Merge pull request #3186 from Ruyk/kokkos-sycl-cmake
Add kokkos-sycl.cmake preset
This commit is contained in:
15
cmake/presets/kokkos-sycl.cmake
Normal file
15
cmake/presets/kokkos-sycl.cmake
Normal file
@ -0,0 +1,15 @@
|
||||
# preset that enables KOKKOS and selects SYCL compilation with OpenMP
|
||||
# enabled as well. Also sets some performance related compiler flags.
|
||||
set(PKG_KOKKOS ON CACHE BOOL "" FORCE)
|
||||
set(Kokkos_ENABLE_SERIAL ON CACHE BOOL "" FORCE)
|
||||
set(Kokkos_ENABLE_OPENMP ON CACHE BOOL "" FORCE)
|
||||
set(Kokkos_ENABLE_CUDA OFF CACHE BOOL "" FORCE)
|
||||
set(Kokkos_ENABLE_SYCL ON CACHE BOOL "" FORCE)
|
||||
set(Kokkos_ARCH_MAXWELL50 on CACHE BOOL "" FORCE)
|
||||
set(BUILD_OMP ON CACHE BOOL "" FORCE)
|
||||
|
||||
set(CMAKE_CXX_COMPILER clang++ CACHE STRING "" FORCE)
|
||||
set(MPI_CXX_COMPILER "mpicxx" CACHE STRING "" FORCE)
|
||||
set(CMAKE_CXX_STANDARD 17 CACHE STRING "" FORCE)
|
||||
set(CMAKE_SHARED_LINKER_FLAGS "-Xsycl-target-frontend -O3" CACHE STRING "" FORCE)
|
||||
set(CMAKE_TUNE_FLAGS "-fgpu-inline-threshold=100000 -Xsycl-target-frontend -O3 -Xsycl-target-frontend -ffp-contract=on -Wno-unknown-cuda-version" CACHE STRING "" FORCE)
|
||||
@ -638,13 +638,14 @@ This list was last updated for version 3.5.0 of the Kokkos library.
|
||||
|
||||
-D CMAKE_CXX_COMPILER=${HOME}/lammps/lib/kokkos/bin/nvcc_wrapper
|
||||
|
||||
To simplify compilation, three preset files are included in the
|
||||
To simplify compilation, four preset files are included in the
|
||||
``cmake/presets`` folder, ``kokkos-serial.cmake``,
|
||||
``kokkos-openmp.cmake``, and ``kokkos-cuda.cmake``. They will
|
||||
enable the KOKKOS package and enable some hardware choice. So to
|
||||
compile with OpenMP host parallelization, CUDA device
|
||||
parallelization (for GPUs with CC 5.0 and up) with some common
|
||||
packages enabled, you can do the following:
|
||||
``kokkos-openmp.cmake``, ``kokkos-cuda.cmake``, and
|
||||
``kokkos-sycl.cmake``. They will enable the KOKKOS package and
|
||||
enable some hardware choice. So to compile with OpenMP host
|
||||
parallelization, CUDA device parallelization (for GPUs with CC 5.0
|
||||
and up) with some common packages enabled, you can do the
|
||||
following:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
|
||||
Reference in New Issue
Block a user