[PATCH] D126885: [SLP]Cost for a constant buildvector.

Fri Jun 3 13:03:29 PDT 2022

xbolva00 added inline comments.

================
Comment at: llvm/test/Transforms/SLPVectorizer/X86/pr46983.ll:146
 ; AVX-NEXT:    [[TMP4:%.*]] = zext i32 [[TMP1:%.*]] to i64
-; AVX-NEXT:    [[TMP5:%.*]] = bitcast i64* [[TMP0:%.*]] to <4 x i64>*
-; AVX-NEXT:    [[TMP6:%.*]] = load <4 x i64>, <4 x i64>* [[TMP5]], align 8, !tbaa [[TBAA5:![0-9]+]]
----------------
ABataev wrote:
> xbolva00 wrote:
> > Regression
> > 
> > https://github.com/llvm/llvm-project/issues/46327
> llvm-mca reports throughputs:
> For scalar code - 5.5
> AVX vector - 8.0
> AVX2 vector - 5.0
>  
> https://godbolt.org/z/rEc74dxza
Ah, right, these checks are for avx.

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D126885/new/

https://reviews.llvm.org/D126885