Shuenhoy/vtk-m2
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge topic 'intel-finites'

Browse Source 
b93ee30ab Enable non-finite values with Intel compiler

Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: Sujin Philip <sujin.philip@kitware.com>
Merge-request: !3213
This commit is contained in:
Kenneth Moreland 2024-05-07 17:07:34 +00:00 committed by Kitware Robot
commit bf0390b8e1
2 changed files with 19 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
CMake/VTKmCompilerFlags.cmake
View File

@ -90,6 +90,19 @@ set_target_properties(
EXPORT_NAME developer_flags
)
# Intel OneAPI compilers >= 2021.2.0 turn on "fast math" at any non-zero
# optimization level. Suppress this non-standard behavior using the
# `-fp-model=precise` flag.
set(intel_oneapi_compiler_version_min "2021.2.0")
set(is_lang "$<COMPILE_LANGUAGE:CXX>")
set(is_intelllvm "$<CXX_COMPILER_ID:IntelLLVM>")
set(is_intelllvm_fastmath_assuming_version "$<VERSION_GREATER_EQUAL:$<CXX_COMPILER_VERSION>,${intel_oneapi_compiler_version_min}>")
set(intel_oneapi_compiler_detections
"$<AND:${is_lang},${is_intelllvm},${is_intelllvm_fastmath_assuming_version}>")
target_compile_options(vtkm_developer_flags
INTERFACE
"$<BUILD_INTERFACE:$<${intel_oneapi_compiler_detections}:-fp-model=precise>>")
# Additional warnings just for Clang 3.5+, and AppleClang 7+
# about failures to vectorize.
if (VTKM_COMPILER_IS_CLANG AND CMAKE_CXX_COMPILER_VERSION VERSION_GREATER 3.4)

6
docs/changelog/intel-finites.md Normal file
View File

@ -0,0 +1,6 @@
# Enable non-finite values with Intel compiler
The Intel compiler by default turns on an optimization that assumes that
all floating point values are finite. This breaks any ligitimate uses of
non-finite values including checking values with functions like `isnan`
and `isinf`. Turn off this feature for the intel compiler.