The Community for Technology Leaders
Proceedings of the 22nd International Conference on Parallel Architectures and Compilation Techniques (2007)
Brasov, Romania
Sept. 15, 2007 to Sept. 19, 2007
ISSN: 1089-795X
ISBN: 0-7695-2944-5
pp: 199-212
Zhiyuan Wang , National University of Defense Technology, China
Jia Jia , National University of Defense Technology, China
Xuejun Yang , National University of Defense Technology, China
Panfeng Wang , National University of Defense Technology, China
Yunfei Du , National University of Defense Technology, China
Hongyi Fu , National University of Defense Technology, China
Guang Suo , National University of Defense Technology, China
ABSTRACT
This paper addresses the issue of fault tolerance in parallel computing, and proposes a new method named parallel recomputing. Such method achieves fault recovery automatically by using surviving processes to recompute the workload of failed processes in parallel. The paper firstly defines the fault tolerant parallel algorithm (FTPA) as the parallel algorithm which tolerates failures by parallel recomputing. Furthermore, the paper proposes the inter-process definition-use relationship analysis method based on the conventional definition-use analysis for revealing the relationship of variables in different processes. Under the guidance of this new method, principles of fault tolerant parallel algorithm design are given. At last, the authors present the design of FTPAs for matrix-matrix multiplication and NPB kernels, and evaluate them by experiments on a cluster system. The experimental results show that the overhead of FTPA is less than the overhead of checkpointing.
INDEX TERMS
null
CITATION
Zhiyuan Wang, Jia Jia, Xuejun Yang, Panfeng Wang, Yunfei Du, Hongyi Fu, Guang Suo, "The Fault Tolerant Parallel Algorithm: the Parallel Recomputing Based Failure Recovery", Proceedings of the 22nd International Conference on Parallel Architectures and Compilation Techniques, vol. 00, no. , pp. 199-212, 2007, doi:10.1109/PACT.2007.73
89 ms
(Ver 3.3 (11022016))