Parallel computation for non-equilibrium dynamics equations using shared memory programming model

OpenMP is one of the major parallel programming models for modern cluster with multi-core CPU,which can obtain speedup on multi-cores on one CPU.But the problem of scalability must be solved on the whole cluster system.Based on the distributed shared memory cluster,this paper presented an OpenMP parallel programming method with explicit data distribution for non-equilibrium dynamics equations.The method first partitioned the whole data into small blocks,then put the blocks to the OpenMP threads and tranferred data between the threads using the copies in the shared memory.Using the method of explicit data distribution,the authors designed the parallel programs of non-equilibrium dynamics equations.The test results show that the OpenMP parallel programs have good scalability while maintaining the readability on the distributed shared memory cluster with Intel Xeon quad-cores processors,and the parallel speedup relative to 64 is 5.61 on 1024 CPU cores.