projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: cf736e2) | patch
[AMDGPU] Allow use of TTMP registers in AMDGPUResourceUsageAnalysis
authorJay Foad <jay.foad@amd.com>
Thu, 13 Apr 2023 14:53:09 +0000 (15:53 +0100)
committerJay Foad <jay.foad@amd.com>
Thu, 13 Apr 2023 15:56:22 +0000 (16:56 +0100)
commit2d39f5b5cdcb87fe968001586ab7949939099cfe
tree0ccbf573cb0cc1a6674b62ee3f20143a4b60b501tree | snapshot
parentcf736e232573c772f96ced4dd45f5d7340c644a1commit | diff
[AMDGPU] Allow use of TTMP registers in AMDGPUResourceUsageAnalysis

With architected SGPRs, workgroup IDs are passed into a compute shader
in TTMP registers. Allow for this in AMDGPUResourceUsageAnalysis instead
of failing an assertion.

Differential Revision: https://reviews.llvm.org/D148239
llvm/lib/Target/AMDGPU/AMDGPUResourceUsageAnalysis.cpp diff | blob | history
llvm/test/CodeGen/AMDGPU/lower-work-group-id-intrinsics.ll diff | blob | history
Domain: System / Toolchain;