Tesla shifts Russian supercomputer into overdrive

Moscow State University is upgrading its already formidable Lomonosov supercomputer with additional Nvidia Tesla GPUs.

The tricked-out system is expected to couple 1,554 Tesla X2070 cards with an identical number of quad-core CPUs, delivering 1.3 petaflops of peak performance and effectively making Lomonosov the number one supercomputer in Russia.

According to Victor Sadovnichy of Moscow State University, the system is currently used for computationally intensive research, such as global climate change, ocean modeling, post-genomic medicine and galaxy formation simulations.

“Our research requires enormous computational resources, and we need to deliver this performance as efficiently as possible,” explained Sadovnichy.

“The only way for us to achieve these twin goals is with a hybrid GPU/CPU based system.”

An Nvidia rep told TG Daily that a number of supercomputing centers in Russia are looking at ways of increasing performance without exceeding power budgets.

For example, Lobachevsky State University of Nizhni Novgorod (NNSU) – Russia’s first CUDA Research Center – actively uses GPUs across projects  as diverse as studying living systems via solid mathematical modeling and large-scale computation.

NNSU is also slated to install a GPU-enabled cluster with a peak performance of 100 teraflops later this year, and plans to increase the cluser to 500 teraflops by the end of 2012.