| commit | 50b55e35c8b3a7a948926d30abcf85e9117e2811 | [log] [tgz] |
|---|---|---|
| author | Lei Zhang <antiagainst@google.com> | Thu Jun 01 17:08:30 2023 -0400 |
| committer | GitHub <noreply@github.com> | Thu Jun 01 21:08:30 2023 +0000 |
| tree | e4d3c786e1a60ee490cf2be66e96fe734ca96d58 | |
| parent | 0d81062a8ffabf3561e9d6cc758b8250b2f607d3 [diff] |
[gpu] Enable fusing input producers after tiling reduction loops (#13806) This commit enables fusing input producers into the serial loop after tiling the matmul K dimension. This enables using the GPU shared memory promoted for A/B matrix slices for weight dequantization, without introducing additional shared memory allocations during bufferization. This is to support int4 weight quantized matmuls in LLMs.
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.