izpis_h1_title_alt

Parallel random number generation
ID Ciglarič, Tadej (Author), ID Štrumbelj, Erik (Mentor) More about this mentor... This link opens in a new window, ID Češnovar, Rok (Co-mentor)

.pdfPDF - Presentation file, Download (492,08 KB)
MD5: 1DCF7941E871176A5F0FA5BAAC4608A3

Abstract
We present a library of 19 pseudo-random number generators, implemented for graphical processing units. The library is implemented in the OpenCL framework and empirically evaluated using the TestU01 library. Most of the presented generators pass the tests. The generators' performance is evaluated on five different devices. The Tyche-i generator is the best choice overall, while on some specific devices other generators are better.

Language:English
Keywords:pseudo-random number generation, parallelization, GPU, OpenCL
Work type:Master's thesis/paper
Organization:FRI - Faculty of Computer and Information Science
Year:2018
PID:20.500.12556/RUL-105387 This link opens in a new window
Publication date in RUL:23.11.2018
Views:1535
Downloads:377
Metadata:XML RDF-CHPDL DC-XML DC-RDF
:
Copy citation
Share:Bookmark and Share

Secondary language

Language:Slovenian
Title:Vzporedno generiranje naključnih števil
Abstract:
Predstavljamo knjižnico devetnajstih generatorjev psevdonaključnih števil. Generatorji so implementirani v programskem jeziku OpenCL in so namenjeni uporabi na grafičnih procesnih enotah. Večina implementiranih generatorjev prestane statistične teste kvalitete naključnosti generiranih števil iz knjižnice TestU01. Hitrost generiranja števil smo ovrednotili na petih različnih računskih napravah. Skupno najboljše rezultate dosega generator Tyche-i, vendar so za nekatere izmed naprav drugi generatorji boljši.

Keywords:generiranje psevdonaključnih števil, paralelizacija, GPE, OpenCL

Similar documents

Similar works from RUL:
Similar works from other Slovenian collections:

Back