Skip to content

[LIBCLC][AMDGCN] Fix get_max_sub_group_size #5386

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

Closed
wants to merge 2 commits into from
Closed

[LIBCLC][AMDGCN] Fix get_max_sub_group_size #5386

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
f324e44
[LIBCLC][AMDGCN] Fix get_max_sub_group_size
npmiller Jan 25, 2022
49d4130
Update libclc/amdgcn-amdhsa/libspirv/workitem/get_max_sub_group_size.cl
npmiller Jan 25, 2022
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
24 changes: 9 additions & 15 deletions libclc/amdgcn-amdhsa/libspirv/workitem/get_max_sub_group_size.cl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,21 +8,15 @@

#include <spirv/spirv.h>

// FIXME: Remove the following workaround once the clang change is released.
// This is for backward compatibility with older clang which does not define
// __AMDGCN_WAVEFRONT_SIZE. It does not consider -mwavefrontsize64.
// See:
// https://github.com/intel/llvm/blob/sycl/clang/lib/Basic/Targets/AMDGPU.h#L414
// and:
// https://github.com/intel/llvm/blob/sycl/clang/lib/Basic/Targets/AMDGPU.cpp#L421
#ifndef __AMDGCN_WAVEFRONT_SIZE
#if __gfx1010__ || __gfx1011__ || __gfx1012__ || __gfx1030__ || __gfx1031__
#define __AMDGCN_WAVEFRONT_SIZE 32
#else
#define __AMDGCN_WAVEFRONT_SIZE 64
#endif
#endif
// The clang driver will define this variable depending on the architecture and
// compile flags by linking in ROCm bitcode defining it to true or false. If
// it's 1 the wavefront size used is 64, if it's 0 the wavefront size used is
// 32.
extern constant unsigned char __oclc_wavefrontsize64;

_CLC_DEF _CLC_OVERLOAD uint __spirv_SubgroupMaxSize() {
return __AMDGCN_WAVEFRONT_SIZE;
if (__oclc_wavefrontsize64 == 1) {
return 64;
}
return 32;
}