Font Size: a A A

The Development Of Distributed Cluster Parallel Simulation Platform Network-based

Posted on:2004-06-03Degree:MasterType:Thesis
Country:ChinaCandidate:Y J XiaoFull Text:PDF
GTID:2168360095956931Subject:Thermal Engineering
Abstract/Summary:PDF Full Text Request
This paper thoroughly analysed the technique of distributed parallel computing and exploied a distributed cluster parallel simulation platform's network-based.We colligate all kinds of knowledge on the distributed computing,parallel computing,representative cluster task manage system and bring forward a platform architecture.This kind of network-based distributed cluster parallel simulation architecture consists of server,node computers.The load balance is a very pivotal question in the cluster computation. A consummate node load computing method is put forward after synthesizing various of load index literature. By using the arithmetic of dynamic adjustive load banlance and introducing the concept of double thresholds, every node can rapidly transfer load in weight load and in idlesses or light load it can rapidly transfer load. The cluster manage system can use GUI tools to keep track of the whole cluster state from vary arrangement. It is very important for a cluster of high performance computing platform to have a good manage sofeware. This paper exploies a system manage platform SIMNOWs that is the part user directly face to. It implemented the system design,task attemper,distribute and unite manage and so on.The whole system manage platform include:system whole function module,task management function module,node management function module,communication management function module.The simulation computing of circulate sulfuration bed's model is a big task's compution program,it is difficult to realize in generic compution environment.At first we run simulation program in server and then we run the same program in the SIMNOWs. We compare the two compution environment and draw a conclusion that the second one has a good efficiency.
Keywords/Search Tags:parallel computing, cluster computing, load balance
PDF Full Text Request
Related items