Design for Interoperability in STAPL: pMatrices and Linear Algebra Algorithms
Conference Paper
Overview
Research
Identity
Additional Document Info
Other
View All
Overview
abstract
The Standard Template Adaptive Parallel Library (stapl) is a high-productivity parallel programming framework that extends C++ and stl with unified support for shared and distributed memory parallelism. stapl provides distributed data structures (pContainers) and parallel algorithms (pAlgorithms) and a generic methodology for extending them to provide customized functionality. To improve productivity and performance, it is essential for stapl to exploit third party libraries, including those developed in programming languages other than C++. In this paper we describe a methodology that enables third party libraries to be used with stapl. This methodology allows a developer to specify when these specialized libraries can correctly be used, and provides mechanisms to transparently invoke them when appropriate. It also provides support for using stapl pAlgorithms and pContainers in external codes. As a concrete example, we illustrate how third party libraries, namely BLAS and PBLAS, can be transparently embedded into stapl to provide efficient linear algebra algorithms for the stapl pMatrix, with negligible slowdown with respect to the optimized libraries themselves. 2008 Springer Berlin Heidelberg.
name of conference
Languages and Compilers for Parallel Computing, 21th International Workshop, LCPC 2008, Edmonton, Canada, July 31 - August 2, 2008, Revised Selected Papers
