| commit | e918678414c8f4b65aeaaa810fae6adcf480509e | [log] [tgz] |
|---|---|---|
| author | Han-Chung Wang <hanhan0912@gmail.com> | Tue Oct 17 17:09:46 2023 -0700 |
| committer | GitHub <noreply@github.com> | Tue Oct 17 17:09:46 2023 -0700 |
| tree | dee8cf7b1c56dc3d2b5f4e9c38fc88c7e6a896cf | |
| parent | 87a1cc6dfd25aaac5580a4e5e45d72b6c4bfd91a [diff] |
[LLVMGPU] Splitting TensorCoreVectorization to two passes. (#15184) It delecates the vectorization step to GenericVectorization pass; renames the pass to LLVMGPUTensorCorePreparation. It also swaps shared memory distribution stage with TensorCorePreparation. Because the generic vectorization converts all linalg ops to vector ops, while we apply specific logics to shared memory distribution. We now handle special cases, vectorize linalg.matmul; prepare the form for tensor core.
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.