Research on F Statistic Parallel Algorithm Based on Gpu

The experimental data is soybean gene variation information file. The individual biological information is extracted from the soybean gene file and the value of the relationship between populations is computed. Then the degree of differentiation of the two populations of wild soybean and cultivated soybean is obtained. Firstly, the easy parallel serial algorithm is designed. Then according to the independence of the operational object, the partial algorithm can be paralleled is parallel processed with GPU multi thread technology. Finally, the result and the time of the serial / parallel algorithm are analyzed and compared. The experimental results show that the GPU algorithm efficiency is much higher than the serial algorithm efficiency. It provides a quick calculation scheme to handle massive biological data.