[PATCH] D93129: [LV] Do not use vector type to compute cost of scalar address comp.
Sanjay Patel via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Tue Jan 19 09:13:47 PST 2021
spatel added a reviewer: pengfei.
spatel added a comment.
I haven't looked at x86 gather/scatter code at all - ping @pengfei for more thoughts on that.
================
Comment at: llvm/test/Transforms/LoopVectorize/X86/gather-cost.ll:16-17
+
+; CHECK-LABEL: @test1
+; CHECK: <4 x float>
----------------
Seems like we should do a bit more checking here to show what is happening. Would it make sense to also add a RUN line with `-mcpu=skylake` or similar, so we can show some difference between the full vectorization with gathers vs. partial without gathers?
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D93129/new/
https://reviews.llvm.org/D93129
More information about the llvm-commits
mailing list