MiniSymposium on Energy Aware Scientific Computing on Low Power and Heterogeneous Architectures

Energy consumption is increasingly becoming one of the most relevant issue concerning the computing platforms for scientific applications and workloads. This has then required the use of energy-efficient processors, ranging from many-core architectures, like GP-GPU and Xeon-Phi, to Systems-on-Chip (SoCs). These processors feature a high performance-per-watt ratio, aimed on one side at energy efficiency, but requiring on the other side careful programming and optimization to provide high computing performance. Also off-the-shelf low-power Systems-onChip, originally designed for the embedded and mobile market, are becoming attractive for scientific and industrial applications given their increasing computing performance coupled with relatively low cost and low electrical power demand. Scientific computing and mobile/embedded sectors, historically very isolated and confined to specific markets, are now experiencing a convergence under the driving forces of high computational power demands, power consumption limitations and cost effectiveness. The mini-symposium on “Energy Aware Scientific Computing on low power and heterogeneous architectures” provided a forum to discuss and share knowledge on developing, running and using computing systems based on low power, energy-efficient, and heterogeneous architectures. It originated from a collaboration between the COSA project (http://www.cosa-project.it/) funded by the Italian Institute for Nuclear Physics (INFN) and the ParCo2017 organizers. Seven contributions have been presented during a whole day (September, 13th) co-located with the ParCo2017 conference. Each presentation triggered active discussion among the attendees and presenters, allowing an interesting exchange of ideas and experiences on the mini-symposium topics.