Broadcast scheduling optimization for heterogeneous cluster systems
Journal of Algorithms
Efficient Gather Operation in Heterogeneous Cluster Systems
HPCS '02 Proceedings of the 16th Annual International Symposium on High Performance Computing Systems and Applications
ISCC '03 Proceedings of the Eighth IEEE International Symposium on Computers and Communications
Complexity Results and Heuristics for Pipelined Multicast Operations on Heterogeneous Platforms
ICPP '04 Proceedings of the 2004 International Conference on Parallel Processing
Pipelining Broadcasts on Heterogeneous Platforms
IEEE Transactions on Parallel and Distributed Systems
Broadcast Trees for Heterogeneous Platforms
IPDPS '05 Proceedings of the 19th IEEE International Parallel and Distributed Processing Symposium (IPDPS'05) - Papers - Volume 01
Scheduling for broadcast operation in heterogeneous parallel computing environments
Systems and Computers in Japan
Scheduling heuristics for efficient broadcast operations on grid environments
IPDPS'06 Proceedings of the 20th international conference on Parallel and distributed processing
Pipelined broadcast on ethernet switched clusters
IPDPS'06 Proceedings of the 20th international conference on Parallel and distributed processing
Hi-index | 0.00 |
With the emergence of the network technologies, heterogeneous computing has become a wide accept paradigm for distributed and network computing. In this paper, we present different algorithms and evaluate their performance on performing atomic one-to-all broadcast in heterogeneous network with one port model. Based on general graph model, two scheduling algorithms, the Nearest Neighbor First and the Maximum Degree Neighbor First are firstly illustrated. The pre-scheduling strategy with constructing message forwarding table for avoiding redundant transmissions is applied as runtime support. By extending graph-based approaches, five tree-based heuris-tic algorithms, the Nearest Neighbor First, the Maximum Degree Neighbor First, the Maximum Height Sub-tree First, the Maximum Sub-Tree First and the Maximum Weighted Sub-tree First, are developed based on different network characteristics. The performance analysis shows that the Maximum Weighted Sub-tree First performs best in high degree heterogeneous environments. Overall speaking, contribution of this study relies on informing significant suggestions for adapting proper broadcasting mechanism in different hetero-geneous platforms.