This is an archive of the discontinued LLVM Phabricator instance.

[mlir][sparse] Favors defined dimension when optimize lattice points.
ClosedPublic

Authored by Peiming on Oct 5 2022, 6:07 PM.

Diff Detail

Event Timeline

Peiming created this revision.Oct 5 2022, 6:07 PM
Herald added a project: Restricted Project. · View Herald Transcript
Peiming requested review of this revision.Oct 5 2022, 6:07 PM
Peiming updated this revision to Diff 465632.Oct 5 2022, 6:09 PM

fix typo

Peiming updated this revision to Diff 465633.Oct 5 2022, 6:10 PM

update comments.

aartbik added inline comments.Oct 5 2022, 6:11 PM
mlir/lib/Dialect/SparseTensor/Utils/Merger.cpp
265

Can you keep this close to the second for-loop, because otherwise it reads a bit out of place here

269

are not -> is not

271

we only need to do this when !reset at this point, right?
otherwise they all go anyway

mlir/unittests/Dialect/SparseTensor/MergerTest.cpp
415

BONUS POINTS for adding a neat test!!!

aartbik added inline comments.Oct 5 2022, 6:13 PM
mlir/lib/Dialect/SparseTensor/Utils/Merger.cpp
278–280

oh, you can use be instead of calling size() each time

Peiming updated this revision to Diff 465634.Oct 5 2022, 6:13 PM
Peiming marked 3 inline comments as done.

address comments.

aartbik accepted this revision.Oct 5 2022, 6:14 PM
This revision is now accepted and ready to land.Oct 5 2022, 6:14 PM
Peiming marked an inline comment as done.Oct 5 2022, 6:14 PM
Peiming updated this revision to Diff 465636.Oct 5 2022, 6:14 PM

address comment.

This revision was landed with ongoing or failed builds.Oct 5 2022, 6:16 PM
This revision was automatically updated to reflect the committed changes.