Matrix multiplication performance on current PCs

Bibliographic Information

Other Title
  • パソコンの行列演算性能の評価

Search this article

Description

We have measured the time of matrix multiply and BLAS1 routines on three personal computers, and compared the performance. The memory bandwidth is improved much in the last year, however, it is still important to use registers and cache memory effectively to reduce memory access latency. So, the research to improve the compilers for generating fast code for the hierarchical memory systems should be encouraged.

Journal

  • IPSJ SIG Notes

    IPSJ SIG Notes 64 7-12, 1996-12-12

    Information Processing Society of Japan (IPSJ)

Citations (1)*help

See more

References(8)*help

See more

Details 詳細情報について

  • CRID
    1572261552107517568
  • NII Article ID
    110002932223
  • NII Book ID
    AN10463942
  • Text Lang
    ja
  • Data Source
    • CiNii Articles

Report a problem

Back to top