Communication studies of single-threaded and multithreaded distributed-memory machines

Andrew Sohn; Yunheung Paek; Jui Yuan Ku; Yuetsu Kodama; Yoshinori Yamaguchi

doi:10.1109/hpca.1999.744384

Communication studies of single-threaded and multithreaded distributed-memory machines

Andrew Sohn, Yunheung Paek, Jui Yuan Ku, Yuetsu Kodama, Yoshinori Yamaguchi

Computer Science

Research output: Contribution to conference › Paper › peer-review

1 Scopus citations

Abstract

This report explicates the communication overlapping capabilities of three distributed-memory machines, SGI/Cray T3E, IBM SP-2 with wide nodes, and the ETL EM-X. Bitonic sorting and Fast Fourier Transform are selected for experiments. Various message sizes are used to determine when, where, how much and why the overlapping takes place. Experimental results with up to 64 processors indicated that the communication performance of EM-X is insensitive to various message sizes while SP-2 is the most sensitive. T3E stayed in between. The EM-X gave the highest communication overlapping capability while T3E did the lowest. The experimental results are compared with the analytical results based on LogP and LogGP communication models.

Original language	English (US)
Pages	310-314
Number of pages	5
DOIs	https://doi.org/10.1109/hpca.1999.744384
State	Published - 1999
Event	Proceedings of the 1999 5th International Symposium on High-Performance Computer Architecture, HPCA - Orlando, FL, USA Duration: Jan 9 1999 → Jan 13 1999

Other

Other	Proceedings of the 1999 5th International Symposium on High-Performance Computer Architecture, HPCA
City	Orlando, FL, USA
Period	1/9/99 → 1/13/99

All Science Journal Classification (ASJC) codes

Hardware and Architecture

Access to Document

10.1109/hpca.1999.744384

Cite this

@conference{077706b2c9584fd7bc83f4c3ad85a2bd,

title = "Communication studies of single-threaded and multithreaded distributed-memory machines",

abstract = "This report explicates the communication overlapping capabilities of three distributed-memory machines, SGI/Cray T3E, IBM SP-2 with wide nodes, and the ETL EM-X. Bitonic sorting and Fast Fourier Transform are selected for experiments. Various message sizes are used to determine when, where, how much and why the overlapping takes place. Experimental results with up to 64 processors indicated that the communication performance of EM-X is insensitive to various message sizes while SP-2 is the most sensitive. T3E stayed in between. The EM-X gave the highest communication overlapping capability while T3E did the lowest. The experimental results are compared with the analytical results based on LogP and LogGP communication models.",

author = "Andrew Sohn and Yunheung Paek and Ku, {Jui Yuan} and Yuetsu Kodama and Yoshinori Yamaguchi",

year = "1999",

doi = "10.1109/hpca.1999.744384",

language = "English (US)",

pages = "310--314",

note = "Proceedings of the 1999 5th International Symposium on High-Performance Computer Architecture, HPCA ; Conference date: 09-01-1999 Through 13-01-1999",

}

TY - CONF

T1 - Communication studies of single-threaded and multithreaded distributed-memory machines

AU - Sohn, Andrew

AU - Paek, Yunheung

AU - Ku, Jui Yuan

AU - Kodama, Yuetsu

AU - Yamaguchi, Yoshinori

PY - 1999

Y1 - 1999

N2 - This report explicates the communication overlapping capabilities of three distributed-memory machines, SGI/Cray T3E, IBM SP-2 with wide nodes, and the ETL EM-X. Bitonic sorting and Fast Fourier Transform are selected for experiments. Various message sizes are used to determine when, where, how much and why the overlapping takes place. Experimental results with up to 64 processors indicated that the communication performance of EM-X is insensitive to various message sizes while SP-2 is the most sensitive. T3E stayed in between. The EM-X gave the highest communication overlapping capability while T3E did the lowest. The experimental results are compared with the analytical results based on LogP and LogGP communication models.

AB - This report explicates the communication overlapping capabilities of three distributed-memory machines, SGI/Cray T3E, IBM SP-2 with wide nodes, and the ETL EM-X. Bitonic sorting and Fast Fourier Transform are selected for experiments. Various message sizes are used to determine when, where, how much and why the overlapping takes place. Experimental results with up to 64 processors indicated that the communication performance of EM-X is insensitive to various message sizes while SP-2 is the most sensitive. T3E stayed in between. The EM-X gave the highest communication overlapping capability while T3E did the lowest. The experimental results are compared with the analytical results based on LogP and LogGP communication models.

UR - http://www.scopus.com/inward/record.url?scp=0032786295&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0032786295&partnerID=8YFLogxK

U2 - 10.1109/hpca.1999.744384

DO - 10.1109/hpca.1999.744384

M3 - Paper

AN - SCOPUS:0032786295

SP - 310

EP - 314

T2 - Proceedings of the 1999 5th International Symposium on High-Performance Computer Architecture, HPCA

Y2 - 9 January 1999 through 13 January 1999

ER -

Communication studies of single-threaded and multithreaded distributed-memory machines

Abstract

Other

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this