| commit | fe30c102c47bd02799b8b289286fe10401e3e31e | [log] [tgz] |
|---|---|---|
| author | MaheshRavishankar <1663364+MaheshRavishankar@users.noreply.github.com> | Fri Jan 06 12:43:46 2023 -0800 |
| committer | GitHub <noreply@github.com> | Fri Jan 06 12:43:46 2023 -0800 |
| tree | 9dd6988b81559775699b18aa2314ff2d7bb75ab5 | |
| parent | 1067b57bef5c7883bd19e8ee6a67079fac33c5dc [diff] |
Modify aggressive fusion heuristics to not create ops where reduction ops also return intermediate results. (#11740) Currently under aggressive fusion, elementwise parallel ops are fused with consumer that are reductions. When the parallel op has other uses the fused op also returns the reduced and non-reduced values. This results in operations that are reductions returning results that have shape similar to the entire iteration space of the op. Such ops are hard to handle in the backend. This patch avoids creating such ops, but adapts the fusion heuristics to still pull these ops into the same dispatch.
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.