| commit | 142894130b8506cb1860729dd2a988b7756fcdca | [log] [tgz] |
|---|---|---|
| author | MaheshRavishankar <1663364+MaheshRavishankar@users.noreply.github.com> | Mon Jan 30 21:44:55 2023 -0800 |
| committer | GitHub <noreply@github.com> | Tue Jan 31 05:44:55 2023 +0000 |
| tree | 52b089948705b5b0c406335a2e909dd439e6c963 | |
| parent | a9b8efb51715c31b6f0177e83ed7afbd1c4b2f62 [diff] |
Add a `--iree-preprocessing-pass-pipeline` to allow user control on preprocessing passes before IREE compilation. (#11986) This PR adds the --iree-preprocessing-pass-pipeline that allows users to run a sequence of passes (using the MLIR -pass-pipeline syntax) to control the sequnece of passes that do preprocessing of the program after conversion from Input dialects (like MHLO/TOSA), before running any of the core IREE compilation pipelines. This allows user control on using things like Winograd/Im2Col transformation etc. that is geared towards specific use cases. This could also be used as a experimental/prototyping space. If there are transformations that apply to all backends and all models, they could be moved out of preprocessing into the core IREE compilation pipelines. This also removes the need to add some global options to iree-compile, reducing the number of -iree-flow-* options. More will be removed in the future.
IREE (Intermediate Representation Execution Environment, pronounced as “eerie”) is an MLIR-based end-to-end compiler and runtime that lowers Machine Learning (ML) models to a unified IR that scales up to meet the needs of the datacenter and down to satisfy the constraints and special considerations of mobile and edge deployments.
See our website for project details, user guides, and instructions on building from source.
IREE is still in its early phase. We have settled down on the overarching infrastructure and are actively improving various software components as well as project logistics. It is still quite far from ready for everyday use and is made available without any support at the moment. With that said, we welcome any kind of feedback on any communication channels!
See our website for more information.
IREE is licensed under the terms of the Apache 2.0 License with LLVM Exceptions. See LICENSE for more information.