high performance computing on graphics processing units: hgpu.org

The paper presents the parallel algorithm for solving the scheduling problem. This algorithm is implemented in the distributed memory multi-computers, and with each machine using CPU – GPU shared memory architecture, so that the time to complete the work as quickly as possible. This algorithm is based on the branching algorithm approach for searching. The experimental results for the scheduling problem were calculated with large data. From that determines the threshold of input data of the problem in order to the computation time is minimum.