Skip to content

Conversation

@sjarus
Copy link
Collaborator

@sjarus sjarus commented Jan 14, 2022

Signed-off-by: Suraj Sudhir suraj.sudhir@arm.com

Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
@sjarus sjarus requested a review from cathyzhyi January 14, 2022 18:50
@sjarus
Copy link
Collaborator Author

sjarus commented Jan 14, 2022

This pass is needed because TOSA-to-LinAlg depends on TOSA and Standard dialect content, with tosa.const being converted to arith.constants by tosa-to-standard. Without it, TOSA-to-LinAlg fails stating it can't recognize tosa.const() . This update fixes @anupgangwar's earlier AtenPowOp implementation in the process.

Copy link
Contributor

@cathyzhyi cathyzhyi left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@sjarus sjarus merged commit 5ded7d0 into llvm:main Jan 14, 2022
@sjarus sjarus deleted the tosa_updates branch January 14, 2022 21:46
qedawkins pushed a commit to nod-ai/torch-mlir that referenced this pull request Oct 3, 2022
* Map ONNX operator names to instructions

This change maps the ONNX operator names (e.g. onnx.Add) to their
corresponding MLIR instructions as  named location information.
Downstream, the operator names can be used to
more accurately propagate the kernel ids that were responsible for generating different MLIR instructions. To avoid extra code churn I only modified a subset of the common operators.

Signed-off-by: kile <kile@microsoft.com>

* formatting and template

Signed-off-by: kile <kile@microsoft.com>

Co-authored-by: Tian Jin <tjingrant@gmail.com>
Co-authored-by: Kevin O'Brien <caomhin@us.ibm.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants