Scalable global and local hashing strategies for duplicate pruning in parallel A graph search

For many applications of the A* algorithm, the state space is a graph rather than a tree. The implication of this for parallel A* algorithms is that different processors may perform significant duplicated work if interprocessor duplicates are not pruned. In this paper, we consider the problem of dup...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE transactions on parallel and distributed systems Jg. 8; H. 7; S. 738 - 756
Hauptverfasser: Mahapatra, N.R., Dutt, S.
Format: Journal Article
Sprache:Englisch
Veröffentlicht: IEEE 01.07.1997
Schlagworte:
ISSN:1045-9219
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Abstract For many applications of the A* algorithm, the state space is a graph rather than a tree. The implication of this for parallel A* algorithms is that different processors may perform significant duplicated work if interprocessor duplicates are not pruned. In this paper, we consider the problem of duplicate pruning in parallel A* graph-search algorithms implemented on distributed-memory machines. A commonly used method for duplicate pruning uses a hash function to associate with each distinct node of the search space a particular processor to which duplicate nodes arising in different processors are transmitted and thereby pruned. This approach has two major drawbacks. First, load balance is determined solely by the hash function. Second, node transmissions for duplicate pruning are global; this can lead to hot spots and slower message delivery. To overcome these problems, we propose two different duplicate pruning strategies: 1) To achieve good load balance, we decouple the task of duplicate pruning from load balancing, by using a hash function for the former and a load balancing scheme for the latter. 2) A novel search-space partitioning scheme that allocates disjoint parts of the search space to disjoint subcubes in a hypercube (or disjoint processor groups in the target architecture), so that duplicate pruning is achieved with only intrasubcube or adjacent intersubcube communication. Thus message latency and hot-spot probability are greatly reduced. The above duplicate pruning schemes were implemented on an nCUBE2 hypercube multicomputer to solve the Traveling Salesman Problem (TSP). For uniformly distributed intercity costs, our strategies yield a speedup improvement of 13 to 35 percent on 1,024-processors over previous methods that do not prune any duplicates, and 13 to 25 percent over the previous hashing-only scheme. For normally distributed data the corresponding figures are 135 percent and 10 to 155 percent. Finally, we analyze the scalability of our parallel A* algorithms on k-ary n-cube networks in terms of the isoefficiency metric, and show that they have isoefficiency lower and upper bounds of /spl Theta/(P log P) and /spl Theta/(Pkn/sup 2/), respectively.
AbstractList The state space is a graph rather than a tree for many applications of the A* algorithm. This means that in parallel A* algorithms, different processors may perform significant duplicated work if interprocessor duplicates are not pruned. The problem of duplicate pruning in parallel A* graph-search algorithms implemented on distributed-memory machines is presented. Two different duplicate pruning strategies are discussed to solve this problem: first, duplicate pruning from load balancing is decoupled by using hash functions to achieve load balance, and second, a search space partitioning scheme that allocates disjoint parts of the search space to disjoint subcubes in hypercubes so that duplicate pruning is achieved.
For many applications of the A* algorithm, the state space is a graph rather than a tree. The implication of this for parallel A* algorithms is that different processors may perform significant duplicated work if interprocessor duplicates are not pruned. In this paper, we consider the problem of duplicate pruning in parallel A* graph-search algorithms implemented on distributed-memory machines. A commonly used method for duplicate pruning uses a hash function to associate with each distinct node of the search space a particular processor to which duplicate nodes arising in different processors are transmitted and thereby pruned. This approach has two major drawbacks. First, load balance is determined solely by the hash function. Second, node transmissions for duplicate pruning are global; this can lead to hot spots and slower message delivery. To overcome these problems, we propose two different duplicate pruning strategies: 1) To achieve good load balance, we decouple the task of duplicate pruning from load balancing, by using a hash function for the former and a load balancing scheme for the latter. 2) A novel search-space partitioning scheme that allocates disjoint parts of the search space to disjoint subcubes in a hypercube (or disjoint processor groups in the target architecture), so that duplicate pruning is achieved with only intrasubcube or adjacent intersubcube communication. Thus message latency and hot-spot probability are greatly reduced. The above duplicate pruning schemes were implemented on an nCUBE2 hypercube multicomputer to solve the Traveling Salesman Problem (TSP). For uniformly distributed intercity costs, our strategies yield a speedup improvement of 13 to 35 percent on 1,024-processors over previous methods that do not prune any duplicates, and 13 to 25 percent over the previous hashing-only scheme. For normally distributed data the corresponding figures are 135 percent and 10 to 155 percent. Finally, we analyze the scalability of our parallel A* algorithms on k-ary n-cube networks in terms of the isoefficiency metric, and show that they have isoefficiency lower and upper bounds of Theta(P log P) and Theta(Pkn(2)), respectively
For many applications of the A* algorithm, the state space is a graph rather than a tree. The implication of this for parallel A* algorithms is that different processors may perform significant duplicated work if interprocessor duplicates are not pruned. In this paper, we consider the problem of duplicate pruning in parallel A* graph-search algorithms implemented on distributed-memory machines. A commonly used method for duplicate pruning uses a hash function to associate with each distinct node of the search space a particular processor to which duplicate nodes arising in different processors are transmitted and thereby pruned. This approach has two major drawbacks. First, load balance is determined solely by the hash function. Second, node transmissions for duplicate pruning are global; this can lead to hot spots and slower message delivery. To overcome these problems, we propose two different duplicate pruning strategies: 1) To achieve good load balance, we decouple the task of duplicate pruning from load balancing, by using a hash function for the former and a load balancing scheme for the latter. 2) A novel search-space partitioning scheme that allocates disjoint parts of the search space to disjoint subcubes in a hypercube (or disjoint processor groups in the target architecture), so that duplicate pruning is achieved with only intrasubcube or adjacent intersubcube communication. Thus message latency and hot-spot probability are greatly reduced. The above duplicate pruning schemes were implemented on an nCUBE2 hypercube multicomputer to solve the Traveling Salesman Problem (TSP). For uniformly distributed intercity costs, our strategies yield a speedup improvement of 13 to 35 percent on 1,024-processors over previous methods that do not prune any duplicates, and 13 to 25 percent over the previous hashing-only scheme. For normally distributed data the corresponding figures are 135 percent and 10 to 155 percent. Finally, we analyze the scalability of our parallel A* algorithms on k-ary n-cube networks in terms of the isoefficiency metric, and show that they have isoefficiency lower and upper bounds of /spl Theta/(P log P) and /spl Theta/(Pkn/sup 2/), respectively.
Author Mahapatra, N.R.
Dutt, S.
Author_xml – sequence: 1
  givenname: N.R.
  surname: Mahapatra
  fullname: Mahapatra, N.R.
  organization: Dept. of Electr. & Comput. Eng., State Univ. of New York, Buffalo, NY, USA
– sequence: 2
  givenname: S.
  surname: Dutt
  fullname: Dutt, S.
BookMark eNqFkD1PwzAURT0UibYwsDJ5QmJIayd2HI9VxZdUiQEYUfTivKRGbhLsZODfkyoVA0JietJ9597hLMisaRsk5IqzFedMrxVfSZ0lIpuROWdCRjrm-pwsQvhgjAvJxJy8vxhwUDiktWsLcBSakrp2DOkewt42NQ29hx5ri4FWrafl0DlrxoR2fmiOgG1oBx6cQ0c3tPbQ7WlA8GZ_Qc4qcAEvT3dJ3u7vXreP0e754Wm72UUmUaKPEiZSWRjQqVFaJxjHAkuRVQyhZExmouQxU6bSDFUlCy6UzNAAcEi1iiuZLMnNtNv59nPA0OcHGww6Bw22Q8jjTCqW6vR_ME20Zmk2grcTaHwbgscq77w9gP_KOcuPdnPF88nuyK5_scb20Nu2Gc1Z92fjempYRPxZPj2_Ad7aiCY
CODEN ITDSEO
CitedBy_id crossref_primary_10_1016_j_artint_2012_10_007
crossref_primary_10_1016_j_asoc_2007_10_011
crossref_primary_10_1016_j_parco_2004_05_001
crossref_primary_10_1109_ACCESS_2020_2973607
crossref_primary_10_1006_jpdc_2000_1664
crossref_primary_10_1016_j_jpdc_2005_05_028
crossref_primary_10_1109_69_755612
Cites_doi 10.1090/dimacs/022/09
10.1109/12.53599
10.1016/0167-6377(89)90038-2
10.1145/174130.174145
10.1109/DMCC.1990.556310
10.1016/0004-3702(89)90010-6
10.1287/opre.11.6.972
10.1016/j.jpdc.2007.05.013
10.1109/FMPC.1990.89450
10.1137/0804046
10.1145/2422.322422
10.1006/jpdc.1994.1106
10.1109/IPPS.1992.222970
10.1007/978-1-4757-2219-2
10.1109/IPPS.1993.262779
ContentType Journal Article
DBID AAYXX
CITATION
7SC
8FD
JQ2
L7M
L~C
L~D
DOI 10.1109/71.598348
DatabaseName CrossRef
Computer and Information Systems Abstracts
Technology Research Database
ProQuest Computer Science Collection
Advanced Technologies Database with Aerospace
Computer and Information Systems Abstracts – Academic
Computer and Information Systems Abstracts Professional
DatabaseTitle CrossRef
Computer and Information Systems Abstracts
Technology Research Database
Computer and Information Systems Abstracts – Academic
Advanced Technologies Database with Aerospace
ProQuest Computer Science Collection
Computer and Information Systems Abstracts Professional
DatabaseTitleList Computer and Information Systems Abstracts
Computer and Information Systems Abstracts

DeliveryMethod fulltext_linktorsrc
Discipline Engineering
Computer Science
EndPage 756
ExternalDocumentID 10_1109_71_598348
598348
GroupedDBID --Z
-~X
.DC
0R~
29I
4.4
5GY
5VS
6IK
97E
AAJGR
AARMG
AASAJ
AAWTH
ABAZT
ABFSI
ABQJQ
ABVLG
ACGFO
ACIWK
AENEX
AETIX
AGQYO
AGSQL
AHBIQ
AI.
AIBXA
AKJIK
AKQYR
ALLEH
ALMA_UNASSIGNED_HOLDINGS
ASUFR
ATWAV
BEFXN
BFFAM
BGNUA
BKEBE
BPEOZ
CS3
DU5
E.L
EBS
EJD
HZ~
H~9
ICLAB
IEDLZ
IFIPE
IFJZH
IPLJI
JAVBF
LAI
M43
MS~
O9-
OCL
P2P
PQQKQ
RIA
RIE
RNI
RNS
RZB
TN5
TWZ
UHB
VH1
AAYXX
CITATION
7SC
8FD
JQ2
L7M
L~C
L~D
ID FETCH-LOGICAL-c374t-30465bca96c7993e224ed48f0ead00584d1207cf90e7f5b14758ecaa1a6972f53
IEDL.DBID RIE
ISICitedReferencesCount 11
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=10_1109_71_598348&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 1045-9219
IngestDate Sun Sep 28 12:04:10 EDT 2025
Wed Oct 01 17:07:45 EDT 2025
Sat Nov 29 03:35:56 EST 2025
Tue Nov 18 20:53:15 EST 2025
Wed Aug 27 02:52:19 EDT 2025
IsPeerReviewed true
IsScholarly true
Issue 7
Language English
License https://ieeexplore.ieee.org/Xplorehelp/downloads/license-information/IEEE.html
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-c374t-30465bca96c7993e224ed48f0ead00584d1207cf90e7f5b14758ecaa1a6972f53
Notes ObjectType-Article-2
SourceType-Scholarly Journals-1
ObjectType-Feature-1
content type line 23
PQID 26399068
PQPubID 23500
PageCount 19
ParticipantIDs crossref_primary_10_1109_71_598348
ieee_primary_598348
proquest_miscellaneous_26399068
crossref_citationtrail_10_1109_71_598348
proquest_miscellaneous_28570696
PublicationCentury 1900
PublicationDate 1997-07-01
PublicationDateYYYYMMDD 1997-07-01
PublicationDate_xml – month: 07
  year: 1997
  text: 1997-07-01
  day: 01
PublicationDecade 1990
PublicationTitle IEEE transactions on parallel and distributed systems
PublicationTitleAbbrev TPDS
PublicationYear 1997
Publisher IEEE
Publisher_xml – name: IEEE
References wah (bibl073828) 1981
bibl073827
rich (bibl073823) 1983
bibl073826
holte (bibl073810) 1994
bibl073811
mahapatra (bibl073818) 1995; 22
bibl07387
knuth (bibl073814) 1973; 3
bibl07388
kumar (bibl073812) 1988; 1
bibl07385
bibl07386
bibl073816
bibl073817
mohan (bibl073822) 1983
bibl07384
hennessey (bibl07389) 1990
kumar (bibl073813) 1989; 1
bibl073819
bibl07382
bibl073824
sen (bibl073825) 1989
lawler (bibl073815) 1976
bibl073821
manzini (bibl073820) 1990
cormen (bibl07383) 1990
anderson (bibl07381) 1987
References_xml – volume: 22
  start-page: 197
  year: 1995
  ident: bibl073818
  article-title: new anticipatory load balancing strategies for parallel a* algorithms
  publication-title: Am Math Society s Proc DIMACS Series in Discrete Math and Theoretical Computer Science
  doi: 10.1090/dimacs/022/09
– volume: 1
  start-page: 603
  year: 1989
  ident: bibl073813
  article-title: load balancing on the hypercube architecture
  publication-title: Proc Fourth Conf Hypercubes Concurrent Computers and Applications
– ident: bibl07384
  doi: 10.1109/12.53599
– ident: bibl073821
  doi: 10.1016/0167-6377(89)90038-2
– ident: bibl073811
  doi: 10.1145/174130.174145
– ident: bibl073824
  doi: 10.1109/DMCC.1990.556310
– start-page: 191
  year: 1983
  ident: bibl073822
  article-title: experience with two parallel programs solving the traveling salesman problem
  publication-title: Proc 1983 Int l Conf Parallel Processing
– ident: bibl07382
  doi: 10.1016/0004-3702(89)90010-6
– start-page: 263
  year: 1994
  ident: bibl073810
  article-title: searching with abstractions: a unifying framework and new high-performance algorithm
  publication-title: Proc 10th Canadian Conf Artificial Intelligence
– start-page: 309
  year: 1987
  ident: bibl07381
  article-title: parallel branch-and-bound algorithms on the hypercube
  publication-title: Proc Second Conf Hypercube Multiprocessors
– ident: bibl073816
  doi: 10.1287/opre.11.6.972
– start-page: 297
  year: 1989
  ident: bibl073825
  article-title: fast recursive formulations for best-first search that allow controlled use of memory
  publication-title: Proc 11th Int l Joint Conf Artificial Intelligence (IJCAI-89)
– year: 1990
  ident: bibl07383
  publication-title: Introduction to Algorithms
– year: 1983
  ident: bibl073823
  publication-title: Artificial Intelligence
– year: 1990
  ident: bibl073820
  article-title: probabilistic performance analysis of heuristic search using parallel hash tables
  publication-title: Proc Int l Symp Artificial Intelligence and Math
– year: 1976
  ident: bibl073815
  publication-title: Combinatorial Optimization Networks and Matroids
– ident: bibl073819
  doi: 10.1016/j.jpdc.2007.05.013
– ident: bibl07388
  doi: 10.1109/FMPC.1990.89450
– ident: bibl07387
  doi: 10.1137/0804046
– ident: bibl073827
  doi: 10.1145/2422.322422
– ident: bibl07386
  doi: 10.1006/jpdc.1994.1106
– ident: bibl073817
  doi: 10.1109/IPPS.1992.222970
– volume: 1
  start-page: 122
  year: 1988
  ident: bibl073812
  article-title: parallel best-first search of state-space graphs: a summary of results
  publication-title: Proc Seventh Nat l Conf Artificial Intelligence (AAAI 88)
– year: 1990
  ident: bibl07389
  publication-title: Computer Architecture A Quantitative Approach
– volume: 3
  year: 1973
  ident: bibl073814
  article-title: sorting and searching
  publication-title: The art of computer programming
– start-page: 239
  year: 1981
  ident: bibl073828
  article-title: manip-a parallel computer system for implementing branch and bound algorithms
  publication-title: Proc Eighth Ann Symp Computer Architecture
– ident: bibl073826
  doi: 10.1007/978-1-4757-2219-2
– ident: bibl07385
  doi: 10.1109/IPPS.1993.262779
SSID ssj0014504
Score 1.6066258
Snippet For many applications of the A* algorithm, the state space is a graph rather than a tree. The implication of this for parallel A* algorithms is that different...
The state space is a graph rather than a tree for many applications of the A* algorithm. This means that in parallel A* algorithms, different processors may...
SourceID proquest
crossref
ieee
SourceType Aggregation Database
Enrichment Source
Index Database
Publisher
StartPage 738
SubjectTerms Algorithm design and analysis
Costs
Delay
Hypercubes
Load management
Scalability
State-space methods
Traveling salesman problems
Tree graphs
Upper bound
Title Scalable global and local hashing strategies for duplicate pruning in parallel A graph search
URI https://ieeexplore.ieee.org/document/598348
https://www.proquest.com/docview/26399068
https://www.proquest.com/docview/28570696
Volume 8
WOSCitedRecordID wos10_1109_71_598348&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
journalDatabaseRights – providerCode: PRVIEE
  databaseName: IEEE Electronic Library (IEL)
  issn: 1045-9219
  databaseCode: RIE
  dateStart: 19900101
  customDbUrl:
  isFulltext: true
  dateEnd: 99991231
  titleUrlDefault: https://ieeexplore.ieee.org/
  omitProxy: false
  ssIdentifier: ssj0014504
  providerName: IEEE
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3PS8MwFA5ueNCD06k4fwbx4CVb0iVNcxRxeJAhqLCLlLRJcDC6sbX-_eZHOxSH4K2Ul7bk5fV7yUu-D4Abi4HU4QDKiMGIykQioYhChKrMKBNb1Pckrk98PE4mE_Fc82z7szBaa7_5TPfdpa_lq3leuaWyARPJkCYt0OI8Dke11gUDyrxSoJ1cMCRsFNYkQgSLASf90PAH9HgtlV8_YI8qo86_vmcf7NXJI7wL3j4AW7rogk4jzADrOO2C3W8sg4fg_cX6wZ2QgoH9A8pCQY9h8CNIKcFV2TBGQJvEQlWFqraGi2XlFk7gtICOJHw20_b10LNcwxAkR-Bt9PB6_4hqVQWUDzktkSuFsiyXIs65TU60xXCtaGKwHVNOZJAqEmGeG4E1Nywj1M4odC4lkbHgkWHDY9Au5oU-AdBkjsomE4JFirLEUYxiLhmV1AiaqaQHbpsOT_OactwpX8xSP_XAIuUkDZ3YA9dr00Xg2dhk1HVOWBs0d68aJ6Y2NlzBQxZ6Xq3SyKVfOP7LwvH7xyI-3fjkM7AT2Grd7txz0C6Xlb4A2_lnOV0tL_0A_ALYk9o4
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1LS8QwEB58gXrwsSq-DeLBS91mN2mao4iiuC6CCl6kpE2CwtKV3a2_30zSLooieCtl0pZMpt8kk3wfwInDQIY4EOXUxhFTqYqkpjqiTOdW28Shvidx7Yl-P31-lvc1z7Y_C2OM8ZvPzBle-lq-HhYVLpW1uUy7LJ2FeRTOqg9rTUsGjHutQDe94JF0cVjTCNFYtgU9C02_gY9XU_nxC_a4crX6ry9ag5U6fSTnwd_rMGPKFqw20gykjtQWLH_hGdyAlwfnCTwjRQL_B1GlJh7FyGsQUyLjScMZQVwaS3QV6tqGvI8qXDohbyVBmvDBwLjXE89zTUKYbMLT1eXjxXVU6ypERVewSYTFUJ4XSiaFcOmJcShuNEtt7EYVygwyTTuxKKyMjbA8p8zNKUyhFFWJFB3Lu1swVw5Lsw3E5khmk0vJO5rxFElGY6E4U8xKlut0B06bDs-KmnQctS8GmZ98xDITNAuduAPHU9P3wLTxm1ELnTA1aO4eNU7MXHRgyUOVZliNsw4mYHHylwUy_Ccy2f31yUeweP1418t6N_3bPVgK3LW4V3cf5iajyhzAQvExeRuPDv1g_ARaf92B
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Scalable+global+and+local+hashing+strategies+for+duplicate+pruning+in+parallel+A+graph+search&rft.jtitle=IEEE+transactions+on+parallel+and+distributed+systems&rft.au=Mahapatra%2C+Nihar+R&rft.au=Dutt%2C+Shantanu&rft.date=1997-07-01&rft.issn=1045-9219&rft.volume=8&rft.issue=7&rft.spage=738&rft.epage=756&rft_id=info:doi/10.1109%2F71.598348&rft.externalDBID=NO_FULL_TEXT
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1045-9219&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1045-9219&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1045-9219&client=summon