IEICE Technical Committee Submission System
Conference Paper's Information
Online Proceedings
[Sign in]
Tech. Rep. Archives
 Go Top Page Go Previous   [Japanese] / [English] 

Paper Abstract and Keywords
Presentation 2017-07-28 11:15
Avoiding redundant data transfer on GPU-enabled Apache Spark
Ryo Asai, Masao Okita, Kenichi Hagihara (Osaka Univ.) CPSY2017-33
Abstract (in Japanese) (See Japanese page) 
(in English) Apache Spark is a fast and general engine for parallel distributed processing on PC cluster. IBMSparkGPU is a Spark plugin that enables GPU offloading. The IBMSparkGPU provides in-VRAM cache instruction for avoiding data transfer between the host PC and the GPU. Programmers are responsible for this cache management. In this paper, we propose an extention of IBMSparkGPU for automatically avoiding redundant data transfer. The proposed method manages in-VRAM cache so that redundant data transfer does not occur be- tween batched processes on the GPU. The batched processes can be identified by analyzing the data flow provided by Spark runtime. Experimental results show that the proposed method avoids redundant data transfer in a logistic regression program with overhead of at most 2.4%.
Keyword (in Japanese) (See Japanese page) 
(in English) CUDA / Design optimization / In-VRAM cache / / / / /  
Reference Info. IEICE Tech. Rep., vol. 117, no. 153, CPSY2017-33, pp. 185-190, July 2017.
Paper # CPSY2017-33 
Date of Issue 2017-07-19 (CPSY) 
ISSN Print edition: ISSN 0913-5685    Online edition: ISSN 2432-6380
Copyright
and
reproduction
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034)
Download PDF CPSY2017-33

Conference Information
Committee CPSY DC IPSJ-ARC  
Conference Date 2017-07-26 - 2017-07-28 
Place (in Japanese) (See Japanese page) 
Place (in English) Akita Atorion-Building (Akita) 
Topics (in Japanese) (See Japanese page) 
Topics (in English) Parallel, Distributed and Cooperative Processing 
Paper Information
Registration To CPSY 
Conference Code 2017-07-CPSY-DC-ARC 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Avoiding redundant data transfer on GPU-enabled Apache Spark 
Sub Title (in English)  
Keyword(1) CUDA  
Keyword(2) Design optimization  
Keyword(3) In-VRAM cache  
Keyword(4)  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Ryo Asai  
1st Author's Affiliation Osaka University (Osaka Univ.)
2nd Author's Name Masao Okita  
2nd Author's Affiliation Osaka University (Osaka Univ.)
3rd Author's Name Kenichi Hagihara  
3rd Author's Affiliation Osaka University (Osaka Univ.)
4th Author's Name  
4th Author's Affiliation ()
5th Author's Name  
5th Author's Affiliation ()
6th Author's Name  
6th Author's Affiliation ()
7th Author's Name  
7th Author's Affiliation ()
8th Author's Name  
8th Author's Affiliation ()
9th Author's Name  
9th Author's Affiliation ()
10th Author's Name  
10th Author's Affiliation ()
11th Author's Name  
11th Author's Affiliation ()
12th Author's Name  
12th Author's Affiliation ()
13th Author's Name  
13th Author's Affiliation ()
14th Author's Name  
14th Author's Affiliation ()
15th Author's Name  
15th Author's Affiliation ()
16th Author's Name  
16th Author's Affiliation ()
17th Author's Name  
17th Author's Affiliation ()
18th Author's Name  
18th Author's Affiliation ()
19th Author's Name  
19th Author's Affiliation ()
20th Author's Name  
20th Author's Affiliation ()
Speaker Author-1 
Date Time 2017-07-28 11:15:00 
Presentation Time 30 minutes 
Registration for CPSY 
Paper # CPSY2017-33 
Volume (vol) vol.117 
Number (no) no.153 
Page pp.185-190 
#Pages
Date of Issue 2017-07-19 (CPSY) 


[Return to Top Page]

[Return to IEICE Web Page]


The Institute of Electronics, Information and Communication Engineers (IEICE), Japan