commit | 575a97d96bf8b88502181271a28655dec7ea094b | [log] [tgz] |
---|---|---|
author | Scott Todd <scotttodd@google.com> | Mon Nov 13 14:58:50 2023 -0800 |
committer | GitHub <noreply@github.com> | Mon Nov 13 22:58:50 2023 +0000 |
tree | 762d7439d7c32fcf20245f4e8526042dc576dcf1 | |
parent | bd52ed05180dfbaa9874f2cb4b426800aec3ba74 [diff] |
Rename "torch_iree" plugin to "input_torch". (#15576) Progress on https://github.com/openxla/iree/issues/15468 Bikeshedding galore. Most of the non-mechanical parts of the code moves for input dialect plugins have been completed, so now we can see the plugins side-by-side and decide on how we want them to look. We have top level plugin folders: * `compiler/plugins/input/Torch` * `compiler/plugins/input/TOSA` * `compiler/plugins/input/StableHLO` * `compiler/plugins/target/CUDA` * `compiler/plugins/target/...` (coming soon) Then each input plugin has `compiler/plugins/input/[DIALECT]/[dialect]-iree/PluginRegistration.cpp` with the `input_[dialect]` plugin name. * I'm also considering making the folder name match what the plugin is called directly, so we'd have e.g. `compiler/plugins/input/Torch/input-torch/PluginRegistration.cpp` (name "input-torch" with a dash?) * Alternately, we could have plugins called `torch-iree`, `stablehlo-iree`, and `tosa-iree`. IMO since these are all IREE compiler plugins the "iree" is redundant and we should instead highlight "input" in the name.  We could further collapse these trees if we wanted, but Torch currently also has sibling `torch-mlir/` and `torch-mlir-dialect/` folders.
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.