Performance Study of Random Walk Simulations

dc.contributor.advisor	Natvig, Lasse	nb_NO
dc.contributor.advisor	Fjeldstad, Ståle	nb_NO
dc.contributor.author	Mahic, Safurudin	nb_NO
dc.date.accessioned	2014-12-19T13:31:27Z
dc.date.available	2014-12-19T13:31:27Z
dc.date.created	2010-09-02	nb_NO
dc.date.issued	2009	nb_NO
dc.identifier	347162	nb_NO
dc.identifier	ntnudaim:4690	nb_NO
dc.identifier.uri	http://hdl.handle.net/11250/250352
dc.description.abstract	Our research shows that generation of random numbers in the benchmark Random Walk application is a significant fraction of the overall computational time in Random Walk simulations (40%). Furthermore, we show that choice of implementation and compiler flags can have a significant impact on performance. Our weakest implementation of the benchmark application performs the computations in 5338 seconds, while the fastest implementation performs the same computations in merely 71 seconds, a 75x speed improvement.	nb_NO
dc.language	eng	nb_NO
dc.publisher	Institutt for datateknikk og informasjonsvitenskap	nb_NO
dc.subject	ntnudaim	no_NO
dc.subject	SIF2 datateknikk	no_NO
dc.subject	Komplekse datasystemer	no_NO
dc.title	Performance Study of Random Walk Simulations	nb_NO
dc.type	Master thesis	nb_NO
dc.source.pagenumber	176	nb_NO
dc.contributor.department	Norges teknisk-naturvitenskapelige universitet, Fakultet for informasjonsteknologi, matematikk og elektroteknikk, Institutt for datateknikk og informasjonsvitenskap	nb_NO