OVERCOMING COMMUNICATION LATENCY BARRIERS IN MASSIVELY PARALLEL SCIENTIFIC COMPUTATION

被引：5

作者：

Dror, Ron O.

Grossman, J. P. ^{[1
]}

Mackenzie, Kenneth M. ^{[2
]}

Towles, Brian ^{[3
]}

Chow, Edmond ^{[4
]}

Salmon, John K. ^{[5
]}

Young, Cliff ^{[6
]}

Bank, Joseph A. ^{[6
]}

Batson, Brannon ^{[7
]}

Deneroff, Martin M. ^{[8
]}

Kuskin, Jeffrey S. ^{[7
]}

Larson, Richard H. ^{[7
]}

Moraes, Mark A. ^{[9
]}

Shaw, David E. ^{[1
]}

机构：

[1] DE Shaw Res, Architecture Design Simulat & Implementat Anton, New York, NY 10036 USA

[2] DE Shaw Res, Architectural Simulat & Software Anton, New York, NY 10036 USA

[3] DE Shaw Res, Architecture Design & Implementat Anton, New York, NY 10036 USA

[4] DE Shaw Res, Software Anton, New York, NY 10036 USA

[5] DE Shaw Res, Algorithm Design Software Dev & Comp Architecture, New York, NY 10036 USA

[6] DE Shaw Res, Architectural Simulat Architecture Design & Softw, New York, NY 10036 USA

[7] DE Shaw Res, Hardware Architecture Design & Implementat Anton, New York, NY 10036 USA

[8] DE Shaw Res, Hardware Design & Engn Team, New York, NY 10036 USA

[9] DE Shaw Res, Anton Software Dev & Syst Teams, New York, NY 10036 USA

来源：

IEEE MICRO | 2011年 / 31卷 / 03期

关键词：

Anton; Data communications; interprocessor communications; multiprocessor systems; network communication; parallel systems; special-purpose hardware;

D O I：

10.1109/MM.2011.38

中图分类号：

TP3 [计算技术、计算机技术];

学科分类号：

0812 ;

摘要：

Anton, a massively parallel special-purpose machine that accelerates molecular dynamics simulations by orders of magnitude, uses a combination of specialized hardware mechanisms and restructured software algorithms to reduce and hide communication latency. Anton delivers end-to-end internode latency significantly lower than any other large-scale parallel machine, and its critical-path communication time for molecular dynamics simulations is less than 3 percent that of the next-fastest platform.

引用

页码：8 / 19

页数：12

共 50 条

[1] An efficient communication strategy for massively parallel computation in CFD
Wan, YunBo
He, Lei
Zhang, Yong
Zhao, Zhong
Liu, Jie
Zhang, HaoYuan
[J]. JOURNAL OF SUPERCOMPUTING, 2023, 79 (07): : 7560 - 7583
[2] An efficient communication strategy for massively parallel computation in CFD
YunBo Wan
Lei He
Yong Zhang
Zhong Zhao
Jie Liu
HaoYuan Zhang
[J]. The Journal of Supercomputing, 2023, 79 : 7560 - 7583
[3] Programming for communication: Overcoming motivational barriers to computation for all
Forte, A
[J]. 200S IEEE SYMPOSIUM ON HUMAN CENTRIC COMPUTING LANGUAGES AND ENVIRONMENTS, 2003, : 285 - 286
[4] HIDING GLOBAL COMMUNICATION LATENCY IN THE GMRES ALGORITHM ON MASSIVELY PARALLEL MACHINES
Ghysels, P.
Ashby, T. J.
Meerbergen, K.
Vanroose, W.
[J]. SIAM JOURNAL ON SCIENTIFIC COMPUTING, 2013, 35 (01): : C48 - C71
[5] Multi-threading: A new dimension to massively parallel scientific computation
Nielsen, IMB
Janssen, CL
[J]. COMPUTER PHYSICS COMMUNICATIONS, 2000, 128 (1-2) : 238 - 244
[6] On the Hardness of Massively Parallel Computation
Chung, Kai-Min
Ho, Kuan-Yi
Sun, Xiaorui
[J]. PROCEEDINGS OF THE 32ND ACM SYMPOSIUM ON PARALLELISM IN ALGORITHMS AND ARCHITECTURES (SPAA '20), 2020, : 153 - 162
[7] Overcoming communication barriers
Sienk, Thorsten
[J]. Konstruktion, 2023, 75 (09): : 29 - 31
[8] EVALUATING ARRAY EXPRESSIONS ON MASSIVELY-PARALLEL MACHINES WITH COMMUNICATION/COMPUTATION OVERLAP
BOUCHITTE, V
BOULET, P
DARTE, A
ROBERT, Y
[J]. INTERNATIONAL JOURNAL OF SUPERCOMPUTER APPLICATIONS AND HIGH PERFORMANCE COMPUTING, 1995, 9 (03): : 205 - 219
[9] Massively Parallel Computation in a Heterogeneous Regime
Fischer, Orr
Horowitz, Adi
Oshman, Rotem
[J]. PROCEEDINGS OF THE 2022 ACM SYMPOSIUM ON PRINCIPLES OF DISTRIBUTED COMPUTING, PODC 2022, 2022, : 345 - 355
[10] PARALLEL SCIENTIFIC COMPUTATION
HILLIS, WD
BOGHOSIAN, BM
[J]. SCIENCE, 1993, 261 (5123) : 856 - 863

← 1 2 3 4 5 →