Estimating Energy Efficiency of Applications on HPC Resources
PLEASE APPLY ONLINE TO THE SCHOOL OF ENGINEERING, PROVIDING THE PROJECT TITLE, NAME OF THE PRIMARY SUPERVISOR AND SELECT THE PROGRAMME CODE "EGPR" (PHD - SCHOOL OF ENGINEERING)
This is a project within the multi-disciplinary EPSRC and ESRC Centre for Doctoral Training (CDT) on Quantification and Management of Risk & Uncertainty in Complex Systems & Environments, within the Institute for Risk and Uncertainty. The studentship is granted for 4 years and includes, in the first year, a Master in Decision Making under Risk & Uncertainty. The project includes extensive collaboration with prime industry to build an optimal basis for employability.
As high performance computing is becoming as one of the core and enabling technology for all domains of sciences and technology, the energy footprint of the large-scale scientific applications is also becoming an increasing concern for the community. Globally, the energy consumption of computing has already overtaken that of airlines. One of the critical challenges of upcoming large-scale systems is the limit on energy consumption.
With respect to large-scale scientific applications, the key difficulty is in selecting a system that is energy efficient. In the absence of any means for estimating the energy consumption or efficiency of a given application on different architectures, only option is to exhaustively execute the application on each of the systems and finding out which one is energy efficient.
The aim of this PhD project is to provide a data driven energy efficiency estimation framework for large-scale applications. The vision is that given a large-scale scientific application (intended to run on supercomputers), the framework should estimate the energy consumption (or efficiency) of the application on different systems without actual execution.
The key idea is to use a set of benchmarks to derive baseline efficiency figures on different systems and then to use these figures in conjunction with the detailed profile data for estimating the energy consumption (or efficiency). The deliverables of this research will be of great practical significance to the HPC community. It is aimed to facilitate better scheduling, cost effective execution, minimising the energy and carbon footprints.
The project will be supported by the Hartree Centre.
The PhD Studentship (Tuition fees + stipend of £ 13,726 annually over 4 years) is available for Home/EU students. In addition, a budget for use in own responsibility will be provided.