Today's trend to use accelerators in heterogeneous systems forces a paradigm shift in programming models. The use of low-level APIs for accelerator programming is tedious and not intuitive for casual programmers. To tackle this problem, recent approaches focused on high-level directive-based models, with a standardization effort made with OpenACC and the directives for accelerator on OpenMP 4.0 release candidate. The pragmas for data management induce some coherence issues in the accelerator memory for code correctnesse. To address this issue, we propose the design for a directory, along with a reduced runtime ABI, to handle correctly data management in these standards. Our design fits a multi-Accelerator system. Also, with our directory, we propose a way to handle correctly pragmas on partially overlapping data intervals. © 2014 Springer-Verlag Berlin Heidelberg.
CITATION STYLE
Jaeger, J., Carribault, P., & Pérache, M. (2014). Data-management directory for OpenMP 4.0 and OpenACC. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 8374 LNCS, pp. 168–177). Springer Verlag. https://doi.org/10.1007/978-3-642-54420-0_17
Mendeley helps you to discover research relevant for your work.