October 11, 2013
EDINBURGH, Scotland, Oct. 11 -- Today, a consortium led by EPCC, the supercomputing centre at the University of Edinburgh, announced the successful launch of the Adept project. Adept brings together software and hardware experts from HPC and Embedded to address the challenge of energy efficient use of parallel technologies. In addition to EPCC, the consortium comprises two world-leading Universities, Uppsala and Ghent, as well as two representatives from industry, Alpha Data Parallel Systems Ltd and Ericsson AB. Adept is supported with €2.8m through the European Commission’s Framework 7 programme.
It's been quite some time since parallel computing was limited to only high-end scientific applications; it is increasingly becoming ubiquitous in all forms of software development and especially so in the Embedded domain. In HPC, the race towards reaching the exascale by 2022 is steering research into new directions, one of which is that of energy efficiency.
Dr. Michèle Weiland, the coordinator of the Adept project which is led by EPCC, explains, “We will address the challenge of energy efficiency in parallel technologies by building on the expertise of HPC software developers in exploiting parallelism for performance, and Embedded Systems engineers in managing power consumption.”
The aim of Adept is to further the understanding of the power usage of parallel software and hardware, and to develop a tool that enables the rapid exploration of power consumption and performance design space across a range of heterogeneous architectures.
“Over the last 15 years, one of our overriding research goals has been to produce new technologies and tools to help programmers understand and optimise their systems for better performance and power efficiency,” says Professor Erik Hagersten from Uppsala University. “This project brings us together with application experience from HPC and large-scale commercial Embedded computing, as well as some of the leading researchers in efficient performance modelling. This is an ideal environment for taking our research results to a new level.”
Professor Lieven Eeckhout from Ghent University, who is working on the modelling aspect of the project, agrees, “The Adept project is a great opportunity for me and my research group to combine the interval model, which we developed here in Ghent, with the cache models developed in Uppsala, and build a powerful tool for software developers to explore hardware options in the Embedded and HPC space.”
The View From Industry
Alpha Data Parallel Systems Ltd, a leading global supplier of high-performance reconfigurable platforms, is one of the commercial partners in Adept. Dr. Andrew McCormick, Design Engineer at Alpha Data, explains why the project is important to the company, “Adept will help bring power management for HPC down from the cluster level to the individual application level. Identifying the total lifetime cost of power hotspots in applications may justify the extra development costs of using lower power customisable hardware accelerators. Alpha Data is pleased to be involved in this project, which could lead to new opportunities for the reconfigurable computing market."
10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.
10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?