In order to reach challenging performance goals, computer architectures willchange significantly in the next future. Heterogeneous chips, equipped with different types of cores and memory will compel application developers to deal with irregularcommunication patterns, high parallelism, and unexpected behaviors. Load balancing among the heterogeneous compute units will be a critical task in order to exploit all the computational power providedby such new architectures. In this highly dynamic scenario, Partitioned Global Address Space (PGAS) languages, like Coarray Fortran (CAF), appear to be a promising alternativeto standard MPI programming using two-sided communications, in particularbecause of their one-sided semantic. In this work, we show how Coarray Fortran can be used for implementingdynamic load balancing algorithms on an exascale compute node and how these algorithms can produce performancebenefits for an Asian option pricing problem, running in symmetricmode on Intel Xeon Phi (KNC).
Cardellini, V., Fanfarillo, A., Filippone, S. (2016). Heterogeneous CAF-based load balancing on Intel Xeon Phi. In Proceedings of the 2016 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW 2016) (pp.702-711). IEEE Computer Society [10.1109/IPDPSW.2016.51].
Heterogeneous CAF-based load balancing on Intel Xeon Phi
CARDELLINI, VALERIA;FANFARILLO, ALESSANDRO;Filippone, S.
2016-01-01
Abstract
In order to reach challenging performance goals, computer architectures willchange significantly in the next future. Heterogeneous chips, equipped with different types of cores and memory will compel application developers to deal with irregularcommunication patterns, high parallelism, and unexpected behaviors. Load balancing among the heterogeneous compute units will be a critical task in order to exploit all the computational power providedby such new architectures. In this highly dynamic scenario, Partitioned Global Address Space (PGAS) languages, like Coarray Fortran (CAF), appear to be a promising alternativeto standard MPI programming using two-sided communications, in particularbecause of their one-sided semantic. In this work, we show how Coarray Fortran can be used for implementingdynamic load balancing algorithms on an exascale compute node and how these algorithms can produce performancebenefits for an Asian option pricing problem, running in symmetricmode on Intel Xeon Phi (KNC).I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
