Neural network classifiers execution on superscalar microprocessors

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

This paper evaluates the contribution of various microprocessor architectural features on the execution of 4 neural networks used for classification problems. In this study, we selected the grnn, pnn, mnn and rbfn networks trained for the Iris data set and simulated with 10,000 elements datasets. Using a superscalar simulator we evaluated various architectural parameters such as IPC, memory hierarchy, branch prediction, functional units configuration. The main contribution of this work is to show that neural network workloads deserve their own characterization which cannot be derived from SPEC95 characteristics.

Original languageEnglish
Title of host publicationHigh Performance Computing - 2nd International Symposium, ISHPC 1999, Proceedings
EditorsKazuki Joe, Akira Fukuda, Constantine Polychronopoulos, Shinji Tomita
PublisherSpringer Verlag
Pages41-54
Number of pages14
ISBN (Print)3540659692, 9783540659693
DOIs
Publication statusPublished - 1 Jan 1999
Externally publishedYes
Event2nd International Symposium on High Performance Computing, ISHPC 1999 - Kyoto, Japan
Duration: 26 May 199928 May 1999

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume1615
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference2nd International Symposium on High Performance Computing, ISHPC 1999
Country/TerritoryJapan
CityKyoto
Period26/05/9928/05/99

Keywords

  • Classification
  • Microarchitecture
  • Neural network
  • Performance
  • Superscalar

Fingerprint

Dive into the research topics of 'Neural network classifiers execution on superscalar microprocessors'. Together they form a unique fingerprint.

Cite this