commit | b6c8f97571caacdbba74c9d90bb25b26fa537176 | [log] [tgz] |
---|---|---|
author | MaheshRavishankar <1663364+MaheshRavishankar@users.noreply.github.com> | Fri Mar 25 16:54:36 2022 -0700 |
committer | GitHub <noreply@github.com> | Fri Mar 25 16:54:36 2022 -0700 |
tree | 3004020a624f0dfce7d80672b5f2fd2a7b8415ad | |
parent | 8b91334edb53a58e15fc986ee6af310794dd8020 [diff] |
Add a pass to bufferize early copy only dispatches. (#8648) For cases where the dispatches are copy only , i.e. data is transfered from one interface binding to another, bufferizing early results in a linalg.generic (i.e. a copy) operation in the dispatch. The backends can use this to generate code. This is the first step in dropping TiledOpInterface for tensor.insert_slice and tensor.extract_slice operations. This commit also adds patterns to fold these operations with the flow.dispatch.tensor.load and flow.dispatch.tensor.store operations. Eventually these patterns will be moved to canonicalizations when the TiledOpInterface implementation for tensor.insert_slice and tensor.extract_slice are dropped. Fixes #8509 Resubmit of PR #8529
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.