7 Hits in 4.3 sec

Work stealing and persistence-based load balancers for iterative overdecomposed applications

Jonathan Lifflander, Sriram Krishnamoorthy, Laxmikant V. Kale
2012 Proceedings of the 21st international symposium on High-Performance Parallel and Distributed Computing - HPDC '12  
In this paper, we consider the design and evaluation of two distinct approaches to addressing this challenge: persistence-based load balancing and work stealing.  ...  We also present an active-message-based retentive work stealing algorithm optimized for iterative applications on distributed memory machines.  ...  With retention, work stealing performs almost as well as persistence-based load balancing. CONCLUSION We presented scalable algorithms for persistence-based load balancing and work stealing.  ... 
doi:10.1145/2287076.2287103 dblp:conf/hpdc/LifflanderKK12 fatcat:wchqotv6vzhtpnhh2hhuychp2u

A distributed dynamic load balancer for iterative applications

Harshitha Menon, Laxmikant Kalé
2013 Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis on - SC '13  
For many applications, computation load varies over time. Such applications require dynamic load balancing to improve performance.  ...  This paper describes a fully distributed algorithm for load balancing that uses partial information about the global state of the system to perform load balancing.  ...  Experiments for this work were performed on Mira and esta, IBM Blue Gene/Q installations at Argonne National Laboratory.  ... 
doi:10.1145/2503210.2503284 dblp:conf/sc/MenonK13 fatcat:uurwcwv5frd3zordofyz7ej7dy

Improving HPC Application Performance in Cloud through Dynamic Load Balancing

A. Gupta, O. Sarood, L. V. Kale, D. Milojicic
2013 2013 13th IEEE/ACM International Symposium on Cluster, Cloud, and Grid Computing  
In this paper, we present a dynamic load balancer for tightlycoupled iterative HPC applications in cloud.  ...  For tightly-coupled iterative applications, one slow processor slows down the entire application, resulting in poor CPU utilization.  ...  ACKNOWLEDGMENTS This work was supported by HP Labs' 2012 IRP award.  ... 
doi:10.1109/ccgrid.2013.65 dblp:conf/ccgrid/GuptaSKM13 fatcat:poatj7saofdvdpijockg4ehmky

Implementation and performance aspects of Kahn process networks

Zeljko Vrba
2010 ACM SIGMultimedia Records  
Lastly, we use Nornir to evaluate several load-balancing methods: static assignment, work-stealing, our improvement of work-stealing, and a method based on graph partitioning.  ...  The understanding brought by this evaluation is significant not only in the context of the Kahn model, but also in the more general context of load-balancing (potentially distributed) applications written  ...  We have implemented three load-balancing methods: static assignment, work-stealing and a method based on graph partitioning.  ... 
doi:10.1145/1874413.1874418 fatcat:sm5wsqpyyfevtarf6336elhyny

Parallel Programming with Migratable Objects: Charm++ in Practice

Bilge Acun, Abhishek Gupta, Nikhil Jain, Akhil Langer, Harshitha Menon, Eric Mikida, Xiang Ni, Michael Robson, Yanhua Sun, Ehsan Totoni, Lukasz Wesolowski, Laxmikant Kale
2014 SC14: International Conference for High Performance Computing, Networking, Storage and Analysis  
In this paper, we leverage our experience with these concepts to demonstrate their applicability and efficacy for real world applications.  ...  The advent of petascale computing has introduced new challenges (e.g. heterogeneity, system failure) for programming scalable parallel applications.  ...  This work also used the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by National Science Foundation grant number ACI-1053575.  ... 
doi:10.1109/sc.2014.58 dblp:conf/sc/AcunGJLMMNRSTWK14 fatcat:tmksxaixivdmbhk2h7lo6tivdu

Using imbalance metrics to optimize task clustering in scientific workflow executions

Weiwei Chen, Rafael Ferreira da Silva, Ewa Deelman, Rizos Sakellariou
2015 Future generations computer systems  
persistence-based load balancers for iterative overdecomposed applica- 39–59. tions, in: Proceedings of the 21st international symposium on High- [36] G. B. Berriman, E.  ...  Kale, Work stealing and M. Shields (Eds.), Workflows for e-Science, Springer London, 2007, pp.  ... 
doi:10.1016/j.future.2014.09.014 fatcat:3wmkdt7h45fyvfe6hc4nrbkila

Progressive load balancing of asynchronous algorithms [article]

Justs Zarins, University Of Edinburgh, Michele Weiland, Bjoern Franke, Lorna Smith
Work stealing, an alternative load balancing approach, is shown to be ineffective. In distributed memory, the method reduces the impact of up to 8 slow nodes out of 15, each slowed [...]  ...  To address this problem, we use the unique properties of asynchronous algorithms to develop a load balancing strategy for iterative convergent asynchronous algorithms in both shared and distributed memory  ...  Finally, I would like to thank the programme's leaders and EPCC for providing support and many opportunities for development.  ... 
doi:10.7488/era/1607 fatcat:ujmsddad3jdj7mmwv7r2gy7h2a