Adaptive Task Size Control on High Level Programming for GPU/CPU Work Sharing

Abstract : On the work sharing among GPUs and CPU cores on GPU equipped clusters, it is a critical issue to keep load balance among these heterogeneous computing resources. We have been developing a runtime system for this problem on PGAS language named XcalableMP- dev/StarPU [1]. Through the development, we found the necessity of adaptive load balancing for GPU/CPU work sharing to achieve the best performance for various application codes. In this paper, we enhance our language system XcalableMP-dev/StarPU to add a new feature which can control the task size to be assigned to these heterogeneous resources dynamically during application execution. As a result of performance evaluation on several benchmarks, we confirmed the proposed feature correctly works and the performance with heterogeneous work sharing provides up to about 40% higher performance than GPU-only utilization even for relatively small size of problems.
Complete list of metadatas

Cited literature [5 references]  Display  Hide  Download

https://hal.inria.fr/hal-00920915
Contributor : Olivier Aumage <>
Submitted on : Thursday, December 19, 2013 - 2:03:16 PM
Last modification on : Tuesday, May 14, 2019 - 11:38:08 AM
Long-term archiving on : Thursday, March 20, 2014 - 11:46:51 AM

File

ADPC2013-117.pdf
Files produced by the author(s)

Identifiers

Citation

Tetsuya Odajima, Taisuke Boku, Mitsuhisa Sato, Toshihiro Hanawa, Yuetsu Kodama, et al.. Adaptive Task Size Control on High Level Programming for GPU/CPU Work Sharing. The 2013 International Symposium on Advances of Distributed and Parallel Computing (ADPC 2013), Dec 2013, Vietri sul Mare, Italy. ⟨10.1007/978-3-319-03889-6_7⟩. ⟨hal-00920915⟩

Share

Metrics

Record views

638

Files downloads

505