commit | 891f438c27f201ae0dde3e8408eefc6fca80bc5f | [log] [tgz] |
---|---|---|
author | Max191 <44243577+Max191@users.noreply.github.com> | Fri Sep 20 13:51:09 2024 -0700 |
committer | GitHub <noreply@github.com> | Fri Sep 20 16:51:09 2024 -0400 |
tree | a8b2faf8432edaf3fb203839946358216e7f92b3 | |
parent | d834aa7357179e0d806f3634d2efe3af2fa45171 [diff] |
[Codegen] Add control options in pack unpack decomposition (#18469) Some early patterns in pack/unpack decomposition avoid generating reshape ops for unit dim packs and unpacks. However, the TileAndFuse pipeline uses these reshape ops to propagate expanded shapes to other fusable ops. This PR adds an option to the DecomposePackUnPackOps pass to create reshape ops anyway for unit dim cases. The reason these unit dims show up right now is that the iree_linalg_ext.im2col op of a unit-batched conv will have a unit dimension in the batch dim. Ultimately, it would be good to allow for batchless im2col ops, but in general it is good to support ops that have required unit dimensions. When prototyping new ops, it can be easiest to not support rank-reducing cases at first (winograd ops are another example), so these unit dims may appear again in the future. This PR also adds an optional control function to the pass options, which controls which packs and unpacks get decomposed. The control function is currently expected to be used when the `useOnlyReshapes` option is true, since there is no control function in some upstream patterns yet, but adding the control function upstream and fixing this is left as a TODO. --------- Signed-off-by: Max Dawkins <max.dawkins@gmail.com>
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
Package | Release status |
---|---|
GitHub release (stable) | |
GitHub release (nightly) | |
Python iree-compiler | |
Python iree-runtime |
Host platform | Build status |
---|---|
Linux | |
macOS | |
Windows |
For the full list of workflows see https://iree.dev/developers/general/github-actions/.
See our website for more information.
Community meeting recordings: IREE YouTube channel
IREE is licensed under the terms of the Apache 2.0 License with LLVM Exceptions. See LICENSE for more information.