An Efficient Parallel Algorithm for Simpson Cumulative Integration on GPU

In this paper, we present an efficient parallel algorithm for calculating cumulative integration based on Simpson´s rule. The proposed parallel algorithm exploits two Blelloch´s prefix sums. The first scan is used to calculate even-index, while the second scan is used to calculate odd-index cumulative integration. We implement the parallel algorithm on NVIDIA CUDA based GPUs. Performance of the proposed parallel algorithm is measured by calculating speedup. We also present accuracy performance of the proposed algorithm. Based on the performance measurements, we can conclude that the parallel proposed algorithm is faster than optimized CPU codes with 3 times speedup.