[PATCH] D89697: * [x86] Implement smarter instruction lowering for FP_TO_UINT from vXf32/vXf64 to vXi32 for SSE2 and AVX2 by using the exact semantic of the CVTTPS2SI instruction.
Sanjay Patel via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Tue Jul 13 09:41:38 PDT 2021
spatel accepted this revision.
spatel added a comment.
This revision is now accepted and ready to land.
In D89697#2874132 <https://reviews.llvm.org/D89697#2874132>, @RKSimon wrote:
> In D89697#2873919 <https://reviews.llvm.org/D89697#2873919>, @spatel wrote:
>
>> Can we make the cost model changes as a preliminary/independent commit?
>
> Are there any specific cost changes that you think can be pulled out? Most of the updates are necessary to match the change in codegen.
Ah, right.
LGTM - some of the test diffs with an extra instruction seem like they would not be wins, but I'm guessing those patterns are rare and the perf diff would be in the noise.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D89697/new/
https://reviews.llvm.org/D89697
More information about the llvm-commits
mailing list