000208036 001__ 208036
000208036 005__ 20180913063119.0
000208036 020__ $$a978-3-319-09581-3
000208036 0247_ $$2doi$$a10.1007/978-3-319-09581-3_4
000208036 037__ $$aCONF
000208036 245__ $$aDisjoint-Access Parallelism Does Not Entail Scalability
000208036 269__ $$a2014
000208036 260__ $$c2014
000208036 336__ $$aConference Papers
000208036 520__ $$aDisjoint Access Parallelism (DAP) stipulates that operations involving disjoint sets of memory words must be able to progress independently, without interfering with each other. In this work we argue towards revising the two decade old wisdom saying that DAP is a binary condition that splits concurrent programs into scalable and non-scalable. We first present situations where DAP algorithms scale poorly, thus showing that not even algorithms that achieve this property provide scalability under all circumstances. Next, we show that algorithms which violate DAP can sometimes achieve the same scalability and performance as their DAP counterparts. We continue to show how by violating DAP and without sacrificing scalability we are able to circumvent three theoretical results showing that DAP is incompatible with other desirable properties of concurrent programs. Finally we introduce a new property called generalized disjoint-access parallelism (GDAP) which estimates how much of an algorithm is DAP. Algorithms having a large DAP part scale similar to DAP algorithms while not being subject to the same impossibility results.
000208036 700__ $$0240335$$aGuerraoui, Rachid$$g105326
000208036 700__ $$0242990$$aLetia, Mihai$$g194745
000208036 7112_ $$aSecond International Conference, NETYS$$cMarrakech, Morocco$$dMay 15-17, 2014
000208036 773__ $$q41-56$$tProceedings of the Second International Conference, NETYS
000208036 8564_ $$s431323$$uhttps://infoscience.epfl.ch/record/208036/files/Disjoint_F978-3-319-09581-3_4.pdf$$yPreprint$$zPreprint
000208036 909C0 $$0252114$$pDCL$$xU10407
000208036 909CO $$ooai:infoscience.tind.io:208036$$pconf$$pIC
000208036 917Z8 $$x166927
000208036 937__ $$aEPFL-CONF-208036
000208036 973__ $$aEPFL$$rREVIEWED$$sPUBLISHED
000208036 980__ $$aCONF