This is an archive of the discontinued LLVM Phabricator instance.

[sparse][mlir] simplify lattice optimization logic
ClosedPublic

Authored by aartbik on Feb 20 2021, 6:46 PM.

Details

Summary

Simplifies the way lattices are optimized with less, but more
powerful rules. This also fixes an inaccuracy where too many
lattices resulted (expecting a non-existing universal index).
Also puts no-side-effects on all proper getters and unifies
bufferization flags order in integration tests (for future,
more complex use cases).

Diff Detail

Event Timeline

aartbik created this revision.Feb 20 2021, 6:46 PM
aartbik requested review of this revision.Feb 20 2021, 6:46 PM
bixia accepted this revision.Feb 22 2021, 2:48 PM
bixia added inline comments.
mlir/integration_test/Sparse/CPU/sparse_sampled_matmul.mlir
3

Shall convert-vector-to-scf go before convert-scf-to-std?

This revision is now accepted and ready to land.Feb 22 2021, 2:48 PM
aartbik marked an inline comment as done.Feb 22 2021, 2:53 PM
aartbik added inline comments.
mlir/integration_test/Sparse/CPU/sparse_sampled_matmul.mlir
3

Yes, good suggestion. Thanks.

aartbik updated this revision to Diff 325580.Feb 22 2021, 2:53 PM
aartbik marked an inline comment as done.

changed flag order

This revision was automatically updated to reflect the committed changes.