| commit | 8dc68201d674088a019f133d75a9968d89dc7060 | [log] [tgz] |
|---|---|---|
| author | Ben Vanik <ben.vanik@gmail.com> | Mon Aug 12 13:36:40 2024 -0700 |
| committer | GitHub <noreply@github.com> | Mon Aug 12 13:36:40 2024 -0700 |
| tree | 7757663fa2c0af7176780fee7a20d3e6e9d93b1d | |
| parent | 76eb9c18f812db75bbe525b3de3bf478eaa05693 [diff] |
Adding simplified HAL dispatch methods. (#18189) These combine push constants and push descriptor sets into the dispatch calls as in practice we have a near 1:1 relationship anyway. Pipeline layouts are still used in HAL interfaces to allow the compiler to map the information but are otherwise not used by the new ops. The `--iree-hal-experimental-dispatch2` flag enables emitting the new ops. Since executables no longer require pipeline layouts in this simplified model the `--iree-hal-experimental-executable-create2` flag can be used to stop passing them; targets that support dispatch2 will ignore them if provided. Future changes will start to add support on targets for the simplified bindings and then remove the existing pipeline layout-based binding model as a breaking ABI change. Current target status: * [x] Local/CPU: executable-create2 and executable-dispatch2 supported (backward compat) * [x] CUDA: executable-dispatch2 supported (backward compat) * [x] HIP: executable-dispatch2 supported (backward compat) * [x] Metal: executable-dispatch2 supported (backward compat) * [x] Vulkan: executable-dispatch2 supported (backward compat) * [x] WebGPU: executable-dispatch2 supported (backward compat) Reworking the CUDA/HIP/Metal/Vulkan/WebGPU flatbuffers to support executable-create2 will be done in a follow-up. Progress on #18154.
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
| Package | Release status |
|---|---|
| GitHub release (stable) |  |
| GitHub release (nightly) |  |
| Python iree-compiler |  |
| Python iree-runtime |  |
| Host platform | Build status |
|---|---|
| Linux |   |
| macOS |  |
| Windows |  |
For the full list of workflows see https://iree.dev/developers/general/github-actions/.
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.