OpenMP static TDG runtime implementation and its usage in heterogeneous computing
Document typeConference report
PublisherBarcelona Supercomputing Center
Rights accessOpen Access
All rights reserved. This work is protected by the corresponding intellectual and industrial property rights. Without prejudice to any existing legal exemptions, reproduction, distribution, public communication or transformation of this work are prohibited without permission of the copyright holder
OpenMP being the standard to use in shared memory parallel programming, it offers the possibility to parallelize sequential program with accelerators by using target directive. However, CUDA Graph as a new, efficient feature is not supported yet. In this work, we present an automatic transformation of OpenMP TDG to CUDA Graph, increasing the programmability of the latter.
CitationYu, C.; Royuela Alcázar, S.; Quiñones, E. OpenMP static TDG runtime implementation and its usage in heterogeneous computing. A: . Barcelona Supercomputing Center, 2020, p. 39-40.
|BSC_SODS-20-14_OpenMP static TDG runtime.pdf||103,2Kb||View/Open|