This is an archive of the discontinued LLVM Phabricator instance.

[LV] Prevent LV to run cost model twice for VF=2
ClosedPublic

Authored by dcaballe on Jun 11 2018, 1:15 PM.

Details

Summary

This is a minor fix for LV cost model, where the cost for VF=2 is computed twice when the vectorization of the loop is forced without specifying a VF. It was reported by @xusx595 in the mailing list.

Diff Detail

Repository
rL LLVM

Event Timeline

dcaballe created this revision.Jun 11 2018, 1:15 PM
hsaito accepted this revision.Jun 11 2018, 2:40 PM

LGTM.

This revision is now accepted and ready to land.Jun 11 2018, 2:40 PM

LGTM. Thanks for this patch and having me as a reviewer. As my work on vectorization is not based on the llvm-trunk but still on top of VPlan, it is not straightforward for me to create patches.

Thank you both for the review!

Ayal added a subscriber: Ayal.Jun 12 2018, 1:02 PM
Ayal added inline comments.
lib/Transforms/Vectorize/LoopVectorize.cpp
5033 ↗(On Diff #150820)

Could have alternatively started from unsigned i = 2 * Width, as the condition above is essentially peeling the first iteration.

Ideally computing expectedCost(1) would also be saved in this case.

Thanks for the comments, Ayal! Please, let me know if you have any other concerns.

lib/Transforms/Vectorize/LoopVectorize.cpp
5033 ↗(On Diff #150820)

Good points!

Could have alternatively started from unsigned i = 2 * Width, as the condition above is essentially peeling the first iteration.

I initially did that but then I had to replicate the LLVM_DEBUG line also in the peeled iteration to be consistent. For that reason I opted for this approach which doesn't need that replication.

Ideally computing expectedCost(1) would also be saved in this case.

Agreed. Unfortunately ScalarCost is necessary below. I also thought that it'd be interesting to have the debug information about the scalar cost for these cases

If no more comments, I'll proceed with the commit.

Thank you all!
Diego

Ayal added inline comments.Jun 14 2018, 3:16 PM
lib/Transforms/Vectorize/LoopVectorize.cpp
5033 ↗(On Diff #150820)

Having a designated LLVM_DEBUG line explaining that the first VF considered is 2 instead of 1, might be helpful.

If cost is to be computed just for supplying it as debug information, it should appear under LLVM_DEBUG.

Anyway, mtcw, feel free to proceed.

dcaballe added inline comments.Jun 14 2018, 3:36 PM
lib/Transforms/Vectorize/LoopVectorize.cpp
5033 ↗(On Diff #150820)

Thanks, Ayal.

If cost is to be computed just for supplying it as debug information, it should appear under LLVM_DEBUG.

Not only for debug. Look at line 5059. Not sure if it would be executed when vectorization is forced, though, but I'd prefer not to change the behavior w.r.t the scalar cost in this patch.

I'll proceed then.

This revision was automatically updated to reflect the committed changes.