[LLVMGPU] Allow reductions with dynamic parallel dimensions through WarpReduce (#14922) This mirrors what is currently done on the SPIR-V side where we naively fully distribute dynamic parallel dimensions to workgroups and use the subgroup reduce path. In the future we will want to use some mix of padding/specialization for cases like contractions where data is reused. Additionally, the pipeline matching logic between SPIR-V and LLVMGPU has nearly converged for subgroup reduction; unifying the two is left as TODO. This also disables fused leading elementwise for the contraction pipelines as this will currently fail to rematerialize the elementwise and overuse shared memory.
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.