Comparison of performance-enhancing strategies for parallel numerical object-oriented frameworks

摘要

Performance short of that of C or FORTRAN 77 is a significant obstacle to general acceptance of object-oriented C++ frameworks in high-performance parallel scientific computing, nonetheless, their value in simplifying complex computations is inarguable. Singular data points of good performance for object- oriented libraries/frameworks have been interesting, but a systematic analysis of the performance issues has not been done. This paper explores just a few of these issues and reports on the use of three mechanisms for enhancing the performance of object-oriented frameworks within numerical computation. The first is the commonly-use of binary overloaded operators (though implemented with substantial internal optimizations), the second is the use of expression templates, and the third is the use of an optimizing preprocessor. The first two have been completely implemented and are available within the A++/P++ array class library, the third, ROSE++, represents work in progress. This paper provides some perspective on the types of optimizations that the authors consider important within their numerical applications using OVERTURE involving complex geometry and AMR on parallel architectures.