TY - GEN
T1 - QCDOC
T2 - 2004 ACM/IEEE Conference on Supercomputing, SC 2004
AU - Boyle, P. A.
AU - Chen, D.
AU - Christ, N. H.
AU - Clark, M.
AU - Cohen, S.
AU - Dong, Zhihua
AU - Gara, A.
AU - Joo, Balint
AU - Jung, Chulwoo
AU - Levkova, L.
AU - Liao, Xiaodong
AU - Liu, Guofeng
AU - Mawhinney, R. D.
AU - Ohta, S.
AU - Petrov, K.
AU - Wettig, T.
AU - Yamaguchi, A.
AU - Cristian, C.
N1 - Publisher Copyright:
© 2004 IEEE.
PY - 2004
Y1 - 2004
N2 - Numerical simulations of the strong nuclear force, known as quantum chromodynamics or QCD, have proven to be a demanding, forefront problem in high-performance computing. In this report, we describe a new computer, QCDOC (QCD On a Chip), designed for optimal price/performance in the study of QCD. QCDOC uses a six-dimensional, low-latency mesh network to connect processing nodes, each of which includes a single custom ASIC, designed by our collaboration and built by IBM, plus DDR SDRAM. Each node has a peak speed of 1Gigaflops and two 12,288node, 10+ Teraflops machines are to be completed in the fall of 2004. Currently, a 512 node machine is running, delivering efficiencies as high as 45% of peak on the conjugate gradient solvers that dominate our calculations and a 4096-node machine with a cost of $1.6M is under construction. This should give us a price/performance less than $1per sustained Megaflops.
AB - Numerical simulations of the strong nuclear force, known as quantum chromodynamics or QCD, have proven to be a demanding, forefront problem in high-performance computing. In this report, we describe a new computer, QCDOC (QCD On a Chip), designed for optimal price/performance in the study of QCD. QCDOC uses a six-dimensional, low-latency mesh network to connect processing nodes, each of which includes a single custom ASIC, designed by our collaboration and built by IBM, plus DDR SDRAM. Each node has a peak speed of 1Gigaflops and two 12,288node, 10+ Teraflops machines are to be completed in the fall of 2004. Currently, a 512 node machine is running, delivering efficiencies as high as 45% of peak on the conjugate gradient solvers that dominate our calculations and a 4096-node machine with a cost of $1.6M is under construction. This should give us a price/performance less than $1per sustained Megaflops.
UR - http://www.scopus.com/inward/record.url?scp=84934279745&partnerID=8YFLogxK
U2 - 10.1109/SC.2004.46
DO - 10.1109/SC.2004.46
M3 - Conference contribution
AN - SCOPUS:84934279745
T3 - Proceedings of the ACM/IEEE SC 2004 Conference: Bridging Communities
BT - Proceedings of the ACM/IEEE SC 2004 Conference
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 6 November 2004 through 12 November 2004
ER -