| commit | 5180ede27567f34a83bb223414d846fdfcb948c3 | [log] [tgz] |
|---|---|---|
| author | Benoit Jacob <jacob.benoit.1@gmail.com> | Fri Mar 01 17:23:42 2024 -0500 |
| committer | GitHub <noreply@github.com> | Fri Mar 01 22:23:42 2024 +0000 |
| tree | 10f85ebb7054bdcb3f57f22501ca05624be95eb3 | |
| parent | 62da59b14cc2cd087bfd4def980925a424061625 [diff] |
mmt4d ukernel: simplification in generic tile funcs: stop using a stack array (#16633) In the mmt4d ukernel, the generic fallback tile-functions have used a stack array to represent the accumulator in an attempt to mimic what a real optimized tile-func would do with registers. That had an initial purpose of providing some starting point to help guide the design of optimized tile-funcs, but by now there are enough optimized tile-funcs that the existing ones are the starting point for new ones. And that led to more lines of code per generic tile-funcs, and some complication with the sizing of the stack arrays, requiring a lengthy comment and some validation code.
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.