onnx / onnx-mlir

Representation and Reference Lowering of ONNX Models in MLIR Compiler Infrastructure
Apache License 2.0
742 stars 315 forks source link

How to use a new graph-level scheduling method in the MLIR project? #1390

Open XuZheng2000 opened 2 years ago

XuZheng2000 commented 2 years ago

We want to test the performance of a new graph-level scheduling method by MLIR, how do we do it? Do we need to build a new Dialect by ourselves?If yes, where should this dialect be, is it on top of the Affine dialect?

AlexandreEichenberger commented 2 years ago

@XuZheng2000 are you sure you did not ment to address this question to the MLIR community? Here we are using MLIR to lower ONNX graphs. We inherit all properties from MLIR, so any new optimizations found there are eventually available within this project as well.