[PATCH] D145614: [AARCH64] Enable STORE of v4i8 to help more vectorization opportunities
Dave Green via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Tue Mar 28 02:04:20 PDT 2023
dmgreen accepted this revision.
dmgreen added a comment.
This revision is now accepted and ready to land.
Thanks. LGTM
> What is t/b instructions?
They are top/bottom instructions. They can do things like trunc and insert into the bottom half of each lane, to extend the top half into a full lane.
================
Comment at: llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h:402
+ // We can vectorize store v4i8.
+ if (ScalarMemTy->isIntegerTy(8) && isPowerOf2_32(VF) && VF >=4)
+ return 4;
----------------
Slight formatting on the `VF >=4` -> `VF >= 4`, but maybe just use `VF == 4` as I believe all other power2 sizes will already be handled?
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D145614/new/
https://reviews.llvm.org/D145614
More information about the llvm-commits
mailing list