COMPARISON OF SINGLE AND DOUBLE FLOATING POINT PRECISION PERFORMANCE FOR TESLA ARCHITECTURE GPUs

The paper compares the single and double floating point precision performance of NVIDIA Tesla GPUs. Double precision is crucial for the accuracy of some applications containing scientific computation. Three representative applications have been chosen in order to compare the performances: matrix mul...

Full description

Saved in:
Bibliographic Details
Published inBulletin of the Transilvania University of Braşov. Series I Engineering Sciences Vol. 4; no. 2; p. 131
Main Authors Itu, L M, Suciu, C, Moldoveanu, F, Postelnicu, A
Format Journal Article
LanguageEnglish
Published Brasov Transilvania University of Brasov 01.07.2011
Subjects
Bandwidth
Conduction
Floating point arithmetic
Heat conduction
Mathematical analysis
Multiplication
Navier-Stokes equations
Steady state
Online AccessGet full text

Cover

More Information
Summary:The paper compares the single and double floating point precision performance of NVIDIA Tesla GPUs. Double precision is crucial for the accuracy of some applications containing scientific computation. Three representative applications have been chosen in order to compare the performances: matrix multiplication, incompressible Navier-Stokes equations and the steady state heat conduction problem. The expected values of performance decrease lie between two for bandwidth limited applications and eight for compute limited applications. Based on the conducted experiments, the type of each application has been identified and the decreases in performance have confirmed the expected theoretical values.
Bibliography:ObjectType-Article-2
SourceType-Scholarly Journals-1
ObjectType-Feature-1
content type line 23
ISSN:2065-2119