Performance evaluation of parallel numerical libraries

The use of multiple processors in parallel gives a remarkable improvement in performance as compared to that of a single processor. Unfortunately, it is still very difficult to write algorithms that actually take advantage of those multiple-processor machines. However a number of parallel libraries have been developed that perform reasonably well on wide range of parallel architectures. Examples of such libraries include SCALAPACK and PETSc. The aim is to study and to evaluate the performance of these parallel numerical libraries. So that if someone runs into a situation where he needs to choose between different available parallel libraries then he might be able to choose the library that gives best performance for his problem.