Creation of computer cluster
Open Technologies, Russian system integrator and the Institute of Mechanics and Mathematics at the Urals Department of the Russian Academy of Science informed about completion of new computing cluster for scientific and technical calculations.
The Institute had to build a highly productive cluster for solving resource consuming tasks and introduction of modern IT intm science researches.
The new computing system was 4.3 times more productive than the old one used by Urals Department of the RAS.
The project made part of Проект выполнялся в рамках RTT № 0120.0852225 and was carried out in close cooperation of Open Technologies Perm Office and RAS UD IMM.
The RTT included the analysis of modern high density computing systems, the equipment and software available at IT market.
Ultimately, the cluster was built with a blade server HP BladeSystem c7000 platform. The final computing system for В RAS UD IMM consisted of 32 dual-processor HP Proliant BL 460c nodes installed on quad-core Intel Xeon 5450 chips with stroking frequency of 3,0 GHz. Red Hat Enterprise Linux 5 and Microsoft Windows Compute Cluster Server 2008 operational systems were chosen. Infiniband technologies were used to build the communication environment.
The peak productivity of the cluster is 3 teraflops and the real one (defined byLinpack test) is 2,36 teraflops.
It is meant for solving fundamental and applied problems in such fields as mathematics and mechanics, physics of metals, electrical and thermal physics, hard bodies chemistry and high temperature electrical chemistry, metallurgy; geology and mining, geochemistry and geophysics, ecology, biology, economy and medicine.
According to Mikhai Ludvogovich Goldsteein, the head computer equipment department at RAS UD IMM, "the created system allows solving a wide circle of issues, their complexity depending on the development of science and technology, enhance requirements to the relevance of physical contents of mathematic models, tough criteria to the results reliability, exclusion of mistakes in discrete representation of initial differentiated models and obtaining decisions with guaranteed result.
The Institute had to build a highly productive cluster for solving resource consuming tasks and introduction of modern IT intm science researches.
The new computing system was 4.3 times more productive than the old one used by Urals Department of the RAS.
The project made part of Проект выполнялся в рамках RTT № 0120.0852225 and was carried out in close cooperation of Open Technologies Perm Office and RAS UD IMM.
The RTT included the analysis of modern high density computing systems, the equipment and software available at IT market.
Ultimately, the cluster was built with a blade server HP BladeSystem c7000 platform. The final computing system for В RAS UD IMM consisted of 32 dual-processor HP Proliant BL 460c nodes installed on quad-core Intel Xeon 5450 chips with stroking frequency of 3,0 GHz. Red Hat Enterprise Linux 5 and Microsoft Windows Compute Cluster Server 2008 operational systems were chosen. Infiniband technologies were used to build the communication environment.
The peak productivity of the cluster is 3 teraflops and the real one (defined byLinpack test) is 2,36 teraflops.
It is meant for solving fundamental and applied problems in such fields as mathematics and mechanics, physics of metals, electrical and thermal physics, hard bodies chemistry and high temperature electrical chemistry, metallurgy; geology and mining, geochemistry and geophysics, ecology, biology, economy and medicine.
According to Mikhai Ludvogovich Goldsteein, the head computer equipment department at RAS UD IMM, "the created system allows solving a wide circle of issues, their complexity depending on the development of science and technology, enhance requirements to the relevance of physical contents of mathematic models, tough criteria to the results reliability, exclusion of mistakes in discrete representation of initial differentiated models and obtaining decisions with guaranteed result.
Modernization of the regional information and computer network core Integrated development and introduction of information security elements
Solutions
High Performance Computing Systems
