December 7th, 2008
We present a novel rails approach so that future E-Science applications can effectively exploit future system architectures, including multi-core and many-core architectures, multiple network cards, multiple graphical processing units and hybrid hierarchical memory architectures. We define “rail” as the co-scheduling of two or more of these resources. This approach enables creation of parallel multi-rails through every aspect of an end system: from processing on the multi- and many cores, to generation of multiple data flows, and streaming over multi-lane NICs connected via a parallel interconnect. We describe a novel open-source multi-rail toolkit and the evaluation of end-system parameters that impact the efficiency of such multi-rail systems, including Interrupt, Memory, Thread, and Core Affinities - key properties for achieving scalable performance.
Vishwanath, V., Leigh, J., Shimizu, T., Nam, S., Renambot, L., Takahashi, H., Takizawa, M., Kamatani, O., The Rails Toolkit (RTK) - Enabling End-System Topology-Aware High End Computing, IEEE Fourth International Conference on e-Science, eScience, 2008, pp. 309-316, December 7th, 2008. http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=4736772&isnumber=4736722