| commit | 59162e38590fbe194e2f5dc11bcfc02bffeb75fc | |
| author | Stanislav Mekhanoshin <Stanislav.Mekhanoshin@amd.com> | |
| Tue, 7 Mar 2023 20:39:49 +0000 (7 12:39 -0800) | ||
| committer | Stanislav Mekhanoshin <Stanislav.Mekhanoshin@amd.com> | |
| Wed, 8 Mar 2023 09:24:20 +0000 (8 01:24 -0800) | ||
| tree | c7cba1f0950d3b1a0904655bcb7da0c9152bdc77 | treesnapshot (tar.gz zip) |
| parent | 173e54c3677420e5774a120a4bc4208093bd5cc8 | commitdiff |
[AMDGPU] Skip buffer_wbl2 before atomic fence acquire
Memory models for gfx90a and gfx940 do not require buffer_wbl2
before the fence for acquire ordering, but we do insert the full
release.
Fixes: SWDEV-386785
Differential Revision: https://reviews.llvm.org/D145524
Memory models for gfx90a and gfx940 do not require buffer_wbl2
before the fence for acquire ordering, but we do insert the full
release.
Fixes: SWDEV-386785
Differential Revision: https://reviews.llvm.org/D145524
| llvm/lib/Target/AMDGPU/SIMemoryLegalizer.cpp | diffblobblamehistory | |
| llvm/test/CodeGen/AMDGPU/memory-legalizer-fence.ll | diffblobblamehistory |