Performance analysis and optimization of MPI collective operations on multi-core clusters Export

@article{citeulike:4405442, abstract = {Abstract\ \ Memory hierarchy on multi-core clusters has twofold characteristics: vertical memory hierarchy and horizontal memory hierarchy. This paper proposes new parallel computation model to unitedly abstract memory hierarchy on multi-core clusters in vertical and horizontal levels. Experimental results show that new model can predict communication costs for message passing on multi-core clusters more accurately than previous models, only incorporated vertical memory hierarchy. The new model provides the theoretical underpinning for the optimal design of MPI collective operations. Aimed at horizontal memory hierarchy, our methodology for optimizing collective operations on multi-core clusters focuses on hierarchical virtual topology and cache-aware intra-node communication, incorporated into existing collective algorithms in MPICH2. As a case study, multi-core aware broadcast algorithm has been implemented and evaluated. The results of performance evaluation show that the above methodology for optimizing collective operations on multi-core clusters is efficient.}, author = {Tu, Bibo and Fan, Jianping and Zhan, Jianfeng and Zhao, Xiaofang}, citeulike-article-id = {4405442}, citeulike-linkout-0 = {http://dx.doi.org/10.1007/s11227-009-0296-3}, citeulike-linkout-1 = {http://www.springerlink.com/content/v3j8j3075845x484}, doi = {10.1007/s11227-009-0296-3}, journal = {The Journal of Supercomputing}, keywords = {communication\_algorithms, mpi, multicore, numa, performance\_modelling}, posted-at = {2009-09-23 15:32:38}, priority = {2}, title = {Performance analysis and optimization of MPI collective operations on multi-core clusters}, url = {http://dx.doi.org/10.1007/s11227-009-0296-3} }

TY - JOUR ID - citeulike:4405442 L3 - citeulike-article-id:4405442 N2 - Abstract  Memory hierarchy on multi-core clusters has twofold characteristics: vertical memory hierarchy and horizontal memory hierarchy. This paper proposes new parallel computation model to unitedly abstract memory hierarchy on multi-core clusters in vertical and horizontal levels. Experimental results show that new model can predict communication costs for message passing on multi-core clusters more accurately than previous models, only incorporated vertical memory hierarchy. The new model provides the theoretical underpinning for the optimal design of MPI collective operations. Aimed at horizontal memory hierarchy, our methodology for optimizing collective operations on multi-core clusters focuses on hierarchical virtual topology and cache-aware intra-node communication, incorporated into existing collective algorithms in MPICH2. As a case study, multi-core aware broadcast algorithm has been implemented and evaluated. The results of performance evaluation show that the above methodology for optimizing collective operations on multi-core clusters is efficient. JF - The Journal of Supercomputing TI - Performance analysis and optimization of MPI collective operations on multi-core clusters KW - communication_algorithms KW - mpi KW - multicore KW - numa KW - performance_modelling AU - Tu, Bibo AU - Fan, Jianping AU - Zhan, Jianfeng AU - Zhao, Xiaofang UR - http://dx.doi.org/10.1007/s11227-009-0296-3 ER -