Building RCCL from source#
To build RCCL as part of the ROCm Core SDK, see TheRock build instructions. TheRock is the recommended way to build ROCm components from source.
Alternatively, you can build RCCL standalone using the following instructions.
To build RCCL directly from the source code, follow these steps. This guide also includes instructions explaining how to test the build. For information on using the quick start install script to build RCCL, see Install RCCL.
Requirements#
The following prerequisites are required to build RCCL:
ROCm-supported GPUs
Having the ROCm stack installed on the system, including the HIP runtime and the HIP-Clang compiler.
Build and install RCCL using the install script#
To quickly install RCCL using the install script, follow these steps. For additional tips, see RCCL usage tips.
Quick start RCCL build#
RCCL directly depends on the HIP runtime plus the HIP-Clang compiler, which are part of the ROCm software stack. For ROCm installation instructions, see the ROCm installation guide.
Use the install.sh helper script,
located in the projects/rccl directory of the rocm-systems repository,
to build and install RCCL with a single command. It uses hard-coded configurations that can be specified directly
when using cmake. However, it’s a great way to get started quickly and provides an
example of how to build and install RCCL.
Run the install script#
To build the library using the install script, use this command:
./install.sh
To build for only for your local machine’s AMD GPU architecture, reducing build time, add the -l flag.
./install.sh -l
For more information on the build options and flags for the install script, run the following command:
./install.sh --help
The RCCL build and installation helper script options are as follows:
RCCL build & installation helper script
Options:
--address-sanitizer Build with address sanitizer enabled
--amdgpu_targets Only compile for specified GPU architecture(s). For multiple targets, separate by ';' (builds for all supported GPU architectures by default)
--cmake-options Pass additional CMake options (e.g. --cmake-options "-DFOO=BAR -DBAZ=ON")
--debug Build debug library
--debug-fast Build debug library with lto optimization disabled (fast build times)
-d|--dependencies Install RCCL dependencies
--device-linker Build with assembly-extract device linker (default)
--disable-roctx Build without ROCTX logging
--disable-sym-kernels Disable symmetric memory kernels
--disable-warp-speed Disable WARP_SPEED kernel optimizations
--dump-asm Disassemble code and dump assembly with inline code
-c|--enable-code-coverage Enable code coverage
--enable_backtrace Build with custom backtrace support
--enable-mpi-tests Enable MPI-based tests (requires --debug and MPI installation; set MPI_PATH if not in /opt/ompi)
-f|--fast Quick-build RCCL (local gpu arch only, no backtrace)
--force-reduce-pipeline Force reduce_copy sw pipeline to be used for every reduce-based collectives and datatypes
-h|--help Prints this help message
-i|--install Install RCCL library (see --prefix argument below)
-j|--jobs Specify how many parallel compilation jobs to run (32 by default)
--kernel-resource-use Dump GPU kernel resource usage (e.g., VGPRs, scratch, spill) at link stage
-l|--local_gpu_only Only compile for local GPU architecture
--log-trace Build with log trace enabled (i.e. NCCL_DEBUG=TRACE)
--no_clean Don't delete files if they already exist
--no-device-linker Disable device linker, use standard -fgpu-rdc
--openmp-test-enable Enable OpenMP in rccl unit tests
-p|--package_build Build RCCL package
--prefix Specify custom directory to install RCCL to (default: `/opt/rocm`)
-q|--quiet-warnings Suppress majority of compiler warnings (not recommended)
--rocshmem Build with rocSHMEM support
--run_tests_all Run all rccl unit tests (must be built already)
-r|--run_tests_quick Run small subset of rccl unit tests (must be built already)
--static Build RCCL as a static library instead of shared library
-t|--tests_build Build rccl unit tests, but do not run
--time-trace Plot the build time of RCCL (requires `ninja-build` package installed on the system)
--ninja Use the Ninja generator instead of Make (requires `ninja-build`; recommended for multi-arch builds)
--verbose Show compile commands
Available RCCL-specific CMake options for --cmake-options:
-DBUILD_EXT_EXAMPLES=ON Build ext-{net,tuner,profiler} example plugins (default: OFF)
-DDWORDX4_INTRINSICS=OFF Disable dwordx4 intrinsics (default: ON)
-DENABLE_COMPRESS=OFF Disable GPU code compression (default: ON)
-DENABLE_IFC=ON Enable indirect function call (default: OFF)
-DFAULT_INJECTION=OFF Disable fault injection (default: ON)
-DRCCL_ROCPROFILER_REGISTER=OFF Disable rocprofiler-register support (default: ON)
-DTIMETRACE=ON Enable time-trace during compilation (default: OFF)
Environment variables:
ONLY_FUNCS Build only specified collective functions (debug builds only).
Restricts GPU kernel generation to the listed collectives, significantly
reducing build time during development. Use '|' to separate multiple functions.
Example: ONLY_FUNCS="AllReduce|SendRecv" ./install.sh --debug -t
Available: AllReduce, Broadcast, Reduce, AllGather, ReduceScatter,
AlltoAllPivot, SendRecv, AlltoAllGda, AlltoAllvGda
Advanced: Specify algo, protocol, redop, and type per collective.
ONLY_FUNCS="AllReduce RING SIMPLE Sum f32|SendRecv"
ROCSHMEM_INSTALL_DIR Path to a pre-built rocSHMEM installation (skips building from source)
Tip
By default, the RCCL install script builds all the GPU targets that are defined in DEFAULT_GPUS in CMakeLists.txt.
To target specific GPUs and potentially reduce the build time, use --amdgpu_targets along with
a semicolon (;) separated string list of the GPU targets.
Building the library using CMake#
RCCL has moved from the standalone ROCm/rccl repository into the
ROCm/rocm-systems monorepo, where it is located at projects/rccl. Clone
the monorepo using either of the following options.
Option A: Clone the full monorepo:
git clone https://github.com/ROCm/rocm-systems.git
cd rocm-systems
git checkout develop
Option B: Clone with sparse checkout. To limit your local checkout to only RCCL
and improve performance with the large monorepo, configure a partial, sparse
checkout so you only fetch the projects/rccl project:
git clone --no-checkout --filter=blob:none https://github.com/ROCm/rocm-systems.git
cd rocm-systems
git sparse-checkout init --cone
git sparse-checkout set projects/rccl
git checkout develop
For the default RCCL build on the develop branch, no git submodule step is
required. Then build the library from the projects/rccl directory:
cd projects/rccl
mkdir build
cd build
cmake ..
make -j 16 # Or some other suitable number of parallel jobs
Note
Optional rocSHMEM builds are configured separately and do not use RCCL git
submodules. Use ./install.sh --rocshmem or pass -DENABLE_ROCSHMEM=ON
together with -DROCSHMEM_INSTALL_DIR=<prefix> or
-DROCSHMEM_SOURCE_DIR=<path>.
rocSHMEM support drives RCCL’s GPU Direct Async (GDA) AllToAll path, which is
validated only on supported GDA platforms (gfx942-class systems with a
supported multi-node NIC and driver setup). Building with --rocshmem on
other GPU architectures is not a supported configuration and may fail. For
the GDA NIC and driver requirements, see the
rocSHMEM documentation.
You can substitute a different installation path by providing the path as a parameter
to CMAKE_INSTALL_PREFIX, for example:
cmake -DCMAKE_INSTALL_PREFIX=$PWD/rccl-install -DCMAKE_BUILD_TYPE=Release ..
Note
Ensure ROCm CMake is installed using the command apt install rocm-cmake. By default,
CMake builds the component in debug mode unless DCMAKE_BUILD_TYPE is specified.
Building the RCCL package and install package:#
After you have cloned the repository and built the library as described in the previous section, use this command to build the package on Debian-based distros:
cd projects/rccl/build
make package
sudo dpkg -i *.deb
Note
The RCCL package install process requires sudo or root access because it creates a directory
named rccl in /opt/rocm/. This is an optional step. RCCL can be used directly by including the path containing librccl.so.
Testing RCCL#
The RCCL unit tests are implemented using the Googletest framework in RCCL. These unit tests require Googletest 1.10
or higher to build and run (this dependency can be installed using the -d option for install.sh).
To run the RCCL unit tests, go to the build folder and the test subfolder,
then run the appropriate RCCL unit test executables.
The RCCL unit test names follow this format:
CollectiveCall.[Type of test]
Filtering of the RCCL unit tests can be done using environment variables
and by passing the --gtest_filter command line flag:
UT_DATATYPES=ncclBfloat16 UT_REDOPS=prod ./rccl-UnitTests --gtest_filter="AllReduce.C*"
This command runs only the AllReduce correctness tests with the float16 datatype.
A list of the available environment variables for filtering appears at the top of every run.
See the Googletest documentation
for more information on how to form advanced filters.
There are also other performance and error-checking tests for RCCL. They are maintained separately at rccl-tests.
Note
For more information on how to build and run rccl-tests, see the rccl-tests README file .