You need JavaScript to view this

Architectural choices for the Columbia 0.8 Teraflops machine

Abstract

We discuss the hardware design choices made in our 16K-node 0.8 Teraflops supercomputer project, a machine architecture optimized for full QCD calculations. The efficiency of the conjugate gradient algorithm in terms of balance of floating-point operations, memory handling and utilization, and communication overhead is addressed. We also discuss the technological innovations and software tools that facilitate hardware design and what opportunities these give to the academic community. ((orig.)).
Authors:
Arsenin, I V [1] 
  1. Columbia Univ., New York, NY (United States). Dept. of Physics
Publication Date:
Apr 01, 1995
Product Type:
Journal Article
Report Number:
CONF-9409269-
Reference Number:
SCA: 990200; PA: AIX-26:064182; EDB-95:133380; SN: 95001458240
Resource Relation:
Journal Name: Nuclear Physics B, Proceedings Supplements; Journal Volume: 42; Conference: Lattice `94, Bielefeld (Germany), 25 Sep - 1 Oct 1994; Other Information: PBD: Apr 1995
Subject:
99 MATHEMATICS, COMPUTERS, INFORMATION SCIENCE, MANAGEMENT, LAW, MISCELLANEOUS; LATTICE FIELD THEORY; SUPERCOMPUTERS; QUANTUM CHROMODYNAMICS; COMPUTER ARCHITECTURE; ALGORITHMS; COMPUTER CALCULATIONS; DATA TRANSMISSION; DATA TRANSMISSION SYSTEMS; EFFICIENCY; MEMORY DEVICES; OPTIMIZATION; PARALLEL PROCESSING
OSTI ID:
100525
Country of Origin:
Netherlands
Language:
English
Other Identifying Numbers:
Journal ID: NPBSE7; ISSN 0920-5632; TRN: NL95FF364064182
Submitting Site:
NLN
Size:
pp. 902-904
Announcement Date:
Oct 04, 1995

Citation Formats

Arsenin, I V. Architectural choices for the Columbia 0.8 Teraflops machine. Netherlands: N. p., 1995. Web. doi:10.1016/0920-5632(95)00416-7.
Arsenin, I V. Architectural choices for the Columbia 0.8 Teraflops machine. Netherlands. doi:10.1016/0920-5632(95)00416-7.
Arsenin, I V. 1995. "Architectural choices for the Columbia 0.8 Teraflops machine." Netherlands. doi:10.1016/0920-5632(95)00416-7. https://www.osti.gov/servlets/purl/10.1016/0920-5632(95)00416-7.
@misc{etde_100525,
title = {Architectural choices for the Columbia 0.8 Teraflops machine}
author = {Arsenin, I V}
abstractNote = {We discuss the hardware design choices made in our 16K-node 0.8 Teraflops supercomputer project, a machine architecture optimized for full QCD calculations. The efficiency of the conjugate gradient algorithm in terms of balance of floating-point operations, memory handling and utilization, and communication overhead is addressed. We also discuss the technological innovations and software tools that facilitate hardware design and what opportunities these give to the academic community. ((orig.)).}
doi = {10.1016/0920-5632(95)00416-7}
journal = {Nuclear Physics B, Proceedings Supplements}
volume = {42}
journal type = {AC}
place = {Netherlands}
year = {1995}
month = {Apr}
}