English

Stream-K: Work-centric Parallel Decomposition for Dense Matrix-Matrix Multiplication on the GPU

Data Structures and Algorithms 2023-01-11 v1 Distributed, Parallel, and Cluster Computing

Abstract

We introduce Stream-K, a work-centric parallelization of matrix multiplication (GEMM) and related computations in dense linear algebra. Whereas contemporary decompositions are primarily tile-based, our method operates by partitioning an even share of the aggregate inner loop iterations among physical processing elements. This provides a near-perfect utilization of computing resources, regardless of how efficiently the output tiling for any given problem quantizes across the underlying processing elements. On GPU processors, our Stream-K parallelization of GEMM produces a peak speedup of up to 14×\times and 6.7×\times, and an average performance response that is both higher and more consistent across 32,824 GEMM problem geometries than state-of-the-art math libraries such as CUTLASS and cuBLAS. Furthermore, we achieve this performance from a single tile size configuration per floating-point precision, whereas today's math libraries employ complex kernel-selection heuristics to select from a large ensemble of kernel variants.

Keywords

Cite

@article{arxiv.2301.03598,
  title  = {Stream-K: Work-centric Parallel Decomposition for Dense Matrix-Matrix Multiplication on the GPU},
  author = {Muhammad Osama and Duane Merrill and Cris Cecka and Michael Garland and John D. Owens},
  journal= {arXiv preprint arXiv:2301.03598},
  year   = {2023}
}

Comments

This work previously appeared in the author's PhD dissertation, available at arXiv:2212.08964

R2 v1 2026-06-28T08:07:56.571Z