This is an archive of the discontinued LLVM Phabricator instance.

[BPI] Add new LazyBPI analysis
ClosedPublic

Authored by anemet on Jul 26 2016, 3:32 PM.

Details

Summary

The motivation is the same as in D22141: In order to add the hotness
attribute to optimization remarks we need BFI to be available in all
passes that emit optimization remarks. BFI depends on BPI so unless we
make this lazy as well we would still compute BPI unconditionally.

The solution is to use the new LazyBPI pass in LazyBFI and only compute
BPI when computation of BFI is requested by the client.

I extended the laziness test using a LoopDistribute test to also cover
BPI.

Diff Detail

Repository
rL LLVM

Event Timeline

anemet updated this revision to Diff 65617.Jul 26 2016, 3:32 PM
anemet retitled this revision from to [BPI] Add new LazyBPI analysis.
anemet updated this object.
anemet added reviewers: davidxl, hfinkel.
anemet added a subscriber: llvm-commits.
davidxl accepted this revision.Jul 28 2016, 10:26 AM
davidxl edited edge metadata.

lgtm

This revision is now accepted and ready to land.Jul 28 2016, 10:26 AM
This revision was automatically updated to reflect the committed changes.