Exploration of CPU/GPU co-execution: From the perspective of performance, energy, and temperature

Seunggu Kang, Hong Jun Choi, Cheol Hong Kim, Sung Woo Jung, Dongseop Kwon, Joong Chae Na

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

In recent computing systems, CPUs have encountered the situations in which they cannot meet the increasing throughput demands. To overcome the limits of CPUs in processing heavy tasks, especially for computer graphics, GPUs have been widely used. Therefore, the performance of up-to-date computing systems can be maximized when the task scheduling between the CPU and the GPU is optimized. In this paper, we analyze the system in the perspective of performance, energy efficiency, and temperature according to the execution methods between the CPU and the GPU. Experimental results show that the GPU leads to better efficiency compared to the CPU when single application is executed. However, when two applications are executed, the GPU does not guarantee superior efficiency than the CPU depending on the application characteristics.

Original languageEnglish
Title of host publicationProceedings of the 2011 ACM Research in Applied Computation Symposium, RACS 2011
Pages38-43
Number of pages6
DOIs
Publication statusPublished - 2011 Dec 1
Event2011 ACM Research in Applied Computation Symposium, RACS 2011 - Miami, FL, United States
Duration: 2011 Nov 22011 Nov 5

Other

Other2011 ACM Research in Applied Computation Symposium, RACS 2011
CountryUnited States
CityMiami, FL
Period11/11/211/11/5

Fingerprint

Program processors
Energy
Computing
Task Scheduling
Computer graphics
Energy Efficiency
Temperature
Throughput
Experimental Results
Energy efficiency
Graphics processing unit
Scheduling
Processing

Keywords

  • CPU
  • CUDA
  • GPU
  • high-performance computing
  • scheduling

ASJC Scopus subject areas

  • Computational Theory and Mathematics
  • Applied Mathematics

Cite this

Kang, S., Choi, H. J., Kim, C. H., Jung, S. W., Kwon, D., & Na, J. C. (2011). Exploration of CPU/GPU co-execution: From the perspective of performance, energy, and temperature. In Proceedings of the 2011 ACM Research in Applied Computation Symposium, RACS 2011 (pp. 38-43) https://doi.org/10.1145/2103380.2103388

Exploration of CPU/GPU co-execution : From the perspective of performance, energy, and temperature. / Kang, Seunggu; Choi, Hong Jun; Kim, Cheol Hong; Jung, Sung Woo; Kwon, Dongseop; Na, Joong Chae.

Proceedings of the 2011 ACM Research in Applied Computation Symposium, RACS 2011. 2011. p. 38-43.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Kang, S, Choi, HJ, Kim, CH, Jung, SW, Kwon, D & Na, JC 2011, Exploration of CPU/GPU co-execution: From the perspective of performance, energy, and temperature. in Proceedings of the 2011 ACM Research in Applied Computation Symposium, RACS 2011. pp. 38-43, 2011 ACM Research in Applied Computation Symposium, RACS 2011, Miami, FL, United States, 11/11/2. https://doi.org/10.1145/2103380.2103388
Kang S, Choi HJ, Kim CH, Jung SW, Kwon D, Na JC. Exploration of CPU/GPU co-execution: From the perspective of performance, energy, and temperature. In Proceedings of the 2011 ACM Research in Applied Computation Symposium, RACS 2011. 2011. p. 38-43 https://doi.org/10.1145/2103380.2103388
Kang, Seunggu ; Choi, Hong Jun ; Kim, Cheol Hong ; Jung, Sung Woo ; Kwon, Dongseop ; Na, Joong Chae. / Exploration of CPU/GPU co-execution : From the perspective of performance, energy, and temperature. Proceedings of the 2011 ACM Research in Applied Computation Symposium, RACS 2011. 2011. pp. 38-43
@inproceedings{9fa5900b3fce4cf285b252d5a59e0297,
title = "Exploration of CPU/GPU co-execution: From the perspective of performance, energy, and temperature",
abstract = "In recent computing systems, CPUs have encountered the situations in which they cannot meet the increasing throughput demands. To overcome the limits of CPUs in processing heavy tasks, especially for computer graphics, GPUs have been widely used. Therefore, the performance of up-to-date computing systems can be maximized when the task scheduling between the CPU and the GPU is optimized. In this paper, we analyze the system in the perspective of performance, energy efficiency, and temperature according to the execution methods between the CPU and the GPU. Experimental results show that the GPU leads to better efficiency compared to the CPU when single application is executed. However, when two applications are executed, the GPU does not guarantee superior efficiency than the CPU depending on the application characteristics.",
keywords = "CPU, CUDA, GPU, high-performance computing, scheduling",
author = "Seunggu Kang and Choi, {Hong Jun} and Kim, {Cheol Hong} and Jung, {Sung Woo} and Dongseop Kwon and Na, {Joong Chae}",
year = "2011",
month = "12",
day = "1",
doi = "10.1145/2103380.2103388",
language = "English",
isbn = "9781450310871",
pages = "38--43",
booktitle = "Proceedings of the 2011 ACM Research in Applied Computation Symposium, RACS 2011",

}

TY - GEN

T1 - Exploration of CPU/GPU co-execution

T2 - From the perspective of performance, energy, and temperature

AU - Kang, Seunggu

AU - Choi, Hong Jun

AU - Kim, Cheol Hong

AU - Jung, Sung Woo

AU - Kwon, Dongseop

AU - Na, Joong Chae

PY - 2011/12/1

Y1 - 2011/12/1

N2 - In recent computing systems, CPUs have encountered the situations in which they cannot meet the increasing throughput demands. To overcome the limits of CPUs in processing heavy tasks, especially for computer graphics, GPUs have been widely used. Therefore, the performance of up-to-date computing systems can be maximized when the task scheduling between the CPU and the GPU is optimized. In this paper, we analyze the system in the perspective of performance, energy efficiency, and temperature according to the execution methods between the CPU and the GPU. Experimental results show that the GPU leads to better efficiency compared to the CPU when single application is executed. However, when two applications are executed, the GPU does not guarantee superior efficiency than the CPU depending on the application characteristics.

AB - In recent computing systems, CPUs have encountered the situations in which they cannot meet the increasing throughput demands. To overcome the limits of CPUs in processing heavy tasks, especially for computer graphics, GPUs have been widely used. Therefore, the performance of up-to-date computing systems can be maximized when the task scheduling between the CPU and the GPU is optimized. In this paper, we analyze the system in the perspective of performance, energy efficiency, and temperature according to the execution methods between the CPU and the GPU. Experimental results show that the GPU leads to better efficiency compared to the CPU when single application is executed. However, when two applications are executed, the GPU does not guarantee superior efficiency than the CPU depending on the application characteristics.

KW - CPU

KW - CUDA

KW - GPU

KW - high-performance computing

KW - scheduling

UR - http://www.scopus.com/inward/record.url?scp=84863119763&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84863119763&partnerID=8YFLogxK

U2 - 10.1145/2103380.2103388

DO - 10.1145/2103380.2103388

M3 - Conference contribution

AN - SCOPUS:84863119763

SN - 9781450310871

SP - 38

EP - 43

BT - Proceedings of the 2011 ACM Research in Applied Computation Symposium, RACS 2011

ER -