FRANKFURT, Germany, June 21, 2017 — At the International Supercomputing Conference 2017, IBM Business Partner and OpenPOWER Foundation member, E4 Computer Engineering, the Italian technology provider of leading-edge solutions for HPC, data analytics and AI, announced that D.A.V.I.D.E. (Development for an Added Value Infrastructure Designed in Europe), a multi-node cluster powered by IBM POWER8 processor technology with NVIDIA Tesla P100 GPU accelerators and NVIDIA NVLink interconnect technology, entered the prestigious TOP500 list.
Twice a year, Top500.org publishes the TOP500 and Green500 lists. The TOP500 ranks supercomputing environments by performance capabilities, as determined by the Linpack benchmark, and recognizes the vendors and technologies that power the most powerful data intensive environments in the world. The Green500 list ranks the top 500 supercomputers in the world by energy efficiency.
D.A.V.I.D.E., developed within the Partnership for Advanced Computing in Europe (PRACE), provides a compelling solution for workloads with highly parallelized code and demanding memory bandwidth requirements such as weather forecasting, QCD, machine learning, computational fluid dynamics and genomic sequencing.
The supercomputer represents the third generation of the Pre-Commercial Procurement project for the development of a Whole-System Design for Energy Efficient HPC, and its innovative design uses the most advanced technologies to create a leading edge HPC cluster that provides powerful performance, low power consumption and ease of use.
D.A.V.I.D.E. was built with best-in-class components. The machine has a total of 45 nodes connected via Infiniband, with a total peak performance of 990 TFlops and an estimated power consumption of less than 2kW per node. Each node is a 2U form factor and hosts two IBM POWER8 Processors with NVIDIA NVLink and four Tesla P100 data center GPUs, with the intra-node communication layout optimized for best performance. Nodes are connected with an efficient EDR 100 Gb/s networking.
The multi-node cluster was fully configured in April 2017 at the E4’s facility in order to perform initial testing, running baseline performance, power and energy benchmarks using standard codes in an aircooled configuration. D.A.V.I.D.E. is currently available for a select number of users for porting applications and profiling energy consumption.
A key feature of the multi-node cluster is an innovative technology for measuring, monitoring and capping the power consumption of the node and of the whole system, through the collection of data from the relevant components (processors, memory, GPUs, fans) to further improve energy efficiency. The technology has been developed in collaboration with the University of Bologna.
“We are delighted to have reached this prestigious result to be included in the TOP500 list. The team worked very hard to design and develop this prototype and is very proud to see the system up and running; we look forward to seeing it fully available to the scientific community,” said Cosimo Gianfreda, CTO, E4 Computer Engineering. “With our work we have demonstrated that it is possible to integrate cost effective technologies to achieve high performance and significantly improve energy efficiency. We thank all our partners for the close collaboration that contributed to this great achievement.”
“HPC and AI are converging and the D.A.V.I.D.E. supercomputer will help the scientific community to run both kinds of workloads on an accelerated system,” said Stefan Kraemer, Director of HPC for EMEA at NVIDIA: “Engery-efficient accelerated computing is the only way to reach the ambitious goals Europe has set for its HPC future.”
About E4 Computer Engineering
Since 2002, E4 Computer Engineering has been innovating and actively encouraging the adoption of new computing and storage technologies. Because new ideas are so important, we invest heavily in research and hence in our future. Thanks to our comprehensive range of hardware, software and services, we are able to offer our customers complete solutions for their most demanding workloads on: HPC, Big-Data, AI, Deep Learning, Data Analytics, Cognitive Computing and for any challenging Storage and Computing requirements. E4. When Performance Matters.
Source: E4 Computer Engineering