projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 936d1e9) | patch
[AMDGPU][GlobalISel] Add support for global atomicrmw fadd
authorJay Foad <jay.foad@amd.com>
Mon, 1 Mar 2021 17:24:57 +0000 (17:24 +0000)
committerJay Foad <jay.foad@amd.com>
Wed, 31 Mar 2021 10:13:00 +0000 (11:13 +0100)
commit5d0e9ddfa512ea3b2dd500e7abe93af30c1d9e11
treeb6f88ea94100c466343adc78104540b3137ccf92tree | snapshot
parent936d1e97a32d05471f845d69496abcbf54687e72commit | diff
[AMDGPU][GlobalISel] Add support for global atomicrmw fadd

This includes gfx908 which only has a no-return version of the
global_atomic_add_f32 instruction, using the same hack that was
previously implemented for selecting from the
llvm.amdgcn.global.atomic.fadd intrinsic.

Differential Revision: https://reviews.llvm.org/D97767
llvm/lib/Target/AMDGPU/AMDGPUInstructionSelector.cpp diff | blob | history
llvm/lib/Target/AMDGPU/AMDGPUInstructionSelector.h diff | blob | history
llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp diff | blob | history
llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-atomicrmw-fadd-global.mir [new file with mode: 0644] blob
llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-atomicrmw-fadd-local.mir [moved from llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-atomicrmw-fadd.mir with 100% similarity] blob | history
Domain: System / Toolchain;