| commit | 09c9f5b417640d42b86f668a27e95c816753adf2 | [log] [tgz] |
|---|---|---|
| author | Lei Zhang <antiagainst@gmail.com> | Wed Dec 20 15:46:55 2023 -0800 |
| committer | GitHub <noreply@github.com> | Wed Dec 20 15:46:55 2023 -0800 |
| tree | 8ae5fe62ac293a89284e2176a65c3d3276d545db | |
| parent | 5b8e870b14bdd88615dfc34738b5ee9d02781cc7 [diff] |
[spirv] Provide same entry point set in variants when linking (#15935) Note that at runtime, for a particular executable, only one variant of it will be loaded. So, all variants of an executable are expected to provide the exact same set of entry points; this way we can guarantee no matter which variant is chosen, we have all entry points to call into. The same entry point in different variants may have different target requirements though. The input to the linking stage are a collection of executables, each may have multiple variants, but only ever provide one entry point. Together with the above restriction, we can link two executables if and only if their variants have the exact same set of target requirements. Under such circumstances, we can make sure for a particular target requirement (loaded as one variant during runtime), we can provide all entry points.
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.