Application assessment of GroupLB load balancer in iterative applications
Abstract
This article presents the results of applying the GroupLB load balancer in real applications. Its purpose is to analyze the performance gains achieved in relation to the other load balancers state of the art. The central bank has been implemented using the Charm ++ programming model. seeking to reduce the iterative application runtime running on parallel shared memory environments. The results of the initial tests showed reductions in total execution time compared to other load balancers state of the art. The initial results of the benchmark lbtest present up to 21.2% compared with the tests without the use of load balancer gains.
References
Bhatelé, A., Kumar, S., Mei, C., Phillips, J. C., Zheng, G., and Kalé, L. V. (2008). Overcoming scaling challenges in biomolecular simulations across multiple platforms. In Proceedings..., pages 1–12. International Symposium on Parallel and Distributed Processing (IPDPS), IEEE.
da Rosa Lizot, G., Mastella, V. M., Pavan, P. J., and Padoin, E. L. (2018). Grouplb: Load balancer proposal to reduce the execution time of parallel applications. In XVI Parallel and Distributed Processing Workshop (WSPPD), pages 1–4, Porto Alegre, RS.
dos Santos, V. R. S., Padoin, E. L., Navaux, P. O. A., and Méhaut, J.-F. (2018). Smartlb: Proposta de um balanceador de carga para redução de tempo de execução de aplicações em ambientes paralelos. In WPERFORMANCE, pages 1–14.
Freytag, G., Arruda, G., Martins, R. S. M., and Padoin, E. L. (2015). Análise de desempenho da paralelização do problema de caixeiro viajante. In XV Escola Regional de Alto Desempenho (ERAD), pages 1–4, Gramado, RS. SBC.
Padoin, E. L., Castro, M., Pilla, L. L., Navaux, P. O., and Méhaut, J.-F. (2014). Saving energy by exploiting residual imbalances on iterative applications. In 2014 21st International Conference on High Performance Computing (HiPC), pages 1–10. IEEE.
Zheng, G., Bhatelé, A., Meneses, E., and Kalé, L. V. (2011). Periodic hierarchical load balancing for large supercomputers. International Journal of High Performance Computing Applications, 25(4):371–385.