[PATCH] D149742: [SLP]Improve isGatherShuffledEntry by trying per-register shuffle.

Thu Jun 29 13:03:55 PDT 2023

ABataev added a comment.

In D149742#4460718 <https://reviews.llvm.org/D149742#4460718>, @vdmitrie wrote:

> In D149742#4460486 <https://reviews.llvm.org/D149742#4460486>, @ABataev wrote:
>
>> In D149742#4401811 <https://reviews.llvm.org/D149742#4401811>, @vdmitrie wrote:
>>
>>> F27851102: test.ll <https://reviews.llvm.org/F27851102>
>>>
>>> command to reproduce: opt -passes=slp-vectorizer -mtriple=x86_64 -mcpu=core-avx2 -S test.ll
>>
>> HMM, llvm-mca shows that the code after patch is better than the code before (with the patch the throughput is 4.0, without - 5.0)
>>
>> https://godbolt.org/z/zvzTKedPd
>
> It puzzled me too actually. It was the only vectorization candidate that changed behavior due to the patch in the test. I could not find any obvious flaw in handling subsequent vectorization sequences.
> Eyeballing of generated code before/after also did not reveal any obvious issues.

I fixed cost estimation, now it results in the same translation (I added your test, no changes)

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D149742/new/

https://reviews.llvm.org/D149742