Mixed Language Programming for High-Performance Computing Applications

Date Added: Jan 2010
Format: PDF

First, different styles of programming in the languages Fortran 77, C, C++, and Python are reviewed and exemplified in the context of representing and computing with sparse matrices. Special emphasis is paid to a promising multi-language environment based on using Python for administrative tasks and Fortran or C/C++ for the most CPU-critical operations. Examples on building similar mixed language, high-performance computing codes for solving systems of partial differential equations are presented. Relevant performance studies of various programming styles constitute an integrated part of the talk.