Eurotech develops industry’s highest-density GPU-accelerated supercomputer cluster with NVIDIA

It will include joint engineering efforts to develop the industry’s highest density Eurotech supercomputer cluster, the Aurora Tigon, featuring 256 NVIDIA Tesla K20 GPUs in an X-rack configuration. The new systems are expected to deliver more than 500 teraflops of performance per rack and above 3.6 GFlops per watt, establishing an unmatched new frontier for energy-efficient supercomputer performance.

“This tremendous power density is made possible by the combination of Eurotech’s proven liquid cooling technology and the unprecedented computing power of the new NVIDIA Kepler GPUs. We are happy to extend our offering with the greenest range of HPC solutions in the world.” says Giampietro Tecchiolli, Eurotech CTO.

Using current-generation NVIDIA Fermi GPUs, the current Aurora supercomputer delivers energy-efficient performance in the range of 2.5 GFlops per watt. Once next-generation NVIDIA Tesla K20 GPUs based on the new NVIDIA Kepler computing architecture are available later this year, the Aurora system is expected to deliver approximately 3.6 GFlops per watt, while keeping power usage effectiveness within the 1.05-1.1 range.

By leveraging NVIDIA Kepler GPUs, the Aurora systems have the horsepower necessary to break through the desirable .5 petaflops per rack barrier.

The new Eurotech product is expected to be the base technology for the development of the PRACE 2IP co-funded prototype action with the project called “Eurora,” which is led by the CINECA supercomputing center in Italy. CINECA will deploy this new prototype system in the fields of computational sciences, including fundamental constituents of matter, condensed matter, astrophysics, life sciences and Earth sciences.

“The new Eurora prototype system will be available to scientists in Europe, who will have the possibility to get access to one of the most advanced and innovative HPC solutions, expanding CINECA offering to the community,” says Sanzio Bassini, CINECA HPC Associate director.

The NVIDIA Tesla K20 GPU is the new flagship of the Tesla GPU product family, designed for the most computationally intensive HPC environments. Expected to be the world’s highest-performance, most energy-efficient accelerator, the Tesla K20 GPU is planned to be available in the fourth quarter of 2012.

“Eurotech’s high-density, liquid-cooled systems provide the highest performance per rack for deploying large-scale systems,” said Sumit Gupta, senior director of the Tesla business unit at NVIDIA. “By setting a new standard for energy-efficient, high performance computing, the new Aurora system promises to broaden the adoption of GPU-accelerated computing for a range of engineering and scientific disciplines.”

For further information visit the Eurotech booth (# 750) at the ISC’12 trade show.