“Titan” may be the most powerful supercomputer in the world at present time, but it is definitely not the one most apt at working full-throttle on the least energy. Italy's “Eurora” seems to have gained that title.
As it happens, NVIDIA's Tesla K2 GPU compute accelerators, which perform parallel processing tasks, are used in Eurora, just like they are present in Titan.
The Cineca Facility in Bologna, Italy, the country's most powerful supercomputing center, managed to get Eurora to work at 3,150 megaflops per watt.
This is a 26% improvement over the top system on the most recent Green500 list.
For those who need a reminder, Top500 is the list of supercomputers according to top performance, while Green500 is the list according to energy efficiency.
Eurora managed to outdo even the best of the best through a mix of the Eurotech Aurora Tigon hardware architecture, Aurora Hot Water Cooling technology (direct hot water cooling on all electronic and electrical components) and, of course, NVIDIA Tesla K20 cards.
Compared to air cooling, Hot-Water cooling reduces TCO (total cost of ownership) by 30-50% for data centers.
As such, scientists (with access to Italy's HPC) researching material science, astrophysics, Earth sciences and life sciences can rest easy knowing they are doing it more efficiently than anyone else.
Eurora was deployed Wednesday, January 30, 2013, at the aforementioned Cineca Facility in Bologna. It is a prototype system developed under the PRACE 2IP initiative.
On a related note, a commercial version of the Eurotech Aurora Tigon supercomputer is available from Eurotech
“Advanced computer simulations that enable scientists to discover new phenomena and test hypotheses require massive amounts of performance, which can consume a lot of power,” said Sanzio Bassini, director of HPC department at Cineca.
“Equipped with the ultra-efficient Aurora system and NVIDIA GPU accelerators, Eurora will give European researchers the computing muscle to study all types of physical and biological systems, while allowing us to keep data center power consumption and costs in check.”