The optimal architecture design of two-dimension matrix multiplication jumping systolic array

Yun Yang, Shinji Kimura

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

This paper proposes an efficient systolic array construction method for optimal planar systolic design of the matrix multiplication. By connection network adjustment among systolic array processing element (PE), the input/output data are jumping in the systolic array for multiplication operation requirements. Various 2-D systolic array topologies, such as square topology and hexagonal topology, have been studied to construct appropriate systolic array configuration and realize high performance matrix multiplication. Based on traditional Kung-Leiserson systolic architecture, the proposed "Jumping Systolic Array (JSA)" algorithm can increase the matrix multiplication speed with less processing elements and few data registers attachment. New systolic arrays, such as square jumping array, redundant dummy latency jumping hexagonal array, and compact parallel flow jumping hexagonal array, are also proposed to improve the concurrent system operation efficiency. Experimental results prove that the JSA algorithm can realize fully concurrent operation and dominate other systolic architectures in the specific systolic array system characteristics, such as band width, matrix complexity, or expansion capability.

Original languageEnglish
Pages (from-to)1101-1111
Number of pages11
JournalIEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences
VolumeE91-A
Issue number4
DOIs
Publication statusPublished - 2008

Fingerprint

Systolic Array
Systolic arrays
Matrix multiplication
Two Dimensions
Hexagon
Topology
Design
Architecture
Array processing
Parallel flow
Concurrent Systems
Latency
Concurrent
Adjustment
Multiplication
High Performance
Bandwidth
Configuration

Keywords

  • Fast process speed
  • Fully concurrent operation
  • Jumping systolic array (JSA)
  • Matrix multiplication
  • Processing element (PE)

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Computer Graphics and Computer-Aided Design
  • Applied Mathematics
  • Signal Processing

Cite this

@article{30794662f0664f98aa516c51430d8fc7,
title = "The optimal architecture design of two-dimension matrix multiplication jumping systolic array",
abstract = "This paper proposes an efficient systolic array construction method for optimal planar systolic design of the matrix multiplication. By connection network adjustment among systolic array processing element (PE), the input/output data are jumping in the systolic array for multiplication operation requirements. Various 2-D systolic array topologies, such as square topology and hexagonal topology, have been studied to construct appropriate systolic array configuration and realize high performance matrix multiplication. Based on traditional Kung-Leiserson systolic architecture, the proposed {"}Jumping Systolic Array (JSA){"} algorithm can increase the matrix multiplication speed with less processing elements and few data registers attachment. New systolic arrays, such as square jumping array, redundant dummy latency jumping hexagonal array, and compact parallel flow jumping hexagonal array, are also proposed to improve the concurrent system operation efficiency. Experimental results prove that the JSA algorithm can realize fully concurrent operation and dominate other systolic architectures in the specific systolic array system characteristics, such as band width, matrix complexity, or expansion capability.",
keywords = "Fast process speed, Fully concurrent operation, Jumping systolic array (JSA), Matrix multiplication, Processing element (PE)",
author = "Yun Yang and Shinji Kimura",
year = "2008",
doi = "10.1093/ietfec/e91-a.4.1101",
language = "English",
volume = "E91-A",
pages = "1101--1111",
journal = "IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences",
issn = "0916-8508",
publisher = "Maruzen Co., Ltd/Maruzen Kabushikikaisha",
number = "4",

}

TY - JOUR

T1 - The optimal architecture design of two-dimension matrix multiplication jumping systolic array

AU - Yang, Yun

AU - Kimura, Shinji

PY - 2008

Y1 - 2008

N2 - This paper proposes an efficient systolic array construction method for optimal planar systolic design of the matrix multiplication. By connection network adjustment among systolic array processing element (PE), the input/output data are jumping in the systolic array for multiplication operation requirements. Various 2-D systolic array topologies, such as square topology and hexagonal topology, have been studied to construct appropriate systolic array configuration and realize high performance matrix multiplication. Based on traditional Kung-Leiserson systolic architecture, the proposed "Jumping Systolic Array (JSA)" algorithm can increase the matrix multiplication speed with less processing elements and few data registers attachment. New systolic arrays, such as square jumping array, redundant dummy latency jumping hexagonal array, and compact parallel flow jumping hexagonal array, are also proposed to improve the concurrent system operation efficiency. Experimental results prove that the JSA algorithm can realize fully concurrent operation and dominate other systolic architectures in the specific systolic array system characteristics, such as band width, matrix complexity, or expansion capability.

AB - This paper proposes an efficient systolic array construction method for optimal planar systolic design of the matrix multiplication. By connection network adjustment among systolic array processing element (PE), the input/output data are jumping in the systolic array for multiplication operation requirements. Various 2-D systolic array topologies, such as square topology and hexagonal topology, have been studied to construct appropriate systolic array configuration and realize high performance matrix multiplication. Based on traditional Kung-Leiserson systolic architecture, the proposed "Jumping Systolic Array (JSA)" algorithm can increase the matrix multiplication speed with less processing elements and few data registers attachment. New systolic arrays, such as square jumping array, redundant dummy latency jumping hexagonal array, and compact parallel flow jumping hexagonal array, are also proposed to improve the concurrent system operation efficiency. Experimental results prove that the JSA algorithm can realize fully concurrent operation and dominate other systolic architectures in the specific systolic array system characteristics, such as band width, matrix complexity, or expansion capability.

KW - Fast process speed

KW - Fully concurrent operation

KW - Jumping systolic array (JSA)

KW - Matrix multiplication

KW - Processing element (PE)

UR - http://www.scopus.com/inward/record.url?scp=78049348338&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78049348338&partnerID=8YFLogxK

U2 - 10.1093/ietfec/e91-a.4.1101

DO - 10.1093/ietfec/e91-a.4.1101

M3 - Article

AN - SCOPUS:78049348338

VL - E91-A

SP - 1101

EP - 1111

JO - IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences

JF - IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences

SN - 0916-8508

IS - 4

ER -