Write-Avoiding Algorithms

Communication, i.e., moving data between levels of a memory hierarchy or between processors over a network, is much more expensive (in time or energy) than arithmetic. There has thus been a recent focus on designing algorithms that minimize communication and, when possible, attain lower bounds on th...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Proceedings - IEEE International Parallel and Distributed Processing Symposium s. 648 - 658
Hlavní autoři: Carson, Erin, Demmel, James, Grigori, Laura, Knight, Nicholas, Koanantakool, Penporn, Schwartz, Oded, Simhadri, Harsha Vardhan
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 01.05.2016
Témata:
ISSN:1530-2075
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Abstract Communication, i.e., moving data between levels of a memory hierarchy or between processors over a network, is much more expensive (in time or energy) than arithmetic. There has thus been a recent focus on designing algorithms that minimize communication and, when possible, attain lower bounds on the total number of reads and writes. However, most previous work does not distinguish between the costs of reads and writes. Writes can be much more expensive than reads in some current and emerging storage devices such as nonvolatile memories. This motivates us to ask whether there are lower bounds on the number of writes that certain algorithms must perform, and whether these bounds are asymptotically smaller than bounds on the sum of reads and writes together. When these smaller lower bounds exist, we then ask when they are attainable, we call such algorithms "write-avoiding" (WA), to distinguish them from "communication-avoiding" (CA) algorithms, which only minimize the sum of reads and writes. We identify a number of cases in linear algebra and direct N-body methods where known CA algorithms are also WA (some are and some aren't). We also identify classes of algorithms, including Strassen's matrix multiplication, Cooley-Tukey FFT, and cache oblivious algorithms for classical linear algebra, where a WA algorithm cannot exist: the number of writes is unavoidably within a constant factor of the total number of reads and writes. We explore the interaction of WA algorithms with cache replacement policies and argue that the Least Recently Used policy works well with the WA algorithms in this paper. We provide empirical hardware counter measurements from Intel's Nehalem-EX microarchitecture to validate our theory. In the parallel case, for classical linear algebra, we show that it is impossible to attain lower bounds both on interprocessor communication and on writes to local memory, but either one is attainable by itself. Finally, we discuss WA algorithms for sparse iterative linear algebra.
AbstractList Communication, i.e., moving data between levels of a memory hierarchy or between processors over a network, is much more expensive (in time or energy) than arithmetic. There has thus been a recent focus on designing algorithms that minimize communication and, when possible, attain lower bounds on the total number of reads and writes. However, most previous work does not distinguish between the costs of reads and writes. Writes can be much more expensive than reads in some current and emerging storage devices such as nonvolatile memories. This motivates us to ask whether there are lower bounds on the number of writes that certain algorithms must perform, and whether these bounds are asymptotically smaller than bounds on the sum of reads and writes together. When these smaller lower bounds exist, we then ask when they are attainable, we call such algorithms "write-avoiding" (WA), to distinguish them from "communication-avoiding" (CA) algorithms, which only minimize the sum of reads and writes. We identify a number of cases in linear algebra and direct N-body methods where known CA algorithms are also WA (some are and some aren't). We also identify classes of algorithms, including Strassen's matrix multiplication, Cooley-Tukey FFT, and cache oblivious algorithms for classical linear algebra, where a WA algorithm cannot exist: the number of writes is unavoidably within a constant factor of the total number of reads and writes. We explore the interaction of WA algorithms with cache replacement policies and argue that the Least Recently Used policy works well with the WA algorithms in this paper. We provide empirical hardware counter measurements from Intel's Nehalem-EX microarchitecture to validate our theory. In the parallel case, for classical linear algebra, we show that it is impossible to attain lower bounds both on interprocessor communication and on writes to local memory, but either one is attainable by itself. Finally, we discuss WA algorithms for sparse iterative linear algebra.
Author Simhadri, Harsha Vardhan
Knight, Nicholas
Carson, Erin
Koanantakool, Penporn
Schwartz, Oded
Demmel, James
Grigori, Laura
Author_xml – sequence: 1
  givenname: Erin
  surname: Carson
  fullname: Carson, Erin
  email: erin.carson@nyu.edu
  organization: Courant Inst. of Math. Sci., New York Univ., New York, NY, USA
– sequence: 2
  givenname: James
  surname: Demmel
  fullname: Demmel, James
  email: demmel@berkeley.edu
  organization: Dept. of Math., Univ. of California, Berkeley, Berkeley, CA, USA
– sequence: 3
  givenname: Laura
  surname: Grigori
  fullname: Grigori, Laura
  email: laura.grigori@inria.fr
  organization: INRIA Paris-Rocquencourt, UPMC - Univ. Paris 6, Paris, France
– sequence: 4
  givenname: Nicholas
  surname: Knight
  fullname: Knight, Nicholas
  email: nknight@nyu.edu
  organization: Courant Inst. of Math. Sci., New York Univ., New York, NY, USA
– sequence: 5
  givenname: Penporn
  surname: Koanantakool
  fullname: Koanantakool, Penporn
  email: penpornk@eecs.berkeley.edu
  organization: Comput. Sci. Div., Univ. of California, Berkeley, Berkeley, CA, USA
– sequence: 6
  givenname: Oded
  surname: Schwartz
  fullname: Schwartz, Oded
  email: odedsc@cs.huji.ac.il
  organization: Sch. of Eng. & Comput. Sci., Hebrew Univ. of Jerusalem, Jerusalem, Israel
– sequence: 7
  givenname: Harsha Vardhan
  surname: Simhadri
  fullname: Simhadri, Harsha Vardhan
  email: harshas@lbl.gov
  organization: Comput. Res. Div., Lawrence Berkeley Nat. Lab., Berkeley, CA, USA
BookMark eNotjF1LAkEUQCcwSM3XIHrpD4zdO3c-9j4uliUIChY9yjp7xyZ0N3Yl6N8n1NPhnIczUoOmbUSpG4QpIvDDYv243kwNoD-7vVATDgU6YDBowQ_UEB2BNhDclRr1_SeAAbI8VLfvXT6JLr_bXOdmf18e9u25fBz7a3WZqkMvk3-O1dv86XX2oper58WsXOpsoDhpF2vPgdNOhGNEDJSEWYpEpnJcWS-RIIKREAViTFGk3rlEZImisNBY3f19s4hsv7p8rLqfbXDowSP9AhnPPaM
CODEN IEEPAD
ContentType Conference Proceeding
DBID 6IE
6IL
CBEJK
RIE
RIL
DOI 10.1109/IPDPS.2016.114
DatabaseName IEEE Electronic Library (IEL) Conference Proceedings
IEEE Xplore POP ALL
IEEE Xplore All Conference Proceedings
IEEE Electronic Library (IEL)
IEEE Proceedings Order Plans (POP All) 1998-Present
DatabaseTitleList
Database_xml – sequence: 1
  dbid: RIE
  name: IEEE Electronic Library (IEL)
  url: https://ieeexplore.ieee.org/
  sourceTypes: Publisher
DeliveryMethod fulltext_linktorsrc
Discipline Computer Science
EISBN 9781509021406
150902140X
EndPage 658
ExternalDocumentID 7516061
Genre orig-research
GroupedDBID 29O
6IE
6IF
6IH
6IK
6IL
6IN
AAJGR
AAWTH
ABLEC
ADZIZ
ALMA_UNASSIGNED_HOLDINGS
BEFXN
BFFAM
BGNUA
BKEBE
BPEOZ
CBEJK
CHZPO
IEGSK
IPLJI
OCL
RIE
RIL
ID FETCH-LOGICAL-i208t-5cd6979fbee9cc1173fe99e8f32a59a46ec30c02e7ce0ccfceedb5f33433ce9e3
IEDL.DBID RIE
ISICitedReferencesCount 24
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000391251800067&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 1530-2075
IngestDate Wed Aug 27 02:11:36 EDT 2025
IsPeerReviewed false
IsScholarly false
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-i208t-5cd6979fbee9cc1173fe99e8f32a59a46ec30c02e7ce0ccfceedb5f33433ce9e3
PageCount 11
ParticipantIDs ieee_primary_7516061
PublicationCentury 2000
PublicationDate 20160501
PublicationDateYYYYMMDD 2016-05-01
PublicationDate_xml – month: 05
  year: 2016
  text: 20160501
  day: 01
PublicationDecade 2010
PublicationTitle Proceedings - IEEE International Parallel and Distributed Processing Symposium
PublicationTitleAbbrev IPDPS
PublicationYear 2016
Publisher IEEE
Publisher_xml – name: IEEE
SSID ssj0020349
Score 1.7562889
Snippet Communication, i.e., moving data between levels of a memory hierarchy or between processors over a network, is much more expensive (in time or energy) than...
SourceID ieee
SourceType Publisher
StartPage 648
SubjectTerms Algorithm design and analysis
communication avoiding algorithms
Hardware
Krylov subspace methods
Linear algebra
Load modeling
lower bounds
N-body methods
Non-volatile memories
Nonvolatile memory
Program processors
Schedules
write complexity
Title Write-Avoiding Algorithms
URI https://ieeexplore.ieee.org/document/7516061
WOSCitedRecordID wos000391251800067&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
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1LSwMxEB5q8eCpaivWF3vwaGw22U02x-IDBSkLPuitdGcnWtCutNv-fpPdtSJ48RaGIa8hTCb55huAc-surSZSHioVaRbZJGfOjyuW58IaSaRkhap8edCjUTIem7QFF5tcGCKqwGd06ZvVX35e4Mo_lQ2064X7WGdLa1Xnam2CK8-zUnOjcmd5HTcEjSE3g_v0On30KC7lmXF_lVGpvMht53_j70LvJx0vSDeOZg9aNN-Hznc9hqA5nl3ou1C7JDZcFzOvGAzfXwsneftY9uD59ubp6o41pQ_YTPCkZDHmymhjMyKDGIZaWjKGEivFNDbTSBFKjlyQRuKI1s8gi62UkZRIhuQBtOfFnA4hEIl16rGIJHoiG5tJYWONGQptQ2WTPnT9SiefNbvFpFnk0d_iY9jxG1lD_k6gXS5WdArbuC5ny8VZZZIvn26LGA
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3dS8MwED-GCvo0dRPnZx98tC7NZ_M41LHhHAWn7G2s6UUHusrW7e-3aetE8MW3cIRwlyNcLvnd7wCubH5p1Vw6qBRXPrdh4udxXPpJQq1miJIVqMqXgRoOw_FYRzW43tTCIGIBPsMbNyz-8pPUrNxTWVvlqxCX62wLzikpq7U26ZVjWinZUUnueyUqisaA6HY_uoueHI5LOm7cX41UijjSrf9Pg31o_hTkedEm1BxADeeHUP_uyOBVB7QBrTzZztDvrNOZm-h13l_TXPL2sWzCc_d-dNvzq-YH_oySMPOFSaRW2saI2pggUMyi1hhaRqdCT7lEw4ghFJVBYox1GsTCMsYZM6iRHcHWPJ3jMXg0tPl0QTkzjsrGxoxaoUxsqLKBtGELGs7SyWfJbzGpjDz5W3wJu73R42Ay6A8fTmHPbWoJADyDrWyxwnPYMetstlxcFO75AhA5jl8
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%3Abook&rft.genre=proceeding&rft.title=Proceedings+-+IEEE+International+Parallel+and+Distributed+Processing+Symposium&rft.atitle=Write-Avoiding+Algorithms&rft.au=Carson%2C+Erin&rft.au=Demmel%2C+James&rft.au=Grigori%2C+Laura&rft.au=Knight%2C+Nicholas&rft.date=2016-05-01&rft.pub=IEEE&rft.issn=1530-2075&rft.spage=648&rft.epage=658&rft_id=info:doi/10.1109%2FIPDPS.2016.114&rft.externalDocID=7516061
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1530-2075&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1530-2075&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1530-2075&client=summon