commit | 61b2bb2082557cf40613ba6a468ca75f3109204e | [log] [tgz] |
---|---|---|
author | MaheshRavishankar <1663364+MaheshRavishankar@users.noreply.github.com> | Sun Apr 17 20:39:04 2022 -0700 |
committer | GitHub <noreply@github.com> | Sun Apr 17 20:39:04 2022 -0700 |
tree | 6b7b490fbbfc13f6125a4bdd3b07c75d8e865cd3 | |
parent | 24baca2c7fa466ca3845244eb0632b0b25043ab6 [diff] |
Change elementwise op fusion heuristics. (#8723) Current fusion heuristics seems to have degraded over time. With the ops moved to different dialects, and changes to op semantics, the control functions used seem to not really capture the original intent. This PR revisits the control functions used for elementwise operation fusion. Fusion of elementwise operations with reshapes by expansion. This change also pulls in the fusion be collapse to clean up some additional reshapes and replaces some of the one-off patterns that were intending to achieve a similar effect. See #8724 and discourse.llvm.org/t/rfc-next-iteration-of-fusion-of-elementwise-operations/59955/4 for discussion of impact of this change.
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.