Performance analysis of graphical processors with calculation of sum of pairs score

The increasing power of the parallel computing capabilities and programming platforms that help the usage of the mentioned computing power of the graphical processing units have lead to emerge new applications to do with the modifications of the existing techniques for executing them on the graphical units. In this study, we investigated the computing capabilities of the Nvidia graphics cards powered by the Compute Unified Device Architecture (CUDA) programming platform for calculation of the Sum of Pairs score that is used to measure the similarity level between biological sequences. Experimental studies showed that the time required to obtain Sum of Pairs score was significantly decreased compared to the time required to obtain the same score on a conventional processor when the thread configuration is set by considering the properties of the graphics cards.