Presentation 2012-05-29
Scalability Analysis of Tightly-Coupled FPGA-Cluster for Lattice Boltzmann Computation
Yoshiaki KONO, Kentaro SANO, Ryotaro CHIBA, Satoru YAMAMOTO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) This paper presents an architecture and its performance model of an LBM accelerator to be implemented on a tightly-coupled FPGA cluster. Our tightly-coupled FPGA cluster has a ID ring for an accelerator-domain network (ADN) which allows FPGAs to send and receive data with low communication overhead. Our LBM accelerator is based on stream computation with scheduling approproate to ADN. We formulate a sustained-performance model of the accelerator. With the model, we show that the network bandwidth is much more important than the memory bandwidth. This result demonstrates the importance of ADN in the tightly-coupled FPGA cluster.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) high-performance computation / lattice boltzmann method / accelerator-domain network / tightly-coupled FPGA cluster / FPGA
Paper # RECONF2012-10
Date of Issue

Conference Information
Committee AI
Conference Date 2012/5/22(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Artificial Intelligence and Knowledge-Based Processing (AI)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Scalability Analysis of Tightly-Coupled FPGA-Cluster for Lattice Boltzmann Computation
Sub Title (in English)
Keyword(1) high-performance computation
Keyword(2) lattice boltzmann method
Keyword(3) accelerator-domain network
Keyword(4) tightly-coupled FPGA cluster
Keyword(5) FPGA
1st Author's Name Yoshiaki KONO
1st Author's Affiliation Graduate School of Information Sciences, Tohoku University()
2nd Author's Name Kentaro SANO
2nd Author's Affiliation Graduate School of Information Sciences, Tohoku University
3rd Author's Name Ryotaro CHIBA
3rd Author's Affiliation Graduate School of Information Sciences, Tohoku University
4th Author's Name Satoru YAMAMOTO
4th Author's Affiliation Graduate School of Information Sciences, Tohoku University
Date 2012-05-29
Paper # RECONF2012-10
Volume (vol) vol.112
Number (no) 70
Page pp.pp.-
#Pages 6
Date of Issue