* Make CAS loops use the TTAS idiom.
* More efficient re-formulation of TTAS.
* Fix typo.
* The typo was not quite a typo
[ROCm/hip commit: 9ba25b42c8]
* [hip] add support for implicit kernel argument for multi-grid sync
* modified code for calculating the prev_sum
* change the impCoopArg type to size_t
* add memory clean up
* launch init_gws and main kernels into two separate loops
[ROCm/hip commit: 359dc79101]
[Algorithm]
[Release]
If CMAKE_INSTALL_PREFIX is set by the user:
If BIN_INSTALL_DIR is set by HIP, use it as CMAKE_INSTALL_PREFIX, otherwise CMAKE_INSTALL_PREFIX is used unchanged.
If the user does not set CMAKE_INSTALL_PREFIX (CMAKE_INSTALL_PREFIX_INITIALIZED_TO_DEFAULT):
If BIN_INSTALL_DIR is set by HIP, use it as CMAKE_INSTALL_PREFIX, otherwise use PROJECT_BINARY_DIR/bin for installation.
[Debug]
If CMAKE_INSTALL_PREFIX is set by the user:
CMAKE_INSTALL_PREFIX is used unchanged.
If the user does not set CMAKE_INSTALL_PREFIX (CMAKE_INSTALL_PREFIX_INITIALIZED_TO_DEFAULT):
use CMAKE_CURRENT_SOURCE_DIR/bin for installation.
Standalone build left unchanged: CMAKE_INSTALL_PREFIX is used if set.
[ROCm/hip commit: 75d70a6714]
By implicit unconditional passing -fno-delayed-template-parsing option (which appeared in LLVM 3.8.0, thus doesn't need compatibility wrapping) to hipify-clang.
[Reason] To parse uncalled template functions otherwise they are not parsed without calling, thus not hipified.
Affects cub_03.cu test, which has uncalled global template function.
[ROCm/hip commit: b6e6f12b54]
[Reason] To support maximum CUDA features in offline tests
+ Add defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 600 restriction for atomicAdd on doubles in atomics.cu.
So if LLVM < 7 and --cuda-gpu-arch doesn't work, __CUDA_ARCH__ is unset too (350 by default in clang);
if LLVM >= 7 --cuda-gpu-arch is used and __CUDA_ARCH__ is set based on it.
[ROCm/hip commit: 39e7d213cf]
[Reason] To support maximum CUDA features in offline tests
+ Add CUDA_VERSION >= 800 restriction for atomics.cu
[TODO] Find a way to use or exclude atomicAdd for doubles if LLVM < 7, because
LLVM 6.0.1 and older do not use --cuda-gpu-arch in clang's Driver code at all (option is only declared)
[ROCm/hip commit: 14b4df126c]
Fixes SWDEV-203394.
Currently in runTest() returns true, even if the texture reference copy does not happen. Using the existing testResult Flag to return from runTest().
[ROCm/hip commit: fbb98f4950]
* Temporary add -D_OPENMP to clang options in hipcc to allow using CMake OpenMP detection with hip-clang (until updated CMake version is available).
[ROCm/hip commit: 596bf4e326]
* guard against division by zero for no VGPR usage (e.g., in an empty kernel)
* fix bracket format
* clean up parenthesis
[ROCm/hip commit: 73ca2b0083]
+ affects cuDNN and CUB tests, paths to libraries of which are defined by CUDA_DNN_ROOT_DIR and CUDA_CUB_ROOT_DIR
+ Warn about excluding and why, for instance:
"WARN: cuDNN tests are excluded due to unset CUDA_DNN_ROOT_DIR"
[ROCm/hip commit: 3444834e3a]
Rahul Garg
Jatin Chaudhary
gandryey
Alex Voicu
satyanveshd
searlmc1
Aryan Salmanpour
Evgeny Mankov
Vladislav Sytchenko
Aaron Enye Shi
kjayapra-amd
vsytch
kpyzhov
Nick Curtis
Xiaozhu Meng
Yaxun (Sam) Liu