The study on the static and dynamic load balancing algorithms has a history over three decades and it is stilla promising field. But because of the uncertainties between the dependencies of the parallel tasks and thei...The study on the static and dynamic load balancing algorithms has a history over three decades and it is stilla promising field. But because of the uncertainties between the dependencies of the parallel tasks and their communi-cation during the compile-time, researchers are more inclined to study the dynamic load balancing algorithms(DLB).There are almost four kinds of DLB algorithms including the centralized DLB, distributed DLB, global DLB and LocalDLB, all of them have their pros and cons. This paper addresses a new dynamic load balancing model based on theMain Load Information Table (MLIT) and its dynamic load balancing algorithm, it not only has the advantages thefour models above mentioned have, but it overcomes some of their disadvantages which lead to a poor performance,thus it boasts a better stability and security and in the end it can improve the performance of the system.展开更多
Dynamic distribution model is one of the best schemes for parallel volume rendering. How- ever, in homogeneous cluster system.since the granularity is traditionally identical, all processors communicate almost simulta...Dynamic distribution model is one of the best schemes for parallel volume rendering. How- ever, in homogeneous cluster system.since the granularity is traditionally identical, all processors communicate almost simultaneously and computation load may lose balance. Due to problems above, a dynamic distribution model with prime granularity for parallel computing is presented. Granularities of each processor are relatively prime, and related theories are introduced. A high parallel performance can be achieved by minimizing network competition and using a load balancing strategy that ensures all processors finish almost simultaneously. Based on Master-Slave-Gleaner ( MSG) scheme, the parallel Splatting Algorithm for volume rendering is used to test the model on IBM Cluster 1350 system. The experimental results show that the model can bring a considerable improvement in performance, including computation efficiency, total execution time, speed, and load balancing.展开更多
文摘The study on the static and dynamic load balancing algorithms has a history over three decades and it is stilla promising field. But because of the uncertainties between the dependencies of the parallel tasks and their communi-cation during the compile-time, researchers are more inclined to study the dynamic load balancing algorithms(DLB).There are almost four kinds of DLB algorithms including the centralized DLB, distributed DLB, global DLB and LocalDLB, all of them have their pros and cons. This paper addresses a new dynamic load balancing model based on theMain Load Information Table (MLIT) and its dynamic load balancing algorithm, it not only has the advantages thefour models above mentioned have, but it overcomes some of their disadvantages which lead to a poor performance,thus it boasts a better stability and security and in the end it can improve the performance of the system.
基金Supported by Natural Science Foundation of China ( No. 60373061).
文摘Dynamic distribution model is one of the best schemes for parallel volume rendering. How- ever, in homogeneous cluster system.since the granularity is traditionally identical, all processors communicate almost simultaneously and computation load may lose balance. Due to problems above, a dynamic distribution model with prime granularity for parallel computing is presented. Granularities of each processor are relatively prime, and related theories are introduced. A high parallel performance can be achieved by minimizing network competition and using a load balancing strategy that ensures all processors finish almost simultaneously. Based on Master-Slave-Gleaner ( MSG) scheme, the parallel Splatting Algorithm for volume rendering is used to test the model on IBM Cluster 1350 system. The experimental results show that the model can bring a considerable improvement in performance, including computation efficiency, total execution time, speed, and load balancing.
