commit | e792c3206cfb12323a689c53dd2cb16d8c342e72 | [log] [tgz] |
---|---|---|
author | Ben Vanik <ben.vanik@gmail.com> | Wed Jul 31 16:09:40 2024 -0700 |
committer | GitHub <noreply@github.com> | Wed Jul 31 23:09:40 2024 +0000 |
tree | d6997edf25293e9e2651daa1ebc8781a4f8dac86 | |
parent | d8d140764c40c328537e374e9143ca6bab8f5a14 [diff] |
Improving global folding and IPO for immutable globals. (#18066) Parameters and constants end up as globals that previously were not getting marked as immutable unless the globals were default initialized. Now the GlobalTable tracks whether a global is exclusively stored within initializers (or functions only called from initializers) in order to mark them as immutable. IPO was updated to support propagating uniform immutable global loads across call edges as if they were constants (as they effectively are just constants stored on the global scope). Required for #17875 (to avoid treating constants/parameters as dynamic binding table values).
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.
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.