DocumentCode :
2501683
Title :
MPI as a coordination layer for communicating HPF tasks
Author :
Foster, Ian T. ; Krishnaiyer, Rakesh
Author_Institution :
Div. of Math. & Comput. Sci., Argonne Nat. Lab., IL
fYear :
1996
fDate :
1-2 Jul 1996
Firstpage :
68
Lastpage :
78
Abstract :
Data-parallel languages such as High Performance Fortran (HPF) present a simple execution model in which a single thread of control performs high-level operations on distributed arrays. These languages can greatly ease the development of parallel programs. Yet there are large classes of applications for which a mixture of task and data parallelism is most appropriate. Such applications can be structured as collections of data-parallel tasks that communicate by using explicit message passing. Because the Message Passing Interface (MPI) defines standardized, familiar mechanisms for this communication model, we propose that HPF tasks communicate by making calls to a coordination library that provides an HPF binding for MPI. The semantics of a communication interface for sequential languages can be ambiguous when the interface is invoked from a parallel language; we show how these ambiguities can be resolved by describing one possible HPF binding for MPI. We then present the design of a library that implements this binding, discuss issues that influenced our design decisions, and evaluate the performance of a prototype HPF/MPI library using a communications microbenchmark and application kernel. Finally, we discuss how MPI features might be incorporated into our design framework
Keywords :
FORTRAN; application program interfaces; libraries; message passing; parallel languages; parallel programming; software libraries; software performance evaluation; utility programs; High Performance Fortran task communication; MPI coordination layer; Message Passing Interface; application kernel; binding; communication interface semantics; communications microbenchmark; coordination library calls; data parallelism; data-parallel languages; distributed array; execution model; explicit message passing; high-level operations; library; parallel program development; performance evaluation; prototype HPF/MPI library; sequential languages; task parallelism; Communication system control; Distributed computing; High performance computing; Libraries; Mathematics; Message passing; Pipelines; Programmable logic arrays; Programming profession; Yarn;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
MPI Developer's Conference, 1996. Proceedings., Second
Conference_Location :
Notre Dame, IN
Print_ISBN :
0-8186-7533-0
Type :
conf
DOI :
10.1109/MPIDC.1996.534096
Filename :
534096
Link To Document :
بازگشت