| commit | 7171014726cf5bcf4fc4a03f924e24166c0a8c92 | [log] [tgz] |
|---|---|---|
| author | Quinn Dawkins <quinn.dawkins@gmail.com> | Mon Mar 04 20:19:18 2024 -0500 |
| committer | GitHub <noreply@github.com> | Mon Mar 04 20:19:18 2024 -0500 |
| tree | dfa4e7ceb0245c681488ba70ab40a13bb2cd5f66 | |
| parent | 77758bdf49d336ad073227062ce9277893225067 [diff] |
[Codegen][ROCDL] Replace custom generalization pass with upstream one (#16662) After tiling in LLVMGPUVectorDistribute, the tiling configuration attributes are no longer necessary. Additionally, the generalization of named ops before vectorization is to make it easier to fold away unit extent dims before vectorizing. At this point, it is best to generalize all named ops to allow the unit dim folding patterns to apply more easily, so we can switch to the upstream pass for that and drop the local one that only applied to convolutions and contractions.
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.