| commit | 380e15479889b2fc6287690150c8a7fbc8aeb989 | [log] [tgz] |
|---|---|---|
| author | Nicolas Vasilache <nicolasvasilache@users.noreply.github.com> | Tue Apr 05 22:58:58 2022 +0200 |
| committer | GitHub <noreply@github.com> | Tue Apr 05 22:58:58 2022 +0200 |
| tree | 91078ff4ef338a84779e608357c3733ef504e066 | |
| parent | fde35efe0c20bbc788518086ee329a5dd1255731 [diff] |
Forward tensor.insert_slice coming from in_parallel lowering to flow.… (#8757) * Forward tensor.insert_slice coming from in_parallel lowering to flow.dispatch.tensor_store This pattern is currently necessary for correctness, it accounts for the fact that InParallel is distributed across multiple workgroups when lowering to HAL but the current implementation connects it to a sequential tensor.insert_slice and only later to flow.dispatch.tensor_store. In the future, all the rewrites in this file this should be done as part of InParallel -> HAL rewrite. But because of dialect dependencies and layering, we have some phase ordering that prevents it atm. A similar layering issue prevents bufferization to be controlled by the transform dialect atm. The effects of the insert_slice -> tensor_store forwarding are best observed before bufferization removes the tensor_store op. To allow more modular testing and separation of concerns, a temporary test-only flag is added to disable bufferization and exhibit the proper forwarding behavior. Finally, a previously failing integration test is now correct. * Refactor and reuse the foldOffsetsSizesAndStrides helper function
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!
| CI System | Build System | Platform | Architecture | Configuration / Component | Status |
|---|---|---|---|---|---|
| Kokoro | Bazel | Linux | x86-64 |  | |
| Kokoro | CMake & Bazel | Linux | x86-64 (swiftshader) | Integrations |  |
| Kokoro | CMake & Bazel | Linux | x86-64 (turing) | Integrations |  |
| Kokoro | CMake | Linux | x86-64 (swiftshader) |  | |
| Kokoro | CMake | Linux | x86-64 (swiftshader) | asan |  |
| Kokoro | CMake | Linux | x86-64 (turing) |  | |
| Kokoro | CMake | Android | arm64-v8a | Runtime (build only) |  |
| Kokoro | CMake | Bare Metal | risc-v-32 | Runtime |  |
| Kokoro | CMake | Linux | risc-v-64 | Runtime |  |
| Buildkite | CMake | Android | arm64-v8a | Runtime |  |
| BuildKite | CMake | Android | arm64-v8a | Runtime Benchmarks |  |
| BuildKite | CMake | Linux | x86-64 | Tracing + Standalone Runtime |  |
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.