[PATCH] D89566: [LV] Epilogue Vectorization with Optimal Control Flow
Sjoerd Meijer via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Wed Nov 4 05:06:26 PST 2020
SjoerdMeijer added a comment.
In D89566#2373443 <https://reviews.llvm.org/D89566#2373443>, @dmgreen wrote:
>> As it stands right now, VPlan can only model control-flow inside a loop. Since epilogue vectorization is concerned with control-flow around (and outside) the loop, there isn't much that can be done today to make the transformation "VPlany". I understand the ultimate goal of vplan is to also model the context around subject loops (eg the entire loop nest or other code surrounding loops). I wondered about whether it's worth to delay this work until that becomes available, but most of the feedback I received was in the direction of let's get it done now and then do it in vplan when it's capable of representing surrounding context.
>
> Yeah. I don't think VPlan should slow this down. The problem is that if no-one pushes on vplan to have those extra features, they will never appear :)
FWIW, I agree with both. The more features we add to the vectoriser, the more difficult it will be for VPlan to catch up. At the same time, I don't think there's an initiative to do this, so waiting for VPlan would be unreasonable. Also, epilogue vectorisation has been an outstanding issue for so long, so it is time it gets addressed.
Just checking/summarising: is there consensus that this is the patch that is going to do epilogue vectorisation, and not D88819 <https://reviews.llvm.org/D88819>? Or does that depend on perf numbers as Florian suggested?
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D89566/new/
https://reviews.llvm.org/D89566
More information about the llvm-commits
mailing list