Nvidia Tesla GPUs to Fuel Russia's Newest Supercomputer

RaS has chosen to update its system with Nvidia Tesla GPUs to boost its parallel computing abilities. The research institute will add, specifically, 128 Tesla M2090 GPUs within HP ProLiant SL390 G7 servers.

RaS institutes perform scientific research in fields including, including hydro-dynamics, geological modeling, genomic analysis, computational mathematics, molecular dynamics, and electromagnetics.

More than 30 RaS facilities, including Keldysh Institute of Applied Mathematics RAS and Institute for Image Processing Systems, currently leverage GPUs to fuel their efforts.

"GPU computing has been instrumental in the development of program and algorithmic simulations and borehole geophysical research," said Vyacheslav Glinskikh, Ph.D. (Geophysics), head of Laboratory at Trofimuk Institute of Petroleum Geology and Geophysics SB RAS, in a statement. "Based on our results, we were able to create new automated geophysical data interpretation systems for the oil and gas industry, promising to dramatically improve efficiencies in oil and gas exploration."

RaS’ adoption of GPUs for high-performance computing (HPC) represents a growing trend within the industry. As more and more research facilities leverage GPUs to fuel their supercomputers, the processors’ role in future HPC and exascale computing initiatives is becoming a hot conversation among manufacturers.

Nevertheless, the complexity of GPU programming, along with the rise of integrated processors, makes the future of these graphic chips somewhat uncertain.

Nvidia said its Tesla GPUs currently power three of the world's five fastest supercomputers, including the U.S. Oak Ridge National Lab’s Titan supercomputer, as well as the most powerful Russian supercomputer, Lomonosov, at Moscow State University.