| commit | 4b4b8546e7b8731d94dc32a61931baf51c7a9036 | [log] [tgz] |
|---|---|---|
| author | Ben Vanik <ben.vanik@gmail.com> | Thu Sep 01 10:13:14 2022 -0700 |
| committer | Ben Vanik <ben.vanik@gmail.com> | Thu Sep 01 12:14:40 2022 -0700 |
| tree | 969e89291798c402125e49148eacbe340a69ce19 | |
| parent | 289762197c246230ec4410dbccdde3e781de8790 [diff] |
Updating VMVX to use per-worker contexts and support workgroup state. Previously one context was shared for all workers and that made it impossible to safely support rwdata inside the modules. Now that each worker has a dedicated context the VMVX module can have workgroup information directly stored on it during execution and available for use by anything VMVX uses to run work. This does have a memory cost when multithreading but it's on the order of ~128B/worker. This required plumbing the expected worker count through local executables but that's likely to be useful with other loaders in the future (wasm/etc).
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.