This was my final year Research & Development individual project and From this research work I was used the massively parallel architecture of GPU as a parallel computing solution, for the improvement of one of the computationally intensive general purpose applications. With that concept I was able to improve dynamic programming based Global genetic Sequence Alignment algorithm with efficient use of heterogeneous memory hierarchy of GPU & CPU. For this research work, CUDA was used because it provides a parallel programming model and an instruction set architecture for Nvidia GPU with an extension of C language. With this work we use two levels of implementations for Needleman-Wunsch Global sequence alignment algorithm. First implementation was based on global memory of GPU and the other one is based on the fast, on-chip, Shared memory of GPU. Use of shared memory improved the performance of the implementation with the blocking strategy for the Needleman-Wunsch algorithm. Rea
(Knowledge + Technology + Life)