E4 Computer Engineering wins third and final phase of PCP commissioned by PRACE, with innovative OpenPOWER-based HPC cluster built on top of IBM POWER8 architecture to be exhibited at SC16


16 Nov 2016 Salt Lake City - PRACE has announced the winners of the third and final phase of its Pre-commercial Procurement (PCP). Out of the three companies involved in the tender, E4 scored the highest points with an innovative and energy efficient OpenPOWER-based HPC cluster that will be exhibited during SC16: D.AV.I.D.E. SUPERCOMPUTER.

PRACE initiated the selection process in 2012 to obtain R&D services which should result in future PRACE HPC systems to become more energy efficient. The strict selection process saw several phases, in which the pre-selected European companies were asked to design HPC solutions that would not only be able to outstandingly perform on the most complex alghoritms used by scientists and researchers, but would also drastically reduce power consumption.

PRACE's selection process is highly competitive and, during Phase II, the three companies produced early prototypes. All of them were invited to bid for Phase III, during which the pilot for the three promising solutions will be built and deployed at the hosting centres. E4 stood out since the early steps of the selection and, thanks to the expertise acquired through years of collaboration with leading European research centres, has proposed a solution that was ranked first place for performance capabilities: D.A.V.I.D.E. SUPERCOMPUTER is a Petaflop-class HPC solution, designed, integrated and tested by E4, that is based on OpenPOWER architecture, using IBM POWER8 processors with NVLink bus and the ultra performing GPGPU NVIDIA TESLA P100 SXM2.

The Open Rack form factor compute nodes were provided by Wistron Corporation to ensure full customization of the solution. Another significant feature is represented by the liquid cooling modules by CoolIT, which help increase energy efficiency compared to the more standard fans. The University of Bologna was also involved in the project with a software that enables power monitoring, profiling and management, while code porting and optimization were performed by the British based firm Doublewoods Consultants Ltd.

Attendees of SC16 in Salt Lake City will be able to see an example of the system at E4's booth.

"D.A.V.I.D.E. SUPERCOMPUTER represents a breakthrough innovation for scientists and researchers who require high performances to enable scientific discovery, while keeping an eye on power consumption to reduce costs and environmental impact", stated Fabrizio Magugliani, Strategic Planning & Business Development at E4. "We are honoured to have been selected for the third phase of the PCP in an open competition with other European companies. OpenPOWER based platforms are the harbinger of a new generation of HPC systems which deliver high performances while being environmentally conscious and allowing scientists to make their dreams become a reality."

"Today's emerging workloads like machine and deep learning, artificial intelligence, accelerated databases, and high performance data analytics require incredible speed through accelerated computing", stated Sumit Gupta, IBM VP, High Performance Computing & Data Analytics. "Delivering the capabilities of the new IBM POWER8 with NVIDIA NVLink-based system through the new E4's cluster brings a highly accelerated computing platform with remarkable energy efficiency to scientists and HPC users requirements."

"Co-finance and actively participate to the technological innovation processes in Europe is part of our institutional mission", stated Sanzio Bassini, Director of Department of Supercomputing Applications and Innovation at CINECA. "We therefore partake of the success achieved by E4 in this PCP initiative and we are ready to promote the new computing system by E4, that will complement the supercomputers in production at our data centre, with the migration activity of computing applications and optimization of the performance of the algorithms in order to make the most of the technological potential of this new prototype that will be launched into production in early 2017."

Source: E4 Computer Engineering