Search

With the expansion and the increasing complexity of computer simulation, high-latency is increasingly becoming the bottleneck limiting the scale of the simulation. On the one hand, high-performance computers and clusters are involved in this area, which have already largely alleviated the hard situation of computer simulation. On the other hand, as one of the simulation standards, HLA simulation was greatly applied in computer simulation. In HLA simulation domain, many RTIs (The Runtime Infrastructure) are designed to support the simulation in LAN/WAN environment. However, traditional RTIs running on high-performance computer are weakly taking advantage of the high speed network and the architecture of the high-performance computers. Therefore, for the problem, this paper wants to design an adaptive communication mechanism based on functional distributed architecture of RTI, which can adapt communication between different RTI components to shared memory, Infiniband network and Ethernet network automatically, thus can greatly improve communication performance. [ABSTRACT FROM PUBLISHER]

Parallel and distributed simulation system is usually composed of a large number of simulation entities, and a large number of interactions exist between these entities. This makes communication performance one of the most important factors which affect the simulation performance. To improve the communication performance on high performance computers, Message Passing Interface (MPI) is usually a preferred choice. In this paper, a new process merge method is proposed to implement the dynamic MPI process management, and message exchanging mechanism is presented. A flow control mechanism is proposed to improve the network bandwidth utilization. By the flow control MPI communication mechanism, the simulation performance can be improved. Based on the mechanism, a high speed communication library using MPI communication is designed. [ABSTRACT FROM PUBLISHER]