Impact of number of angles on the performance of the data vortex optical interconnection network

被引:13
作者
Hawkins, Cory [1 ]
Wills, D. Scott [1 ]
机构
[1] Georgia Inst Technol, Sch Elect & Comp Engn, Atlanta, GA 30332 USA
关键词
data vortex; optical packet switching; photonic packet switch; supercomputers;
D O I
10.1109/JLT.2006.879508
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
Reducing communication latency in multiprocessor interconnection networks can increase system performance on a broad range of applications. The data vortex photonic network reduces message latency by utilizing all-optical end-to-end transparent links and deflection routing. Cylinders replace node storage for buffering messages. The cylinder circumference (measured as number of angles) has a significant impact on the message acceptance rate and average message latency. A new symmetric mode of usage for the data vortex is discussed in which a fraction of the angles is used for input/output (I/O), and the remainder is used for "virtual buffering" of messages. For single-angle injection, six total angles provide the best performance. Likewise, the same ratio of 5: 1 purely routing nodes versus I/O nodes is shown to produce greater than 99% acceptance, under normal loading conditions for all other network sizes studied. It is shown that for a given network I/O size, a shorter height and wider circumference data vortex organization provides acceptable latency with fewer total nodes than a taller but narrower data vortex. The performance versus system cost is discussed and evaluated, and the 5:1 noninjection-to-injection angle ratio is shown to be cost effective when constructing a system in current optical technology.
引用
收藏
页码:3288 / 3294
页数:7
相关论文
共 15 条
[1]  
CHAN S, 1994, P IEEE GLOBECOM 94 S, V3, P1935
[2]   How much does network contention affect distributed shared memory performance? [J].
Dai, D ;
Panda, DK .
PROCEEDINGS OF THE 1997 INTERNATIONAL CONFERENCE ON PARALLEL PROCESSING, 1997, :454-461
[3]  
HAWKINS C, IN PRESS IEEE T PARA
[4]   Physical layer scalability of WDM optical packet interconnection networks [J].
Liboiron-Ladouceur, O ;
Small, BA ;
Bergman, K .
JOURNAL OF LIGHTWAVE TECHNOLOGY, 2006, 24 (01) :262-270
[5]   Scowl: A tool for characterization of parallel workload and its use on SPLASH-2 application suite [J].
Marinov, D ;
Magdic, D ;
Milenkovic, A ;
Protic, J ;
Tartalja, I ;
Milutinovic, V .
8TH INTERNATIONAL SYMPOSIUM ON MODELING, ANALYSIS AND SIMULATION OF COMPUTER AND TELECOMMUNICATION SYSTEMS, PROCEEDINGS, 2000, :207-213
[6]   Optical switching: Switch fabrics, techniques, and architectures [J].
Papadimitriou, GI ;
Papazoglou, C ;
Pomportsis, AS .
JOURNAL OF LIGHTWAVE TECHNOLOGY, 2003, 21 (02) :384-405
[7]   Characterizing parallel workloads to reduce multiple writer overhead in Shared Virtual Memory systems [J].
Petit, S ;
Sahuquillo, J ;
Pont, A .
10TH EUROMICRO WORKSHOP ON PARALLEL, DISTRIBUTED AND NETWORK-BASED PROCESSING, PROCEEDINGS, 2002, :261-268
[8]  
REED CS, 1999, Patent No. 5996020
[9]   A fully implemented 12x12 data vortex optical packet switching interconnection network [J].
Shacham, A ;
Small, BA ;
Liboiron-Ladouceur, O ;
Bergman, K .
JOURNAL OF LIGHTWAVE TECHNOLOGY, 2005, 23 (10) :3066-3075
[10]   Slot timing considerations in optical packet switching networks [J].
Small, BA ;
Bergman, K .
IEEE PHOTONICS TECHNOLOGY LETTERS, 2005, 17 (11) :2478-2480