Workshop on Power Optimization in High-Performance Computing

Designing high-performance computing systems is a multi-dimensional challenge. Power optimization is an important concern right from the HPC system design to HPC system operations. A hypothetical Exascale computer with over 64 K nodes of 16 teraops, each would consume nearly 4 MW per hour for just computing – even at 64 watts per node. Fine-grained DVFS (Dynamic Voltage and Frequency Scaling) and Running Average Power Limit (RAPL) can help achieve substantial energy savings for different workloads. Distributed, adaptive and autonomic power Management in HPC systems has become one of the major research challenges in order to reduce overall energy consumption. The scope of this workshop includes

Power optimized systems – Power perspective in HPC, Adaptive and Autonomic Power Management Methodology, Power Profiling of the HPC applications, fine-grained measurement techniques, the evolution of power-aware subsystems [i.e. processor, memory, and disk] and fine-grained control mechanism for power manageable components.

Power optimized nodes – Synthetic Processors and application optimized computing along with the research opportunities in emerging HPC modelling languages and standards to drive programmer productivity as well as high performance, low cost, low power computing of the future.

Programme Schedule