Symbolic verification and error prediction methodology

A SIMD platform provides higher computing power by executing multiple data simultaneously, but this feature is making the design and verification harder. This paper aims at helping designer with an efficient methodology for verifying and evaluating the performance of a SIMD PLX platform design. Using Mathematica, a computer algebra system (CAS), and reverse engineering techniques, the correctness of and errors accumulated in running multimedia applications on this PLX platform can be precisely evaluated, which would not be easily done without human interventions before. The proposed methodology can be easily automated and adapted to other SIMD platforms.