This workshop will address the issues and potential solutions for power capacity planning and management in an HPC facility. Two main issues are managing stranded and trapped capacity. Stranded and trapped capacity issues often manifest when facility managers are given highly inaccurate power budget estimations. Power and other physical resources are allocated based on these estimations, which results in overprovisioning and underutilization of these resources. Currently many HPC users are not concerned about power usage effectiveness, but that is certain to change in the near future as the HPC community gears up to meet the exa-scale challenge. Power consumption and management and related infrastructure requirements are considered to be some of the main hurdles to overcome in order to achieve sustained exa-scale operations on future computing platforms and key applications.
This workshop will be a multi-year series to address different aspects of power capacity management and will include topics such as energy efficient hardware design, green programming models, energy-aware job scheduling, monitoring, modeling, evaluation etc. For 2016, the theme of this workshop will be "knowledge discovery."
If you plan on attending, please send us an email and include your name, affiliation and contact information.