projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 056a8ba) | patch
InstCombineSimplifyDemanded: Allow v3 results for AMDGCN buffer and image intrinsics
authorTim Renouf <tpr.llvm@botech.co.uk>
Fri, 22 Mar 2019 15:53:50 +0000 (15:53 +0000)
committerTim Renouf <tpr.llvm@botech.co.uk>
Fri, 22 Mar 2019 15:53:50 +0000 (15:53 +0000)
commit94c163c34e4d1d9f2fd7887f198a6fac3c71e271
tree5e01a80f5df9d56ed107568cbac8f90e60da0173tree | snapshot
parent056a8ba48378a0704bb873d2f49cbff8ba4a28e3commit | diff
InstCombineSimplifyDemanded: Allow v3 results for AMDGCN buffer and image intrinsics

This helps to avoid the situation where RA spots that only 3 of the
v4f32 result of a load are used, and immediately reallocates the 4th
register for something else, requiring a stall waiting for the load.

Differential Revision: https://reviews.llvm.org/D58906

Change-Id: I947661edfd5715f62361a02b100f14aeeada29aa
llvm-svn: 356768
llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp diff | blob | history
llvm/test/Transforms/InstCombine/AMDGPU/amdgcn-demanded-vector-elts.ll diff | blob | history
Domain: System / Toolchain;