Shuenhoy/vtk-m
1
0
Fork 0
mirror of https://gitlab.kitware.com/vtk/vtk-m synced 2024-10-08 11:29:02 +00:00
Code Issues 2 Actions Packages Projects Releases Wiki Activity

Fix CUDA compilation error with Lagrangian filter

Browse Source 
CUDA architecture has a limited amount of memory available for
constants. The CUDA compiler uses this space to hold constants for some
optimizations. However, for large kernels, the number of constants
needed might be larger than the constant space available. For these
conditions, you have to disable this form of optimization with the `-
Xptxas --disable-optimizer-constants` flags.

Currently, the only file that seems to have this issue is the test for
the Lagrangian filter. Someone should take a closer look to see if this
filter in particular is making unnecessarily large worklet/kernel. (In
particular, why does the Lagrangian filter have a larger kernel than the
streamline and stream surface filters?)

If this occurance happens more often, we might need to add some ways to
configure it in the build.
This commit is contained in:
Kenneth Moreland 2021-02-15 12:38:21 -07:00
parent aec9890e96
commit 047d79672a
2 changed files with 32 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
examples/lagrangian/CMakeLists.txt
View File

@ -12,6 +12,19 @@ cmake_minimum_required(VERSION 3.12...3.15 FATAL_ERROR)
#Find the VTK-m package
find_package(VTKm REQUIRED QUIET)
if ((TARGET vtkm::cuda) OR (TARGET vtkm::kokkos_cuda))
# CUDA architecture has a limited amount of memory available for constants. The CUDA
# compiler uses this space to hold constants for some optimizations. However, for large
# kernels, the number of constants needed might be larger than the constant space
# available. For these conditions, you have to disable this form of optimization with
# the -Xptxas --disable-optimizer-constants flags.
# TODO: Find a more elegant way to do this. Either figure out a way around this problem
# or add more general flags to vtkm_library/vtkm_unit_tests for sources with "large" kernels.
set_source_files_properties(lagrangian.cxx PROPERTIES
COMPILE_OPTIONS "-Xptxas;--disable-optimizer-constants"
)
endif()
add_executable(Lagrangian lagrangian.cxx ABCfield.h)
target_link_libraries(Lagrangian PRIVATE vtkm_filter)
vtkm_add_target_information(Lagrangian

19
vtkm/filter/testing/CMakeLists.txt
View File

@ -94,6 +94,25 @@ if (VTKm_ENABLE_RENDERING)
)
endif()
if ((TARGET vtkm::cuda) OR (TARGET vtkm::kokkos_cuda))
# CUDA architecture has a limited amount of memory available for constants. The CUDA
# compiler uses this space to hold constants for some optimizations. However, for large
# kernels, the number of constants needed might be larger than the constant space
# available. For these conditions, you have to disable this form of optimization with
# the -Xptxas --disable-optimizer-constants flags.
# TODO: Find a more elegant way to do this. Either figure out a way around this problem
# or add more general flags to vtkm_library/vtkm_unit_tests for sources with "large" kernels.
set(large_kernel_sources
RegressionTestStreamline.cxx
UnitTestLagrangianFilter.cxx
UnitTestStreamlineFilter.cxx
UnitTestStreamSurfaceFilter.cxx
)
set_source_files_properties(${large_kernel_sources} PROPERTIES
COMPILE_OPTIONS "-Xptxas;--disable-optimizer-constants"
)
endif()
vtkm_unit_tests(
SOURCES ${unit_tests}
LIBRARIES ${libraries}