Extension of the DVM parallel programming model for clusters with heterogeneous nodes

Authors

V.A. Bakhtin, B.N. Chetverushkin, V.A. Krukov, E. V. Shilnikov

Annotation

The HMPP, hiCUDA, and PGI_APM highlevel models for GPUs developed in recent years are studied; a new model DVMH (designed at Keldysh Institute of Applied Mathemat ics, Russian Academy of Sciences) for clusters with heterogeneous nodes  and extensions of the FORTRAN DVM language and of the support system LibDVM for parallel implementation of programs are described. A study of characteristics of test and reallife applications are performed.

Language

English

Reference

V.A. Bakhtin, B.N. Chetverushkin, V.A. Krukov, E. V. Shilnikov. Extension of the DVM parallel programming model for clusters with heterogeneous nodes // DOKLADY MATHEMATICS , Moscow: Pleiades Publishing, Ltd, 2011, Vol. 84, Issue 3, P. 879-881