Computational resources

Storage system

Visualization

Тelecommunications

MVS-15000BM (924 CPU) The cluster MVS-15000BM is used to solve complex scientific and technical problems. Peak perfomance of cluster is 8.13 Teraflops. Total amount of the RAM is 1848 Gbytes.

The peak performance of cluster will rise to 15 Teraflops in the feature.

MVS-1000M (768 CPU) Peak performance is 1 TFlops. Supercomputer consist of 6 base blocks, each block includes 64 dual-CPU modules. Modules are build on Alpha21264A, 667 MHz CPU and connected via high performance Myrinet (2 Gbit/s) network, used only for inter-module communication during parallel programs execution.

Hewlett Packard SuperDome (64 CPU) is symmetric multiprocessor system with peak performance 141,3 GFlops. It is targeted for solution of problems requiring big amount of shared memory and high CPU power. System is build on HP PA-8600 550 Mhz CPUs connected to 64 GBytes of shared memory.

System is operating under BXA restrictions.

Hewlett Packard V2250 (16 CPU) is symmetric multiprocessor complex with peak performance 16 GFlops.

System is operating under BXA restrictions.

Cluster "KRYPTON" (16 CPU, Intel Zeon) used for parallel file-systems. Cluster is connected to SAN.

Peak performance of the cluster is 76,8 GFlops.

Cluster "KAPPA" (16 CPU, AMD Athlon) used for teaching high-performance computational technologies to students.

Peak performance of the cluster is 49 GFlops.

Visualization system is based on HP workstation J2240 and J5000 with powerful 3D graphics accelerators.

Storage system.

Data storage system for computational clusters is built on Storage Area Network technology. This system consists of four disk servers IBM FAStT700, two Fiber Channel switches and eight dual-CPU Input/Output nodes. Total disk memory is 10 TBytes, archiving memory is 60 TBytes (on magnetic tapes).

Archiving system.

Data storage repository system is build on dual-CPU servers. It uses Tivoli software and supports three levels of hierarchy of external memory up to 100 Tbytes of capacity with software system for data distribution optimisation.

All computational resources are connected to local area network using Gigabit/Fast Ethernet (1000/100 MBit/s) and ATM (622/155 MBit/s) technologies.

With use of main trunks of Russian Academy of Sciences network (RASnet), built on Gigabit Ethernet and ATM technologies, JSCC is connected with Russian Backbone Network (RBNet), Russian University Network (RUNNet), Moscow University Network (MSUnet) and with dozens of universities and scientific institutions in Moscow.

Currently JSCC is a member оf the GÉANT-2 project and represents scientific and educational network community of the Russian Federation in it. The GÉANT-2 project is the collaboration between the European Commission and national and educational networks of 30 countries. Its principal purpose is to develop Multi-Gigabit European Academic Network for research and education. The project also covers a number of network relating activities: network testing, development of new technologies and support for research projects with high network capacity requirements (like GRID). GÉANT is the name given to both the project and the network that emerged from it.