The last decade has observed an unprecedented growth in computational capabilities provided by the most modern supercomputers of today. For high-end computing (HEC), many science domains leverage simulation as an imperative tool along, with theory and experimentation, driving the need for these extreme-scale architectures. Advanced research in HEC architectures is being performed to break the exascale barrier by the end of this decade. Many forms of multi-core architectures (Intel, AMD) with accelerators (graphical processing units [GPUs]), field-programmable gate arrays (FPGAs), high-performance communication networks (InfiniBand, 10 Gigabit Ethernet, iWARP) and integrated systems (IBM, Cray) with various computation and communication tradeoffs have been introduced to meet the requirements. At the same time, designing high-performance software components is imperative to leveraging the computational capabilities of HEC architectures. Programming models, and systems software, which address the issues of performance while maintaining productivity, are imperative to meeting grand challenges of designing high-performance and scalable applications.