commit | 02cfcd1e5561c0db0470d09644a1f527c259c208 | [log] [tgz] |
---|---|---|
author | Stella Laurenzo <stellaraccident@gmail.com> | Sun Nov 14 13:20:53 2021 -0800 |
committer | GitHub <noreply@github.com> | Sun Nov 14 13:20:53 2021 -0800 |
tree | 29132a9cfe44db1adfc2569209d31b9f2893f7bb | |
parent | 026cdda0c4be738476dd263c44db22d5ba4211c4 [diff] |
Moves the linalg_ext dialect to iree_linalg_ext under the iree-dialects project. (#7657) Mostly NFC: * Pre-factors iree-dialects into a consistent state wrt namespaces and directory layouts. * Moves linalg_ext to a new dialect under iree-dialects. Does some adaptation to upstream style along the way. * Redirects everything that was using it to use the new one. * Enables tests for iree-dialects in the cmake CI (they were not enabled) and fixes some type prefixing that had drifted. (Will follow up with enabling them in the internal builds to guard against further regression). Non-NFC: *When tiling, the old pass was directly generating flow.dispatch.workgroup... ops to query the current workgroup id. We had been planning to add those to the input dialect, so I pulled part of that patch forward. * Since these now newly lack lowerings, I was expecting to hit some test failures that would guide me to where to adapt, but this appears to be dead code (outside of integration tests?). We'll see what the CI says. * Can keep coding on this patch to adapt whatever is needed.
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.