摘要

We describe a new design of parallel algorithm for solving the two-dimensional longest common substring (2D LCS) problem, taking advantage of the multi-core graphic processing unit architecture offered by Compute Unified Device Architecture (CUDA). In this article we also define the 2D LCS problem as finding the largest common 4-connected component from two input matrices and present an algorithm which can exactly solve this problem in 0 (mnst/P) time with a P-core GPU.

全文