[PATCH] D143787: [X86] Try to use `{v}shufps` instead of `vpermilps` for common float shuffles.
Simon Pilgrim via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Sat Feb 11 03:19:18 PST 2023
RKSimon added a comment.
Please can you split off the patch adding the free domain switch tuning flags? Its been something we've been missing from combineX86ShuffleChain for some time and should be addressed first.
================
Comment at: llvm/lib/Target/X86/X86.td:533
+ "NoBypassDelay","true",
+ "Has no bypass delay when using the 'wrong' domain">;
+
----------------
Maybe rename to TuningNoDomainDelay/TuningNoDomainDelayShuffle - bypass could mean many things.....
================
Comment at: llvm/lib/Target/X86/X86.td:1005
TuningPadShortFunctions,
TuningInsertVZEROUPPER];
----------------
Bonnell has no domain penalty either (and REALLY likes the shorted instruction encodings).
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D143787/new/
https://reviews.llvm.org/D143787
More information about the llvm-commits
mailing list