A Capabilities-Aware Programming Model for Asymmetric High-End Systems

In this research, we investigate and address the challenges of asymmetry in High-End Computing (HEC) systems comprising heterogeneous architectures with varying I/O and computation capacities. We focus on developing a flexible, scalable and easy-to-use programming model that automatically adapts to the capabilities of the system resources on largescale asymmetric clusters. Furthermore, we aim to develop innovative and efficient workload distribution techniques that bridge the asymmetry between system components. In particular, we intent to design tools and technologies that enable quick and efficient utilization of high-end asymmetric clusters in large-scale settings for modern scientific and enterprise computing. Keywords-Accelerator-based systems; heterogeneous clusters; programming asymmetric clusters; capability-aware task distribution

• Type: Conference
• Year: 2010
• Where: CCGRID
• Authors: M. Mustafa Rafique
• Keywords: Asymmetric Clusters, CCGRID 2010, Distributed And Parallel Computing, Heterogeneous Clusters, Largescale Asymmetric Clusters