projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 361f2a7) | patch
AMDGPU/GlobalISel: Select v2s32->v2s16 G_TRUNC
authorMatt Arsenault <Matthew.Arsenault@amd.com>
Sun, 16 Feb 2020 00:18:47 +0000 (19:18 -0500)
committerMatt Arsenault <arsenm2@gmail.com>
Mon, 17 Feb 2020 14:20:13 +0000 (09:20 -0500)
commite5805529bf037d692d6ae84f5b612b103f1d3cae
tree3be3218c140331c980c65b17fc0dfde9ce80c63atree | snapshot
parent361f2a78185b489ce495a2f381bf1444d4f7860ecommit | diff
AMDGPU/GlobalISel: Select v2s32->v2s16 G_TRUNC

It would be nice if there was a way to avoid the tied operand, but as
far as I can tell there isn't a way to use or with op_sel to achieve
this
llvm/lib/Target/AMDGPU/AMDGPUInstructionSelector.cpp diff | blob | history
llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-trunc.v2s16.mir [new file with mode: 0644] blob
llvm/test/CodeGen/AMDGPU/GlobalISel/trunc.ll [new file with mode: 0644] blob
Domain: System / Toolchain;