國立虎尾科技大學 |

Novel Techniques for Graph Algorithm Acceleration.

Record Type:	Language materials, manuscript : Monograph/item
Title/Author:	Novel Techniques for Graph Algorithm Acceleration./
Author:	Liu, Hang.
Description:	1 online resource (167 pages)
Notes:	Source: Dissertation Abstracts International, Volume: 79-08(E), Section: B.
Contained By:	Dissertation Abstracts International79-08B(E).
Subject:	Computer engineering. -
Online resource:	click for full text (PQDT)
ISBN:	9780355828474

Novel Techniques for Graph Algorithm Acceleration.
Liu, Hang.

Novel Techniques for Graph Algorithm Acceleration. - 1 online resource (167 pages)

Source: Dissertation Abstracts International, Volume: 79-08(E), Section: B.

Thesis (Ph.D.)--The George Washington University, 2017.

Includes bibliographical references

The concept of graph has been around since Euler brought up the Seven Bridges of Knigsberg problem in 1736. Recent years have seen graph computing regains its momentum because of many emerging graph relevant applications, e.g., World-Wide-Web (WWW) networks, social and computer networks, metabolic interactions and chemical compound design graphs. This dissertation strives to provide graph computing systems which are able to quickly compute very large graph datasets with relatively low cost and expose easy programming interface to programmers..

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2018

Mode of access: World Wide Web

ISBN: 9780355828474Subjects--Topical Terms:

569006
Computer engineering.
Index Terms--Genre/Form:

554714
Electronic books.

Novel Techniques for Graph Algorithm Acceleration.
LDR:03290ntm a2200373Ki 4500 001 920659
005 20181203094031.5
006 m o u
007 cr mn||||a|a||
008 190606s2017 xx obm 000 0 eng d
020 $a 9780355828474
035 $a (MiAaPQ)AAI10621649
035 $a (MiAaPQ)gwu:13796
035 $a AAI10621649
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Liu, Hang. $3 1195523
245 1 0 $a Novel Techniques for Graph Algorithm Acceleration.
264 0 $c 2017
300 $a 1 online resource (167 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertation Abstracts International, Volume: 79-08(E), Section: B.
500 $a Advisers: H. Howie Huang; Ahmed Louri.
502 $a Thesis (Ph.D.)--The George Washington University, 2017.
504 $a Includes bibliographical references
520 $a The concept of graph has been around since Euler brought up the Seven Bridges of Knigsberg problem in 1736. Recent years have seen graph computing regains its momentum because of many emerging graph relevant applications, e.g., World-Wide-Web (WWW) networks, social and computer networks, metabolic interactions and chemical compound design graphs. This dissertation strives to provide graph computing systems which are able to quickly compute very large graph datasets with relatively low cost and expose easy programming interface to programmers..
520 $a The first part of this dissertation introduces the Graphics Processing Units (GPUs) accelerated graph traversal which consists of two projects---Enterprise and iBFS. Particularly, Enterprise is the first work that achieves atomic operation free Breadth-First Search (BFS) on GPUs and iBFS is the first to conduct multiple traversals together on GPUs. Both projects achieve orders of magnitude speedup over state-of-the-art.
520 $a The second part introduces SIMD-X, a graph framework that supports a variety of graph algorithms on GPUs. SIMD-X not only provides a simple Active-Compute-Combine (ACC) programming model for end users to express graph algorithms on Single Instruction Multiple Data (SIMD) GPUs, but also creates opportunities for system-level optimizations. Together, SIMD-X allows programmers to develop a typical graph algorithm with less than 100 Lines Of Code (LOCs) and achieve an order of magnitude speedup over Gunrock.
520 $a Finally, this dissertation describes Graphene which can tackle trillion-edge graphs on a single machine with an array of Solid State Drives (SSDs). To enhance the bandwidth utilization of such an array of SSDs, we introduce a bitmap based IO request management component that improves bandwidth efficiency by 4--8x and a row-column 2D graph partition approach to balance the graph data access across the disks. Notably, Graphene achieves comparable performance to in-memory systems, e.g., Galois with merely 10% of memory consumption.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2018
538 $a Mode of access: World Wide Web
650 4 $a Computer engineering. $3 569006
650 4 $a Engineering. $3 561152
655 7 $a Electronic books. $2 local $3 554714
690 $a 0464
690 $a 0537
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a The George Washington University. $b Computer Engineering. $3 1195524
773 0 $t Dissertation Abstracts International $g 79-08B(E).
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=10621649 $z click for full text (PQDT)