Resources

 

SONY DSCIn December 2013, UNN installed its Lobachevsky supercomputer, a computer system based on NVIDIA Kepler architecture and Intel Xeon Phi coprocessors with a peak performance of 573 Tflops. According to the 20th edition of the Russian Top50 list, Lobachevcky is ranked # 4 as of April 1, 2014. The supercomputer consists of 2 segments:

  1. Segment running on Linux (CentOS 6.4) and consisting of
    • 120 computer nodes;
      • Two Intel Sandy Bridge E5-2660 2.2 GHz processors (8 cores);
      • 64 GB RAM;
      • Three NVIDIA Kepler K20 graphic processors (2688 streaming processors, 6 GB GDDR5)
    • 10 computer nodes
      • Two Intel Sandy Bridge E5-2660 2.2 GHz processors (8 cores);
      • 64 GB RAM;
      • Two Intel Xeon Phi 5110P coprocessors (8GB, 60 core, 240 streaming processors).
    • 10 computer nodes
      • Two Intel Sandy Bridge E5-2660 2.2 GHz processors (8 cores);
      • 64 GB RAM.
  1. Segment running on Windows(HPC Server 2008) and consisting of
    • 10 computer nodes
      • Two Intel Sandy Bridge E5-2660 2.2 GHz processors (8 cores);
      • 64 GB RAM;
      • Three NVIDIA Tesla M2090 graphic processors (512 streaming processors, 6 GB GDDR5)
    • 50 computer nodes
      • Two Intel Sandy Bridge E5-2660 2.2 GHz processors (8 cores);
      • 64 GB RAM;
      • Two NVIDIA Tesla M2090 graphic processors (512 streaming processors, 6 GB GDDR5)

The supercomputer also includes a data storage system with a total capacity of 120 TB. The cluster has an InfiniBand-based computer network.

In addition to Lobachevsky, the Center operates two HPC clusters acquired by UNN in 2007 and 2010 as part of an innovative educational program within the Education National Project.

The 2007 cluster has a SKIF-based architecture and includes 64 two-processors and two-core Intel Xeon servers 3.2 GHz, 4 Gb RAM and 12 workstations equipped with Intel Core 2 Quad 2.4 GHz, 4 Gb RAM and a Gigabit Ethernet network. The cluster’s peak performance is 3 Tflops.

The 2010 cluster consists of 16 computer nodes with a hybrid architecture. Each node has two Intel Xeon 2.13 GHz, 24 Gb RAM processors and two NVIDIA Tesla X2070 graphic processors. The network equipment is InfiniBand-based. The cluster’s peak performance is 17 Tflops.