FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2732567711> ?p ?o ?g. }

• W2732567711 abstract "Performance portability of computer programs, andprogrammer productivity in writing them are key expectations insoftware engineering. These expectations lead to the followingquestions: Can programmers write code once, and execute it atoptimal speed on any machine configuration? Can programmers writeparallel code to simple models that hide the complex details ofparallel programming? This thesis addresses these questions forcertain classes of computer programs. It describes autotuningtechniques that achieve performance portability for serialdivide-and-conquer programs, and an abstraction that improvesprogrammer productivity in writing parallel code for a class ofprograms called Star. We present a pruned-exhaustive autotunercalled Ztune that optimizes the performance of serialdivide-and-conquer programs for a given machine configuration.Whereas the traditional way of autotuning divide-and-conquerprograms involves simply coarsening the base case of recursionoptimally, Ztune searches for optimal divide-and-conquer trees.Although Ztune, in principle, exhaustively enumerates the searchdomain, it uses pruning properties that greatly reduce the size ofthe search domain without significantly sacrificing the quality ofthe autotuned code. We illustrate how to autotunedivide-and-conquer stencil computations using Ztune, and presentperformance comparisons with state-of-the-art heuristicautotuning. Not only does Ztune autotune significantly faster thana heuristic autotuner, the Ztuned programs also run faster onaverage than their heuristic autotuner tuned counterparts.Surprisingly, for some stencil benchmarks, Ztune actually autotunedfaster than the time it takes to execute the stencil computationonce. We introduce the Star class that includes many seeminglydifferent programs like solving symmetric, diagonally-dominanttridiagonal systems, executing cuts on graphs, samplesort, fast multipole computations, and all-prefix-sums and itsvarious applications. We present a programming model, which is alsocalled Star, to generate and execute parallel code for the Starclass of programs. The Star model abstracts the pattern ofcomputation and interprocessor communication in the Star class ofprograms, hides low-level parallel programming details, and offersease of expression, thereby improving programmer productivity inwriting parallel code. Besides, we also present parallelalgorithms, which offer asymptotic improvements over prior art, fortwo programs in the Star class - a Trip algorithm for solvingsymmetric, diagonally-dominant tridiagonal systems, and a Waspalgorithm for executing watershed cuts on graphs. The Star model isimplemented in the Julia programming language, and leveragesJulia's capabilities in expressing parallelism in code concisely,and in supporting both shared-memory and distributed-memoryparallel programming alike." @default.
• W2732567711 created "2017-07-14" @default.
• W2732567711 creator A5069582736 @default.
• W2732567711 date "2017-01-01" @default.
• W2732567711 modified "2023-09-28" @default.
• W2732567711 title "Portable and productive high-performance computing" @default.
• W2732567711 cites W1504320321 @default.
• W2732567711 cites W1513400187 @default.
• W2732567711 cites W1520821748 @default.
• W2732567711 cites W1525394366 @default.
• W2732567711 cites W1567510187 @default.
• W2732567711 cites W1570895503 @default.
• W2732567711 cites W1582222436 @default.
• W2732567711 cites W1595783387 @default.
• W2732567711 cites W1649645444 @default.
• W2732567711 cites W1788180225 @default.
• W2732567711 cites W1963695588 @default.
• W2732567711 cites W1968013322 @default.
• W2732567711 cites W1968347730 @default.
• W2732567711 cites W1979457157 @default.
• W2732567711 cites W1986022261 @default.
• W2732567711 cites W1989741444 @default.
• W2732567711 cites W1993879355 @default.
• W2732567711 cites W1997147891 @default.
• W2732567711 cites W2003798513 @default.
• W2732567711 cites W2017838066 @default.
• W2732567711 cites W2028067303 @default.
• W2732567711 cites W2034761517 @default.
• W2732567711 cites W2041097097 @default.
• W2732567711 cites W2045992017 @default.
• W2732567711 cites W2055253125 @default.
• W2732567711 cites W2059710204 @default.
• W2732567711 cites W2081422781 @default.
• W2732567711 cites W2084423328 @default.
• W2732567711 cites W2095875205 @default.
• W2732567711 cites W2099625934 @default.
• W2732567711 cites W2100218206 @default.
• W2732567711 cites W2102182691 @default.
• W2732567711 cites W2104512032 @default.
• W2732567711 cites W2113190809 @default.
• W2732567711 cites W2118746782 @default.
• W2732567711 cites W2119042753 @default.
• W2732567711 cites W2120833345 @default.
• W2732567711 cites W2121082877 @default.
• W2732567711 cites W2124931755 @default.
• W2732567711 cites W2128000934 @default.
• W2732567711 cites W2128249697 @default.
• W2732567711 cites W2129471558 @default.
• W2732567711 cites W2131975293 @default.
• W2732567711 cites W2132536544 @default.
• W2732567711 cites W2135653967 @default.
• W2732567711 cites W2135682468 @default.
• W2732567711 cites W2136628731 @default.
• W2732567711 cites W2154786353 @default.
• W2732567711 cites W2154894831 @default.
• W2732567711 cites W2163491234 @default.
• W2732567711 cites W2164197394 @default.
• W2732567711 cites W2173213060 @default.
• W2732567711 cites W2306558989 @default.
• W2732567711 cites W2575060280 @default.
• W2732567711 cites W2750218732 @default.
• W2732567711 cites W2963242393 @default.
• W2732567711 cites W76123042 @default.
• W2732567711 hasPublicationYear "2017" @default.
• W2732567711 type Work @default.
• W2732567711 sameAs 2732567711 @default.
• W2732567711 citedByCount "0" @default.
• W2732567711 crossrefType "dissertation" @default.
• W2732567711 hasAuthorship W2732567711A5069582736 @default.
• W2732567711 hasConcept C154945302 @default.
• W2732567711 hasConcept C168773036 @default.
• W2732567711 hasConcept C173608175 @default.
• W2732567711 hasConcept C173801870 @default.
• W2732567711 hasConcept C177264268 @default.
• W2732567711 hasConcept C199360897 @default.
• W2732567711 hasConcept C2776760102 @default.
• W2732567711 hasConcept C2777212361 @default.
• W2732567711 hasConcept C2778514511 @default.
• W2732567711 hasConcept C41008148 @default.
• W2732567711 hasConcept C459310 @default.
• W2732567711 hasConcept C63000827 @default.
• W2732567711 hasConcept C71559656 @default.
• W2732567711 hasConcept C76752949 @default.
• W2732567711 hasConcept C80444323 @default.
• W2732567711 hasConceptScore W2732567711C154945302 @default.
• W2732567711 hasConceptScore W2732567711C168773036 @default.
• W2732567711 hasConceptScore W2732567711C173608175 @default.
• W2732567711 hasConceptScore W2732567711C173801870 @default.
• W2732567711 hasConceptScore W2732567711C177264268 @default.
• W2732567711 hasConceptScore W2732567711C199360897 @default.
• W2732567711 hasConceptScore W2732567711C2776760102 @default.
• W2732567711 hasConceptScore W2732567711C2777212361 @default.
• W2732567711 hasConceptScore W2732567711C2778514511 @default.
• W2732567711 hasConceptScore W2732567711C41008148 @default.
• W2732567711 hasConceptScore W2732567711C459310 @default.
• W2732567711 hasConceptScore W2732567711C63000827 @default.
• W2732567711 hasConceptScore W2732567711C71559656 @default.
• W2732567711 hasConceptScore W2732567711C76752949 @default.
• W2732567711 hasConceptScore W2732567711C80444323 @default.
• W2732567711 hasLocation W27325677111 @default.

• next