| commit | c15b610542b538f468d7249a645be16a6def3c18 | [log] [tgz] |
|---|---|---|
| author | Simon Camphausen <simon.camphausen@iml.fraunhofer.de> | Fri Feb 23 06:01:18 2024 +0100 |
| committer | GitHub <noreply@github.com> | Thu Feb 22 21:01:18 2024 -0800 |
| tree | ef6ffab4f75df46a27246b72dd85344c242813ca | |
| parent | 1d7fb8e1eb696b1cf450b62496208bec170263a5 [diff] |
[EmitC] Remove the forked emitter and generate all the code in the conversion pass (#16357) This resolves the longstanding painpoint of having a custom fork of the EmitC dialect to C emitter infrastructure in IREE. This is done by generating `emitc.verbatim` ops for the missing pieces in the conversion. Things that need to be addressed: - Move `createModuleStructure` into a separate file. Skipped for now as it introduces hundreds lines of diff noise - The creation of the descriptor arrays are a 1:1 copy from the old CModuleTarget. Helper structs/functions around these would help readability. - Convert the `vm.module` to a `builtin.module` in the conversion. (Would need to make the conversion pass run on the outer `builtin.module` and nest the pass manager correctly, I think?) - The `FuncAnalysis` became kind of a sink where different pieces of information are mashed together with subtle bugs around uninitialized variables from the different constructos. --------- Co-authored-by: Marius Brehler <marius.brehler@iml.fraunhofer.de>
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.