Recent High Performance Computing (HPC) platforms can be classified as Shared-Memory Multiprocessors (SMP), Massively Parallel Processors (MPP), and Clusters of computing nodes. These platforms are deployed in many scientific and engineering applications which require very high demand on computing power. In order to realize an optimal performance for these applications, it is crucial to find and use the suitable computing platforms and programming paradigms. In this paper, we use SPEC HPC 2002 benchmark suite developed in various parallel programming models (MPI, OpenMP, and hybrid of MPI/OpenMP) to find an optimal computing environments and programming paradigms for them through their performance analyses.