[PATCH] D74775: [Utils][x86] add an option to reduce scrubbing of shuffles with memops
Simon Pilgrim via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Tue Feb 18 09:21:43 PST 2020
RKSimon added a comment.
I think we we discussed recently about always exposing stack math - I don't care about seeing constant labels but tbh I think we probably need to see all register based pointer math on x86 (maybe with some way to merge X86/X64 registers?).
================
Comment at: llvm/test/CodeGen/X86/extractelement-load.ll:233
; X32-SSE2-NEXT: cmpeqss %xmm0, %xmm1
-; X32-SSE2-NEXT: movss {{.*#+}} xmm2 = mem[0],zero,zero,zero
+; X32-SSE2-NEXT: movss {{\.LCPI.*}}, %xmm2 # xmm2 = mem[0],zero,zero,zero
; X32-SSE2-NEXT: andps %xmm1, %xmm2
----------------
this is a constant - ideally we'd hide the LCPI and the comment would be the actual constant value
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D74775/new/
https://reviews.llvm.org/D74775
More information about the llvm-commits
mailing list