[PATCH] D83156: [X86] Teach lowerShuffleAsBlend to use bit blend for v16i8/v32i8/v16i16 when avx512vl is enabled but not avx512bw.
Simon Pilgrim via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Sat Jul 4 04:17:38 PDT 2020
RKSimon added a comment.
In D83156#2131280 <https://reviews.llvm.org/D83156#2131280>, @craig.topper wrote:
> Thanks Simon. Thoughts on whether we should do this with BWI too? I think the bit blend would also work better with shuffle combining?
Yes, my only concern is that currently we don't anything to combine variable select patterns OTHER than bit blend - ternlog/vselect are neither handled as faux shuffles or combined to at the moment which they probably need to be.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D83156/new/
https://reviews.llvm.org/D83156
More information about the llvm-commits
mailing list