Fast quasi double-precision method with single-precision hardware to accelerate scientific applications

Tetsu Narumi; Tsuyoshi Hamada; Keigo Nitadori; Ryuji Sakamaki; Kenji Yasuoka

doi:10.1142/S0219876211002708

Fast quasi double-precision method with single-precision hardware to accelerate scientific applications

Tetsu Narumi, Tsuyoshi Hamada, Keigo Nitadori, Ryuji Sakamaki, Kenji Yasuoka

Department of Mechanical Engineering

Research output: Contribution to journal › Article › peer-review

2 Citations (Scopus)

Abstract

The recent commodity hardware, such as the Cell processor in PLAYSTATION 3 or GeForce GTX280 GPU, has much more peak performance in single-precision than that of CPU of PCs, while the double-precision performance of these computers is comparable to it. Even though a quasi double-precision method achieves comparable accuracy to double-precision, the performance of it is relatively low. In this paper, the quasi double-precision method is modified, and it can deliver more performance than that by native double-precision of the GPU. Calculation of a Mandelbrot set and molecular dynamics (MD) simulation are used to test the method.

Original language	English
Pages (from-to)	561-581
Number of pages	21
Journal	International Journal of Computational Methods
Volume	8
Issue number	3
DOIs	https://doi.org/10.1142/S0219876211002708
Publication status	Published - 2011 Sept

Keywords

Mixed-precision method
graphics processing unit
molecular dynamics simulation
quasi double-precision method

ASJC Scopus subject areas

Computer Science (miscellaneous)
Computational Mathematics

Access to Document

10.1142/S0219876211002708

Cite this

@article{c0ff6785f3894abd83150656020ec3f5,

title = "Fast quasi double-precision method with single-precision hardware to accelerate scientific applications",

abstract = "The recent commodity hardware, such as the Cell processor in PLAYSTATION 3 or GeForce GTX280 GPU, has much more peak performance in single-precision than that of CPU of PCs, while the double-precision performance of these computers is comparable to it. Even though a quasi double-precision method achieves comparable accuracy to double-precision, the performance of it is relatively low. In this paper, the quasi double-precision method is modified, and it can deliver more performance than that by native double-precision of the GPU. Calculation of a Mandelbrot set and molecular dynamics (MD) simulation are used to test the method.",

keywords = "Mixed-precision method, graphics processing unit, molecular dynamics simulation, quasi double-precision method",

author = "Tetsu Narumi and Tsuyoshi Hamada and Keigo Nitadori and Ryuji Sakamaki and Kenji Yasuoka",

year = "2011",

month = sep,

doi = "10.1142/S0219876211002708",

language = "English",

volume = "8",

pages = "561--581",

journal = "International Journal of Computational Methods",

issn = "0219-8762",

publisher = "World Scientific Publishing Co. Pte Ltd",

number = "3",

}

TY - JOUR

T1 - Fast quasi double-precision method with single-precision hardware to accelerate scientific applications

AU - Narumi, Tetsu

AU - Hamada, Tsuyoshi

AU - Nitadori, Keigo

AU - Sakamaki, Ryuji

AU - Yasuoka, Kenji

PY - 2011/9

Y1 - 2011/9

N2 - The recent commodity hardware, such as the Cell processor in PLAYSTATION 3 or GeForce GTX280 GPU, has much more peak performance in single-precision than that of CPU of PCs, while the double-precision performance of these computers is comparable to it. Even though a quasi double-precision method achieves comparable accuracy to double-precision, the performance of it is relatively low. In this paper, the quasi double-precision method is modified, and it can deliver more performance than that by native double-precision of the GPU. Calculation of a Mandelbrot set and molecular dynamics (MD) simulation are used to test the method.

AB - The recent commodity hardware, such as the Cell processor in PLAYSTATION 3 or GeForce GTX280 GPU, has much more peak performance in single-precision than that of CPU of PCs, while the double-precision performance of these computers is comparable to it. Even though a quasi double-precision method achieves comparable accuracy to double-precision, the performance of it is relatively low. In this paper, the quasi double-precision method is modified, and it can deliver more performance than that by native double-precision of the GPU. Calculation of a Mandelbrot set and molecular dynamics (MD) simulation are used to test the method.

KW - Mixed-precision method

KW - graphics processing unit

KW - molecular dynamics simulation

KW - quasi double-precision method

UR - http://www.scopus.com/inward/record.url?scp=80052601885&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=80052601885&partnerID=8YFLogxK

U2 - 10.1142/S0219876211002708

DO - 10.1142/S0219876211002708

M3 - Article

AN - SCOPUS:80052601885

SN - 0219-8762

VL - 8

SP - 561

EP - 581

JO - International Journal of Computational Methods

JF - International Journal of Computational Methods

IS - 3

ER -

Fast quasi double-precision method with single-precision hardware to accelerate scientific applications

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this