[PATCH] D74976: [CostModel][X86] Improve extract/insert element costs (PR43605)
Simon Pilgrim via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Fri Feb 21 11:39:20 PST 2020
RKSimon marked an inline comment as done.
RKSimon added inline comments.
================
Comment at: llvm/test/Transforms/SLPVectorizer/X86/commutativity.ll:37
-; CHECK-NEXT: [[TMP15:%.*]] = insertelement <16 x i8> [[TMP14]], i8 [[C]], i32 14
-; CHECK-NEXT: [[TMP16:%.*]] = insertelement <16 x i8> [[TMP15]], i8 [[C]], i32 15
-; CHECK-NEXT: [[TMP17:%.*]] = insertelement <2 x i8> undef, i8 [[A:%.*]], i32 0
----------------
It doesn't look like we were ever detecting this correctly as an insert+broadcast, we were relying on the cheap insert cost.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D74976/new/
https://reviews.llvm.org/D74976
More information about the llvm-commits
mailing list