Skip to content

Fix erroneous MPI_Bcast sizing in autotune.cc and remove O1 optimization cap. #153

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
romerojosh merged 2 commits into from
Jul 15, 2026
+8 −9
Merged

Fix erroneous MPI_Bcast sizing in autotune.cc and remove O1 optimization cap. #153

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
9126bf2
Remove forced O1 optimization cap for host code compilation.
romerojosh Jul 15, 2026
4f8fa05
Fix incorrect MPI_Bcast sizing.
romerojosh Jul 15, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 6 additions & 7 deletions CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -35,13 +35,6 @@ endif()
string(REPLACE "compilers/bin/nvc++" "cmake" NVHPC_CMAKE_DIR ${NVHPC_CXX_BIN})
set(CMAKE_PREFIX_PATH ${NVHPC_CMAKE_DIR})

# Limit optimization levels of CPU code compilation
set(CMAKE_CXX_FLAGS_RELEASE "-O1 -DNDEBUG")
set(CMAKE_CXX_FLAGS_RELWITHDEBINFO "-O1 -g -DNDEBUG")
set(CMAKE_Fortran_FLAGS_RELEASE "-O1 -DNDEBUG")
# Note: Removing "-g" flag from Fortran RelWithDebInfo to enable mixed compilation with cc100 and older compute capabilities
set(CMAKE_Fortran_FLAGS_RELWITHDEBINFO "-O1 -DNDEBUG")

if (CUDECOMP_BUILD_FORTRAN)
set(LANGS CXX CUDA Fortran)
else()
Expand All @@ -50,6 +43,12 @@ endif()

project(cudecomp LANGUAGES ${LANGS})

if (CUDECOMP_BUILD_FORTRAN)
# Remove debug information to enable mixed Fortran compilation with cc100 and older compute capabilities.
string(REGEX REPLACE "(^| )-g(opt)?($| )" "\\3" CMAKE_Fortran_FLAGS_RELWITHDEBINFO
"${CMAKE_Fortran_FLAGS_RELWITHDEBINFO}")
endif()

# Set up CUDA compute capabilities by CUDA version. Users can override defaults with CUDECOMP_CUDA_CC_LIST
if (NOT CUDECOMP_CUDA_CC_LIST)
if (${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 13.0)
Expand Down
4 changes: 2 additions & 2 deletions src/autotune.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -588,7 +588,7 @@ void autotuneTransposeBackend(cudecompHandle_t handle, cudecompGridDesc_t grid_d

// Set handle to best option (broadcast from rank 0 for consistency)
CHECK_MPI(MPI_Bcast(&comm_backend_best, sizeof(cudecompTransposeCommBackend_t), MPI_CHAR, 0, handle->mpi_comm));
CHECK_MPI(MPI_Bcast(pdims_best, 2 * sizeof(int), MPI_INT, 0, handle->mpi_comm));
CHECK_MPI(MPI_Bcast(pdims_best, 2, MPI_INT, 0, handle->mpi_comm));

grid_desc->config.transpose_comm_backend = comm_backend_best;
grid_desc->config.pdims[0] = pdims_best[0];
Expand Down Expand Up @@ -966,7 +966,7 @@ void autotuneHaloBackend(cudecompHandle_t handle, cudecompGridDesc_t grid_desc,

// Set handle to best option (broadcast from rank 0 for consistency)
CHECK_MPI(MPI_Bcast(&comm_backend_best, sizeof(cudecompHaloCommBackend_t), MPI_CHAR, 0, handle->mpi_comm));
CHECK_MPI(MPI_Bcast(pdims_best, 2 * sizeof(int), MPI_INT, 0, handle->mpi_comm));
CHECK_MPI(MPI_Bcast(pdims_best, 2, MPI_INT, 0, handle->mpi_comm));

grid_desc->config.halo_comm_backend = comm_backend_best;
grid_desc->config.pdims[0] = pdims_best[0];
Expand Down
Loading