Title :
Optimised global reduction on QsNetII
Author :
Roweth, Duncan ; Pittman, Ashley
Author_Institution :
Quadrics Ltd., Bristol, UK
Abstract :
In this paper we describe how QsNetII supports reduction, a key collective for massively parallel applications. Results from jobs run on a 512-node quad CPU cluster show excellent scaling, with the average time to execute a 2048 process global sum being 22 microsecs.
Keywords :
microprocessor chips; parallel architectures; reduced instruction set computing; workstation clusters; 2048 processor; QsNet; cluster; optimisation; parallel application; Bandwidth; Broadcasting; Communication switching; Communication system control; Delay; Engines; Hardware; Protocols; SDRAM; Switches;
Conference_Titel :
High Performance Interconnects, 2005. Proceedings. 13th Symposium on
Print_ISBN :
0-7695-2449-4
DOI :
10.1109/CONECT.2005.28