[PATCH] D111546: [X86][LV] X86 does *not* prefer vectorized addressing
Roman Lebedev via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Fri Oct 15 00:29:18 PDT 2021
lebedev.ri marked an inline comment as done.
lebedev.ri added a comment.
In D111546#3066122 <https://reviews.llvm.org/D111546#3066122>, @pengfei wrote:
> In D111546#3065618 <https://reviews.llvm.org/D111546#3065618>, @craig.topper wrote:
>
>> I think this makes sense.
>>
>> @pengfei do you want to check this on Intel's benchmarks?
>
> Yes, I can check our internal benchmark.
Thank you.
I'm not sure how this measures standalone,
my main hope this unblocks D111460 <https://reviews.llvm.org/D111460>.
> I guess this only affect non AVX512 targets, right?
Yes.
================
Comment at: llvm/lib/Target/X86/X86TargetTransformInfo.cpp:5017-5019
+bool X86TTIImpl::supportsEfficientVectorElementLoadStore() const {
+ return false;
+}
----------------
pengfei wrote:
> Do we need to override here? I think it returns false by default.
I think we want to be explicit here.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D111546/new/
https://reviews.llvm.org/D111546
More information about the llvm-commits
mailing list