Page MenuHomePhabricator

[mlir] Move memref.[tensor_load|buffer_cast|clone] to "bufferization" dialect.
ClosedPublic

Authored by pifon2a on Wed, Nov 24, 11:10 AM.

Diff Detail

Event Timeline

pifon2a created this revision.Wed, Nov 24, 11:10 AM
pifon2a requested review of this revision.Wed, Nov 24, 11:10 AM
Herald added a project: Restricted Project. · View Herald TranscriptWed, Nov 24, 11:10 AM
mehdi_amini added inline comments.Wed, Nov 24, 11:28 AM
mlir/include/mlir/Dialect/Bufferization/IR/BufferizationBase.td
20

This is what will show up on the website, I think we need a more extensive doc here, to set up the context and also linking to the other existing docs.

springerm added inline comments.
mlir/include/mlir/Dialect/Linalg/ComprehensiveBufferize/BufferizableOpInterface.h
386

This should probably not be part of this change.

herhut accepted this revision.Thu, Nov 25, 12:39 AM

Thank you for cleaning this up. Just some nits and please address the comments others already had.

I would be OK with improving documentation in a follow up, just to avoid merge conflicts and rebase.

mlir/lib/Dialect/Bufferization/IR/BufferizationOps.cpp
185

This comment is now off.

268

This is not dim of cast but dim of ToMemref.

mlir/lib/Dialect/Linalg/ComprehensiveBufferize/ComprehensiveBufferize.cpp
1051

nit: variable name

1062

nit: variable name.

pifon2a accepted this revision.Thu, Nov 25, 2:43 AM
pifon2a marked 5 inline comments as done.
pifon2a added inline comments.
mlir/include/mlir/Dialect/Bufferization/IR/BufferizationBase.td
20

Good point, also updated the names of the ops in Bufferization.md and BufferDeallocationInternals.md

pifon2a resigned from this revision.Thu, Nov 25, 2:43 AM
pifon2a updated this revision to Diff 389707.Thu, Nov 25, 2:43 AM

address the comments.

This revision was not accepted when it landed; it landed in state Needs Review.Thu, Nov 25, 2:57 AM
This revision was landed with ongoing or failed builds.
This revision was automatically updated to reflect the committed changes.

LGTM for the sparse part (there are some dangling references in the doc now to "memref" but since this is submitted, I will fix that in a follow up).

LGTM for the sparse part (there are some dangling references in the doc now to "memref" but since this is submitted, I will fix that in a follow up).

Thank you, Aart!