Overlapping computation and communication in SMT clusters with commodity interconnects

Goumas, G; Anastopoulos, N; Koziris, N; Ioannou, N

dc.contributor.author	Goumas, G	en
dc.contributor.author	Anastopoulos, N	en
dc.contributor.author	Koziris, N	en
dc.contributor.author	Ioannou, N	en
dc.date.accessioned	2014-03-01T02:46:17Z
dc.date.available	2014-03-01T02:46:17Z
dc.date.issued	2009	en
dc.identifier.issn	15525244	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/32642
dc.subject	Overlapping	en
dc.subject	SMT architecture	en
dc.subject.other	Advection equations	en
dc.subject.other	Gbit Ethernet	en
dc.subject.other	Helper threading	en
dc.subject.other	Overlapping	en
dc.subject.other	Performance improvements	en
dc.subject.other	Simultaneous multithreading processors	en
dc.subject.other	Test case	en
dc.subject.other	Cluster computing	en
dc.subject.other	Computer science	en
dc.subject.other	Ethernet	en
dc.subject.other	Technical presentations	en
dc.subject.other	Multitasking	en
dc.title	Overlapping computation and communication in SMT clusters with commodity interconnects	en
heal.type	conferenceItem	en
heal.identifier.primary	10.1109/CLUSTR.2009.5289174	en
heal.identifier.secondary	http://dx.doi.org/10.1109/CLUSTR.2009.5289174	en
heal.identifier.secondary	5289174	en
heal.publicationDate	2009	en
heal.abstract	In this paper we focus on optimizing the performance in a cluster of Simultaneous Multithreading (SMT) processors connected with a commodity interconnect (e.g. Gbit Ethernet), by applying overlapping of computation with communication. As a test case we consider the parallelized advection equation and discuss the steps that need to be followed to semantically allow overlapping to occur.We propose an implementation based on the concept of Helper Threading that distributes computation and communication in the two sibling threads of an SMT processor, thus creating an asymmetric pair of execution patterns in each hardware context. Our experimental results in an 8-node cluster interconnected with commodity Gbit Ethernet demonstrate that the proposed implementation is able to achieve substantial performance improvements that can exceed 20% in some cases, by efficiently utilizing the available resources of the SMT processors. ©2009 IEEE.	en
heal.journalName	Proceedings - IEEE International Conference on Cluster Computing, ICCC	en
dc.identifier.doi	10.1109/CLUSTR.2009.5289174	en