Psta.psiras.ru/

The paper describes a parallel algorithm for searching for MD5 hashfunction collisions and its implementation with run results on a computational cluster. The parallel collision search program is implemented in C ++ language using the MPI library. The source code of the program is based on a consistent version of the program of the search for collisions from the Dutch scientist Mark Stevens. The author is sure that the parallelization algorithm can be used to develop efficient parallel programs for finding collisions of hash functions whose algorithm is based on the difference method of Wan. In the course of this study, using a high-performance cluster, a new pair of single-block messages is opened, MD5-digests of which coincide (they form a collision).