Results 1 -
3 of
3
Improving Speedup and Response Times by Replicating Parallel Programs on a SNOW
- In Proceedings of the 10th Workshop on Job Scheduling Strategies for Parallel Processing
, 2004
"... this paper we consider a SNOW where parallel jobs are run in an opportunistic fashion as in Condor. In such an environment, workstation owner processes have preemptive priority over batched parallel programs. Owner process workstation reclamations stop execution of the batch job task and hence may s ..."
Abstract
-
Cited by 5 (0 self)
- Add to MetaCart
this paper we consider a SNOW where parallel jobs are run in an opportunistic fashion as in Condor. In such an environment, workstation owner processes have preemptive priority over batched parallel programs. Owner process workstation reclamations stop execution of the batch job task and hence may significantly impact the parallel job response time
Competitive Execution of Sequential Programs on a Network of Shared Processors
- In IEEE 2nd International Conference on Algorithms and Architectures for Parallel Processing
, 1996
"... We propose an alternative to process migration, called competition, to speed up distributed programs in the background on a network of shared processors. Competition protocols are transparent operating system facilities that involve creating multiple instances (called clones) p 1 , p 2 , etc. of a p ..."
Abstract
-
Cited by 1 (0 self)
- Add to MetaCart
We propose an alternative to process migration, called competition, to speed up distributed programs in the background on a network of shared processors. Competition protocols are transparent operating system facilities that involve creating multiple instances (called clones) p 1 , p 2 , etc. of a process P on different processors, and making clones "compete", i.e., attempting to guarantee that the output of the clone that is farthest "ahead" is fed to the rest of the computation, and that the entire application's performance tracks that of the clone which is farthest ahead. One clone may be ahead of or behind others depending on the current foreground loads. If for any reason there is variation in the progress of the clones, so that one clone is ahead at some times, but another is ahead at other times, then a set of competing clones may outperform any single copy. In this paper, we focus on sequential programs, but our results for sequential programs can be extended to distributed pro...
Speedup of Distributed Programs on a Network of Shared Processors
, 1996
"... This paper presents an alternative to process migration, called competition, to speed up distributed programs in the background on a network of processors. Competition protocols are transparent operating system facilities that involve creating multiple instances (called clones) p 1 , p 2 , etc. of a ..."
Abstract
- Add to MetaCart
This paper presents an alternative to process migration, called competition, to speed up distributed programs in the background on a network of processors. Competition protocols are transparent operating system facilities that involve creating multiple instances (called clones) p 1 , p 2 , etc. of a process P on different processors, and making clones "compete", i.e., attempting to guarantee that the output of the clone that is farthest "ahead" is fed to the rest of the computation, and that the entire application's performance tracks that of the clone which is farthest ahead. We show that competition protocols offer performance benefits that are as good as or better than migration protocols for distributed programs under comparable assumptions. We also demonstrate that competition protocols offer even more speedup for distributed programs than for sequential programs. 1 Introduction The cost/performance ratio of workstations has shown a dramatic improvement in the past, and this tren...

