Reducing the communication cost via chain pattern scheduling

Ciorba, FM; Andronikos, T; Drositis, L; Papakonstantinou, G; Tsanakas, P

dc.contributor.author	Ciorba, FM	en
dc.contributor.author	Andronikos, T	en
dc.contributor.author	Drositis, L	en
dc.contributor.author	Papakonstantinou, G	en
dc.contributor.author	Tsanakas, P	en
dc.date.accessioned	2014-03-01T02:43:29Z
dc.date.available	2014-03-01T02:43:29Z
dc.date.issued	2005	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/31441
dc.subject	Communication cost	en
dc.subject	Compute-intensive applications	en
dc.subject	Data locality	en
dc.subject	Nested loop scheduling	en
dc.subject	Patterns	en
dc.subject	Uniform dependence loop	en
dc.subject.other	Computational geometry	en
dc.subject.other	Computer programming languages	en
dc.subject.other	Conformal mapping	en
dc.subject.other	Digital signal processing	en
dc.subject.other	Parallel processing systems	en
dc.subject.other	Communication cost	en
dc.subject.other	Compute-intensive applications	en
dc.subject.other	Data locality	en
dc.subject.other	Nested loop scheduling	en
dc.subject.other	Uniform dependence loop	en
dc.subject.other	Telecommunication services	en
dc.title	Reducing the communication cost via chain pattern scheduling	en
heal.type	conferenceItem	en
heal.identifier.primary	10.1109/NCA.2005.43	en
heal.identifier.secondary	http://dx.doi.org/10.1109/NCA.2005.43	en
heal.identifier.secondary	1565952	en
heal.publicationDate	2005	en
heal.abstract	This paper deals with general nested loops and proposes a novel scheduling methodology for reducing the communication cost of parallel programs. General loops contain complex loop bodies (consisting of arbitrary program statements, such as assignments, conditions and repetitions) that exhibit uniform loop-carried dependencies. Therefore it is now possible to achieve efficient parallelization for a vast class of loops, mostly found in DSP, PDEs, signal and video coding. We use computational geometry methods, that exploit efficiently the regularity of nested loops index spaces, in order to significantly reduce the communication cost, which in most cases is the main drawback of parallel programs ' performance. Through extensive testing, we show that the proposed method outperforms in all cases the classic cyclic mapping, succeeding to reduce the communication by 15%35%. This significant reduction of the communication volume makes our method a promising candidate to be incorporated into existing automatic parallel code generation tools. © 2005 IEEE.	en
heal.journalName	Proceedings - Fourth IEEE International Symposium on Network Computing and Applications, NCA 2005	en
dc.identifier.doi	10.1109/NCA.2005.43	en
dc.identifier.volume	2005	en
dc.identifier.spage	186	en
dc.identifier.epage	193	en