2013 42nd International Conference on Parallel Processing (2013)
Oct. 1, 2013 to Oct. 4, 2013
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/ICPP.2013.86
Thierry Gautier , INRIA, Sophia-Antipolis, France
Fabien Lementec , INRIA, Sophia-Antipolis, France
Vincent Faucher , DEN, CEA, Gif-sur-Yvette, France
Bruno Raffin , INRIA, Sophia-Antipolis, France
The paper presents X-KAAPI, a compact runtime for multicore architectures that brings multi parallel paradigms (parallel independent loops, fork-join tasks and dataflow tasks) in a unified framework without performance penalty. Comparisons on independent loops with OpenMP and on dense linear algebra with QUARK/PLASMA confirm our design decisions. Applied to EUROPLEXUS, an industrial simulation code for fast transient dynamics, we show that X-KAAPI achieves high speedups on multicore architectures by efficiently parallelizing both independent loops and dataflow tasks.
Runtime, Multicore processing, Plasmas, Load modeling, Computational modeling, Instruction sets, Adaptation models
T. Gautier, F. Lementec, V. Faucher and B. Raffin, "X-kaapi: A Multi Paradigm Runtime for Multicore Architectures," 2013 42nd International Conference on Parallel Processing(ICPP), Lyon France, 2014, pp. 728-735.