ROCm math and compute libraries

ROCm math and compute libraries#

ROCm math and compute libraries provide GPU-accelerated implementations of common numerical operations including dense and sparse linear algebra, FFTs, random number generation, and deep learning primitives.

Libraries prefixed with roc* are native, high-performance implementations written in HIP specifically for AMD GPUs. Libraries prefixed with hip* are portable wrappers that implement NVIDIA CUDA-equivalent APIs, allowing CUDA applications to be ported to AMD GPUs with minimal code changes.

  • Composable Kernel 1.2.0 – Provides a programming model for writing performance critical kernels for machine learning workloads across multiple architectures.

  • hipBLAS 3.6.0 – BLAS-marshalling library that supports rocBLAS and cuBLAS backends.

  • hipBLASLt 1.4.1 – Provides general matrix-matrix operations with a flexible API and extends functionalities beyond traditional BLAS library.

  • hipCUB 4.6.0 – Thin header-only wrapper library on top of rocPRIM or CUB that allows project porting using the CUB library to the HIP layer.

  • hipFFT 1.0.25 – Fast Fourier Transforms (FFT)-marshalling library that supports rocFFT or cuFFT backends.

  • hipRAND 3.4.0 – Ports CUDA applications that use the cuRAND library into the HIP layer.

  • hipSOLVER 3.6.0 – LAPACK-marshalling library that supports rocSOLVER and cuSOLVER backends.

  • hipSPARSE 4.7.0 – SPARSE-marshalling library that supports rocSPARSE and cuSPARSE backends.

  • hipSPARSELt 0.2.9 – Sparse matrix-matrix operations library.

  • MIOpen 3.6.0 – An open source deep-learning library.

  • rocBLAS 5.6.0 – BLAS implementation (in the HIP programming language) on the ROCm runtime and toolchains.

  • rocFFT 1.0.39 – Software library for computing Fast Fourier Transforms (FFTs) written in HIP.

  • rocPRIM 4.6.0 – Header-only library for HIP parallel primitives.

  • rocRAND 5.0.0 – Provides functions that generate pseudorandom and quasirandom numbers.

  • rocSOLVER 3.36.0 – An implementation of LAPACK routines on ROCm software, implemented in the HIP programming language and optimized for AMD’s latest discrete GPUs.

  • rocSPARSE 5.0.0 – Exposes a common interface that provides BLAS for sparse computation implemented on ROCm runtime and toolchains (in the HIP programming language).

  • rocThrust 4.6.0 – Parallel algorithm library.

  • rocWMMA 2.2.1 – C++ library for accelerating mixed-precision matrix multiply-accumulate (MMA) operations.