Simplify how mmt4d ukernels deal with the K=0 case. (#16137) The case of K=0 is vacuous when accumulating, so we just early-return. This doesn't change. When not accumulating, it amounts to clearing the accumulator. We used to have dedicated code for that in `mmt4d.c`. That then allowed our ukernels tile-functions to `IREE_UK_ASSUME(K >= 1);`. That makes the asm a little simpler. However, that only really mattered when we thought we would be hand-writing that assembly. Now that we're not, it doesn't really matter, and the slightly smaller tile function code size is outweighed by the additional code in `mmt4d.c`. Overall, this PR drops 40 lines of source code and makes bitcode 2% smaller.
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.