| commit | 8e86156cd647f4afd6412c41e081ebf9518c5927 | [log] [tgz] |
|---|---|---|
| author | Stanley Winata <68087699+raikonenfnu@users.noreply.github.com> | Fri Apr 19 13:09:43 2024 -0700 |
| committer | GitHub <noreply@github.com> | Fri Apr 19 13:09:43 2024 -0700 |
| tree | b8e760b636e10e3fd2a14eddaef54344c9640450 | |
| parent | e87ff178a9a7b210ec2c0e6726024adb0b0124e6 [diff] |
[LLVMGPU] Modify layouts to be able to handle dequant operation. (#17113) This PR solves 2 issues to enable dequant-gemm to work on vector distribution: 1. During dequant-gemm, we will have reads of scale and zero-point that will feed into a broadcast operation which will generate a vector layout conflict operation. To resolve this, we do not anchor on these transfer_read and depend on the layout from the transfer_read of quants. 2. We have not applied permutation on the subgroup and thread active IDs which resulted in numerical issues during dequant-gemm. This PR applies them and added a test for them.
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.