mat_B;
rocsparse_dnmat_descr mat_C;

rocsparse_create_csr_descr(&mat_A, m, k, nnz, csr_row_ptr, csr_col_ind, csr_val, itype, jtype, rocsparse_index_base_zero, ttype);
rocsparse_create_dnmat_descr(&mat_B, k, n, k, B, ttype, rocsparse_order_column);
rocsparse_create_dnmat_descr(&mat_C, m, n, m, C, ttype, rocsparse_order_column);

rocsparse_csr_set_strided_batch(mat_A, batch_count_A, offsets_batch_stride_A, columns_values_batch_stride_A);
rocsparse_dnmat_set_strided_batch(B, batch_count_B, batch_stride_B);
rocsparse_dnmat_set_strided_batch(C, batch_count_C, batch_stride_C);

// Query SpMM buffer
size_t buffer_size;
rocsparse_spmm(handle,
               rocsparse_operation_none,
               rocsparse_operation_none,
               &alpha,
               mat_A,
               mat_B,
               &beta,
               mat_C,
               ttype,
               rocsparse_spmm_alg_default,
               rocsparse_spmm_stage_buffer_size,
               &buffer_size,
               nullptr));

// Allocate buffer
void* buffer;
hipMalloc(&buffer, buffer_size);

rocsparse_spmm(handle,
               rocsparse_operation_none,
               rocsparse_operation_none,
               &alpha,
               mat_A,
               mat_B,
               &beta,
               mat_C,
               ttype,
               rocsparse_spmm_alg_default,
               rocsparse_spmm_stage_preprocess,
               &buffer_size,
               buffer));

// Pointer mode host
rocsparse_spmm(handle,
               rocsparse_operation_none,
               rocsparse_operation_none,
               &alpha,
               mat_A,
               mat_B,
               &beta,
               mat_C,
               ttype,
               rocsparse_spmm_alg_default,
               rocsparse_spmm_stage_compute,
               &buffer_size,
               buffer));

// Clear up on device
hipFree(csr_row_ptr);
hipFree(csr_col_ind);
hipFree(csr_val);
hipFree(B);
hipFree(C);
hipFree(temp_buffer);

rocsparse_destroy_spmat_descr(mat_A);
rocsparse_destroy_dnmat_descr(mat_B);
rocsparse_destroy_dnmat_descr(mat_C);

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Currently, only trans_A == rocsparse_operation_none is supported for COO and Blocked ELL formats.

Note

Currently, only CSR, COO and Blocked ELL sparse formats are supported.

Note

Different algorithms are available which can provide better performance for different matrices. Currently, the available algorithms are rocsparse_spmm_alg_csr, rocsparse_spmm_alg_csr_row_split or rocsparse_spmm_alg_csr_merge for CSR matrices, rocsparse_spmm_alg_bell for Blocked ELL matrices and rocsparse_spmm_alg_coo_segmented or rocsparse_spmm_alg_coo_atomic for COO matrices. Additionally, one can specify the algorithm to be rocsparse_spmm_alg_default. In the case of CSR matrices this will set the algorithm to be rocsparse_spmm_alg_csr, in the case of Blocked ELL matrices this will set the algorithm to be rocsparse_spmm_alg_bell and for COO matrices it will set the algorithm to be rocsparse_spmm_alg_coo_atomic. When A is transposed, rocsparse_spmm will revert to using rocsparse_spmm_alg_csr for CSR format and rocsparse_spmm_alg_coo_atomic for COO format regardless of algorithm selected.

Note

This function writes the required allocation size (in bytes) to buffer_size and returns without performing the SpMM operation, when a nullptr is passed for temp_buffer.

Note

SpMM requires three stages to complete. The first stage rocsparse_spmm_stage_buffer_size will return the size of the temporary storage buffer that is required for subsequent calls to rocsparse_spmm (or rocsparse_spmm_ex). The second stage rocsparse_spmm_stage_preprocess will preprocess data that would be saved in the temporary storage buffer. In the final stage rocsparse_spmm_stage_compute, the actual computation is performed.

Note

If rocsparse_spmm_stage_auto is selected, rocSPARSE will automatically detect which stage is required based on the following indicators: If temp_buffer is equal to nullptr, the required buffer size will be returned. Else, the SpMM preprocess and the SpMM algorithm will be executed.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans_A[in] matrix operation type.

  • trans_B[in] matrix operation type.

  • alpha[in] scalar \(\alpha\).

  • mat_A[in] matrix descriptor.

  • mat_B[in] matrix descriptor.

  • beta[in] scalar \(\beta\).

  • mat_C[in] matrix descriptor.

  • compute_type[in] floating point precision for the SpMM computation.

  • alg[in] SpMM algorithm for the SpMM computation.

  • stage[in] SpMM stage for the SpMM computation.

  • buffer_size[out] number of bytes of the temporary storage buffer. buffer_size is set when temp_buffer is nullptr.

  • temp_buffer[in] temporary storage buffer allocated by the user. When a nullptr is passed, the required allocation size (in bytes) is written to buffer_size and function returns without performing the SpMM operation.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointeralpha, mat_A, mat_B, mat_C, beta, or buffer_size pointer is invalid.

  • rocsparse_status_not_implementedtrans_A, trans_B, compute_type or alg is currently not supported.

rocsparse_spgemm()#

rocsparse_status rocsparse_spgemm(rocsparse_handle handle, rocsparse_operation trans_A, rocsparse_operation trans_B, const void *alpha, const rocsparse_spmat_descr A, const rocsparse_spmat_descr B, const void *beta, const rocsparse_spmat_descr D, rocsparse_spmat_descr C, rocsparse_datatype compute_type, rocsparse_spgemm_alg alg, rocsparse_spgemm_stage stage, size_t *buffer_size, void *temp_buffer)#

Sparse matrix sparse matrix multiplication.

rocsparse_spgemm multiplies the scalar \(\alpha\) with the sparse \(m \times k\) matrix \(A\) and the sparse \(k \times n\) matrix \(B\) and adds the result to the sparse \(m \times n\) matrix \(D\) that is multiplied by \(\beta\). The final result is stored in the sparse \(m \times n\) matrix \(C\), such that

\[ C := \alpha \cdot op(A) \cdot op(B) + \beta \cdot D, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if trans_A == rocsparse_operation_none} \\ A^T, & \text{if trans_A == rocsparse_operation_transpose} \\ A^H, & \text{if trans_A == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]
and
\[\begin{split} op(B) = \left\{ \begin{array}{ll} B, & \text{if trans_B == rocsparse_operation_none} \\ B^T, & \text{if trans_B == rocsparse_operation_transpose} \\ B^H, & \text{if trans_B == rocsparse_operation_conjugate_transpose} \end{array} \right. \end{split}\]

Note

SpGEMM requires three stages to complete. The first stage rocsparse_spgemm_stage_buffer_size will return the size of the temporary storage buffer that is required for subsequent calls to rocsparse_spgemm. The second stage rocsparse_spgemm_stage_nnz will determine the number of non-zero elements of the resulting \(C\) matrix. If the sparsity pattern of \(C\) is already known, this stage can be skipped. In the final stage rocsparse_spgemm_stage_compute, the actual computation is performed.

Note

If rocsparse_spgemm_stage_auto is selected, rocSPARSE will automatically detect which stage is required based on the following indicators: If temp_buffer is equal to nullptr, the required buffer size will be returned. Else, if the number of non-zeros of \(C\) is zero, the number of non-zero entries will be computed. Else, the SpGEMM algorithm will be executed.

Note

If \(\alpha == 0\), then \(C = \beta \cdot D\) will be computed.

Note

If \(\beta == 0\), then \(C = \alpha \cdot op(A) \cdot op(B)\) will be computed.

Note

If rocsparse_spgemm_stage_symbolic is selected then the symbolic computation is performed only.

Note

If rocsparse_spgemm_stage_numeric is selected then the numeric computation is performed only.

Note

\(\alpha == beta == 0\) is invalid.

Note

It is allowed to pass the same sparse matrix for \(C\) and \(D\), if both matrices have the same sparsity pattern.

Note

Currently, only trans_A == rocsparse_operation_none is supported.

Note

Currently, only trans_B == rocsparse_operation_none is supported.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Note

Please note, that for rare matrix products with more than 4096 non-zero entries per row, additional temporary storage buffer is allocated by the algorithm.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • trans_A[in] sparse matrix \(A\) operation type.

  • trans_B[in] sparse matrix \(B\) operation type.

  • alpha[in] scalar \(\alpha\).

  • A[in] sparse matrix \(A\) descriptor.

  • B[in] sparse matrix \(B\) descriptor.

  • beta[in] scalar \(\beta\).

  • D[in] sparse matrix \(D\) descriptor.

  • C[out] sparse matrix \(C\) descriptor.

  • compute_type[in] floating point precision for the SpGEMM computation.

  • alg[in] SpGEMM algorithm for the SpGEMM computation.

  • stage[in] SpGEMM stage for the SpGEMM computation.

  • buffer_size[out] number of bytes of the temporary storage buffer. buffer_size is set when temp_buffer is nullptr.

  • temp_buffer[in] temporary storage buffer allocated by the user. When a nullptr is passed, the required allocation size (in bytes) is written to buffer_size and function returns without performing the SpGEMM operation.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointeralpha and beta are invalid, A, B, D, C or buffer_size pointer is invalid.

  • rocsparse_status_memory_error – additional buffer for long rows could not be allocated.

  • rocsparse_status_not_implementedtrans_A != rocsparse_operation_none or trans_B != rocsparse_operation_none.

rocsparse_sddmm_buffer_size()#

rocsparse_status rocsparse_sddmm_buffer_size(rocsparse_handle handle, rocsparse_operation opA, rocsparse_operation opB, const void *alpha, const rocsparse_dnmat_descr A, const rocsparse_dnmat_descr B, const void *beta, rocsparse_spmat_descr C, rocsparse_datatype compute_type, rocsparse_sddmm_alg alg, size_t *buffer_size)#

Calculate the size in bytes of the required buffer for the use of rocsparse_sddmm and rocsparse_sddmm_preprocess.

rocsparse_sddmm_buffer_size returns the size of the required buffer to execute the SDDMM operation from a given configuration.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • opA[in] dense matrix \(A\) operation type.

  • opB[in] dense matrix \(B\) operation type.

  • alpha[in] scalar \(\alpha\).

  • A[in] dense matrix \(A\) descriptor.

  • B[in] dense matrix \(B\) descriptor.

  • beta[in] scalar \(\beta\).

  • C[inout] sparse matrix \(C\) descriptor.

  • compute_type[in] floating point precision for the SDDMM computation.

  • alg[in] specification of the algorithm to use.

  • buffer_size[out] number of bytes of the temporary storage buffer.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_value – the value of trans_A or trans_B is incorrect.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointeralpha and beta are invalid, A, B, D, C or buffer_size pointer is invalid.

  • rocsparse_status_not_implementedopA == rocsparse_operation_conjugate_transpose or opB == rocsparse_operation_conjugate_transpose.

rocsparse_sddmm_preprocess()#

rocsparse_status rocsparse_sddmm_preprocess(rocsparse_handle handle, rocsparse_operation opA, rocsparse_operation opB, const void *alpha, const rocsparse_dnmat_descr A, const rocsparse_dnmat_descr B, const void *beta, rocsparse_spmat_descr C, rocsparse_datatype compute_type, rocsparse_sddmm_alg alg, void *temp_buffer)#

Preprocess data before the use of rocsparse_sddmm.

rocsparse_sddmm_preprocess executes a part of the algorithm that can be calculated once in the context of multiple calls of the rocsparse_sddmm with the same sparsity pattern.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • opA[in] dense matrix \(A\) operation type.

  • opB[in] dense matrix \(B\) operation type.

  • alpha[in] scalar \(\alpha\).

  • A[in] dense matrix \(A\) descriptor.

  • B[in] dense matrix \(B\) descriptor.

  • beta[in] scalar \(\beta\).

  • C[inout] sparse matrix \(C\) descriptor.

  • compute_type[in] floating point precision for the SDDMM computation.

  • alg[in] specification of the algorithm to use.

  • temp_buffer[in] temporary storage buffer allocated by the user. The size must be greater or equal to the size obtained with rocsparse_sddmm_buffer_size.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_value – the value of trans_A or trans_B is incorrect.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointeralpha and beta are invalid, A, B, D, C or temp_buffer pointer is invalid.

  • rocsparse_status_not_implementedopA == rocsparse_operation_conjugate_transpose or opB == rocsparse_operation_conjugate_transpose.

rocsparse_sddmm()#

rocsparse_status rocsparse_sddmm(rocsparse_handle handle, rocsparse_operation opA, rocsparse_operation opB, const void *alpha, const rocsparse_dnmat_descr A, const rocsparse_dnmat_descr B, const void *beta, rocsparse_spmat_descr C, rocsparse_datatype compute_type, rocsparse_sddmm_alg alg, void *temp_buffer)#

Sampled Dense-Dense Matrix Multiplication.

rocsparse_sddmm multiplies the scalar \(\alpha\) with the dense \(m \times k\) matrix \(A\), the dense \(k \times n\) matrix \(B\), filtered by the sparsity pattern of the \(m \times n\) sparse matrix \(C\) and adds the result to \(C\) scaled by \(\beta\). The final result is stored in the sparse \(m \times n\) matrix \(C\), such that

\[ C := \alpha ( op(A) \cdot op(B) ) \cdot spy(C) + \beta C, \]
with
\[\begin{split} op(A) = \left\{ \begin{array}{ll} A, & \text{if op(A) == rocsparse_operation_none} \\ A^T, & \text{if op(A) == rocsparse_operation_transpose} \\ \end{array} \right. \end{split}\]
,
\[\begin{split} op(B) = \left\{ \begin{array}{ll} B, & \text{if op(B) == rocsparse_operation_none} \\ B^T, & \text{if op(B) == rocsparse_operation_transpose} \\ \end{array} \right. \end{split}\]
and
\[\begin{split} spy(C)_ij = \left\{ \begin{array}{ll} 1 \text{ if i == j}, & 0 \text{ if i != j} \\ \end{array} \right. \end{split}\]

Note

opA == rocsparse_operation_conjugate_transpose is not supported.

Note

opB == rocsparse_operation_conjugate_transpose is not supported.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • opA[in] dense matrix \(A\) operation type.

  • opB[in] dense matrix \(B\) operation type.

  • alpha[in] scalar \(\alpha\).

  • A[in] dense matrix \(A\) descriptor.

  • B[in] dense matrix \(B\) descriptor.

  • beta[in] scalar \(\beta\).

  • C[inout] sparse matrix \(C\) descriptor.

  • compute_type[in] floating point precision for the SDDMM computation.

  • alg[in] specification of the algorithm to use.

  • temp_buffer[in] temporary storage buffer allocated by the user. The size must be greater or equal to the size obtained with rocsparse_sddmm_buffer_size.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_value – the value of trans_A, trans_B, compute_type or alg is incorrect.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointeralpha and beta are invalid, A, B, D, C or temp_buffer pointer is invalid.

  • rocsparse_status_not_implementedopA == rocsparse_operation_conjugate_transpose or opB == rocsparse_operation_conjugate_transpose.

rocsparse_dense_to_sparse()#

rocsparse_status rocsparse_dense_to_sparse(rocsparse_handle handle, const rocsparse_dnmat_descr mat_A, rocsparse_spmat_descr mat_B, rocsparse_dense_to_sparse_alg alg, size_t *buffer_size, void *temp_buffer)#

Dense matrix to sparse matrix conversion.

rocsparse_dense_to_sparse rocsparse_dense_to_sparse performs the conversion of a dense matrix to a sparse matrix in CSR, CSC, or COO format.

Note

This function writes the required allocation size (in bytes) to buffer_size and returns without performing the dense to sparse operation, when a nullptr is passed for temp_buffer.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • mat_A[in] dense matrix descriptor.

  • mat_B[in] sparse matrix descriptor.

  • alg[in] algorithm for the sparse to dense computation.

  • buffer_size[out] number of bytes of the temporary storage buffer. buffer_size is set when temp_buffer is nullptr.

  • temp_buffer[in] temporary storage buffer allocated by the user. When a nullptr is passed, the required allocation size (in bytes) is written to buffer_size and function returns without performing the dense to sparse operation.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointermat_A, mat_B, or buffer_size pointer is invalid.

rocsparse_sparse_to_dense()#

rocsparse_status rocsparse_sparse_to_dense(rocsparse_handle handle, const rocsparse_spmat_descr mat_A, rocsparse_dnmat_descr mat_B, rocsparse_sparse_to_dense_alg alg, size_t *buffer_size, void *temp_buffer)#

Sparse matrix to dense matrix conversion.

rocsparse_sparse_to_dense rocsparse_sparse_to_dense performs the conversion of a sparse matrix in CSR, CSC, or COO format to a dense matrix

Note

This function writes the required allocation size (in bytes) to buffer_size and returns without performing the sparse to dense operation, when a nullptr is passed for temp_buffer.

Note

This function is non blocking and executed asynchronously with respect to the host. It may return before the actual computation has finished.

Parameters:
  • handle[in] handle to the rocsparse library context queue.

  • mat_A[in] sparse matrix descriptor.

  • mat_B[in] dense matrix descriptor.

  • alg[in] algorithm for the sparse to dense computation.

  • buffer_size[out] number of bytes of the temporary storage buffer. buffer_size is set when temp_buffer is nullptr.

  • temp_buffer[in] temporary storage buffer allocated by the user. When a nullptr is passed, the required allocation size (in bytes) is written to buffer_size and function returns without performing the sparse to dense operation.

Return values:
  • rocsparse_status_success – the operation completed successfully.

  • rocsparse_status_invalid_handle – the library context was not initialized.

  • rocsparse_status_invalid_pointermat_A, mat_B, or buffer_size pointer is invalid.