English

SIMD$^2$: A Generalized Matrix Instruction Set for Accelerating Tensor Computation beyond GEMM

Hardware Architecture 2022-09-02 v3

Abstract

Matrix-multiplication units (MXUs) are now prevalent in every computing platform. The key attribute that makes MXUs so successful is the semiring structure, which allows tiling for both parallelism and data reuse. Nonetheless, matrix-multiplication is not the only algorithm with such attributes. We find that many algorithms share the same structure and differ in only the core operation; for example, using add-minimum instead of multiply-add. Algorithms with a semiring-like structure therefore have potential to be accelerated by a general-purpose matrix operation architecture, instead of common MXUs. In this paper, we propose SIMD2^2, a new programming paradigm to support generalized matrix operations with a semiring-like structure. SIMD2^2 instructions accelerate eight more types of matrix operations, in addition to matrix multiplications. Since SIMD2^2 instructions resemble a matrix-multiplication instruction, we are able to build SIMD2^2 architecture on top of any MXU architecture with minimal modifications. We developed a framework that emulates and validates SIMD2^2 using NVIDIA GPUs with Tensor Cores. Across 8 applications, SIMD2 provides up to 38.59×\times speedup and more than 10.63×\times on average over optimized CUDA programs, with only 5% of full-chip area overhead.

Keywords

Cite

@article{arxiv.2205.01252,
  title  = {SIMD$^2$: A Generalized Matrix Instruction Set for Accelerating Tensor Computation beyond GEMM},
  author = {Yunan Zhang and Po-An Tsai and Hung-Wei Tseng},
  journal= {arXiv preprint arXiv:2205.01252},
  year   = {2022}
}

Comments

To Appear in the 49th International Symposium on Computer Architecture (ISCA'22), June 18--22, 2022, New York, NY, USA

R2 v1 2026-06-24T11:05:26.190Z