Research Article

Emmerald: a fast matrix–matrix multiply using Intel's SSE instructions

  1. Douglas Aberdeen*,
  2. Jonathan Baxter

Article first published online: 27 FEB 2001

DOI: 10.1002/cpe.549

Issue

Concurrency and Computation: Practice and Experience

Concurrency and Computation: Practice and Experience

Volume 13, Issue 2, pages 103–119, February 2001

Additional Information

How to Cite

Aberdeen, D. and Baxter, J. (2001), Emmerald: a fast matrix–matrix multiply using Intel's SSE instructions. Concurrency and Computation: Practice and Experience, 13: 103–119. doi: 10.1002/cpe.549

Author Information

  1. Research School of Information Sciences and Engineering, Australian National University, Canberra, Australia

*Research School of Information Sciences and Engineering, Australian National University, Canberra, Australia

Publication History

  1. Issue published online: 27 FEB 2001
  2. Article first published online: 27 FEB 2001
  3. Manuscript Revised: 27 JUL 2000
  4. Manuscript Received:

Funded by

  • Australian Research Council

SEARCH

SEARCH BY CITATION

ARTICLE TOOLS

Get PDF (272K)

Keywords:

  • GEMM;
  • SIMD;
  • SSE;
  • matrix multiply;
  • deep memory hierarchy

Abstract

Generalized matrix–matrix multiplication forms the kernel of many mathematical algorithms, hence a faster matrix–matrix multiply immediately benefits these algorithms. In this paper we implement efficient matrix multiplication for large matrices using the Intel Pentium single instruction multiple data (SIMD) floating point architecture. The main difficulty with the Pentium and other commodity processors is the need to efficiently utilize the cache hierarchy, particularly given the growing gap between main-memory and CPU clock speeds. We give a detailed description of the register allocation, Level 1 and Level 2 cache blocking strategies that yield the best performance for the Pentium III family. Our results demonstrate an average performance of 2.09 times faster than the leading public domain matrix–matrix multiply routines and comparable performance with Intel's SIMD small matrix–matrix multiply routines. Copyright © 2001 John Wiley & Sons, Ltd.

Get PDF (272K)