It appears that JavaScript is disabled in your web browser. Please enable it to have full system functionality available.

156

Technical DataSheet MVS-10P

Technical Info MVS-10P

  • Performance: 
    0.524 PFLOPS (Peak) / 0.376 PFLOPS (LINPACK)
  • Nodes / Сores (CPU + CoCPU) / Threads (CPU + CoCPU): 
    208 / 3328 + 24960 / 6656 + 101504
  • CPU: 
    416x Intel® Xeon® E5-2690 (8 cores, 2.9 GHz)
  • Coprocessor: 
    416x Intel® Xeon Phi™ SE10X (60 cores, 1.1 GHz, 8 GB)
  • Motherboard: 
    Intel® Server Board S2600JF(F)
  • Total RAM of the nodes: 
    13.2 TB (DDR3-1600, low voltage green memory)
  • Total disk storage of the nodes: 
    24.8 TB, Solid State Drives Intel® SSD
  • Interconnect: 
    InfiniBand FDR, 56 Gb/s
  • Service network: 
    2x 1GigE independent sensor and control networks
  • Occupied Area: 
    20.00 м.кв. (215.28 sq. ft.)

RSC group has deployed a prototype of a 10 pflops-range supercomputing system at Joint Supercomputer Center (JSCC) of the Russian academy of science (RAS). MVS-10P supercomputer based on RSC tornado architecture is the largest system with latest intel® Xeon phi™ coprocessors in Russia, CIS and Europe. The new supercomputer at JSCC has set several records in different nominations. The current peak performance of this powerful system, which includes 208 computing nodes based on innovative RSC Tornado architecture with direct liquid cooling, Intel® server boards, Intel® Xeon® E5-2690 processors and Intel Xeon Phi coprocessors, is 523.8 tflops. Its energy efficiency has reached Russia’s and CIS record level of 1,949.3 mflops/w, which is 5.5 times increased comparing to the previous energy efficiency record for the region (also achieved by an RSC Tornado-based system with liquid cooling). New MVS-10P supercomputer takes 30rd place in green500 rating (November, 2012) as well as 59th position at top500 list (November, 2012) while holding 2nd place at top50 of Russia/CIS the most powerful systems (April, 2013).

JSCC RAS is state of the art high performance computing center with competent staff running the highly ranked top500 and green500 powerful and energy efficient supercomputer. The facility uses RSC tornado based system with innovative liquid cooling and newest Intel Xeon Phi coprocessors which provide impressive high performance capabilities and energy efficient solutions to solve very demanding science research and engineering problems.”, said Jack Dongarra, one of the leading worldwide HPC experts,  after  JSCC RAS  visiting in April, 2013. mr. Dongarra has been very impressed by high level of energy efficiency and world record computing (up to 181 tflops per rack) and power (up to 100 kW per rack) density while having a very small footprint because of RSC Tornado liquid cooling technology implemented.

The new powerful supercomputer MVS-10P based on RSC Tornado architecture  triples  the   performance of JSCC’s computing complex, allows achieving some unique  characteristics in performance, energy efficiency and manageability, supports the research and solving of cutting-edge supercomputing tasks and will help solve urgent and fundamental scientific problems at the Russian Academy of Science with the goal to further develop scientific and economic potential of Russia.” Gennady Savin, Academician, RAS Presidium Member, JSCC RAS General Manager

JOINT SUPERCOMPUTER CENTER
Joint Supercomputer Center the Russian Academy of Sciences (founded in 1999 and having branches in st. Petersburg and Kazan) is one of the most powerful supercomputing enter providing high-performance  computing  services in Russia for scientific and educational institutions. JSCC employs highly skilled scientists, programmers and engineers. JSCC's main objectives are:
• supporting scientific research in insti- tutes with modern computing and tele- communication  resources;
• providing methodology to researchers that need high-performance comput- ing systems and information process- ing tools; helping prepare highly quali- fied scientific specialists;
• providing access to modern digital li- braries, archives, databases and data collections;
• performing research in the development of system and application software, as well solving highly complex tasks.

CHALLENGES
Request for performance. JSCC provides its services to over 100 organizations; 181 research groups work in the center with more than 900 tasks in fundamental and applied areas which require more and more performance. so the jscc was urgently needed a new computing power for high-performance supercomputing system having an essential performance reserve to solve not only current actual tasks but for the nearest future too.
Small footprint, energy efficiency and low TCO. because of future 10 ptflos scaling, the most important require- ments to jscc supercomputer are minimization of space and energy consumption, while providing high computing power and energy efficiency, as well as low total cost of ownership (TCO).
 
SOLUTIONS
• High system performance. to achieve planned objectives jscc made its choice in favor a computing system based on high-performance Intel® Xeon® E5-2690 processors with latest Intel®  Xeon Phi™ coprocessors. the performance of com- munication interconnect is insured by the modern Infiniband FDR technology.
• The highest computing density. MVS-10P supercomputer system based on innovative RSC Tornado architecture and developed by  RSC Group.  The use of an advanced direct liquid cooling system allows reaching industry’s the highest computing density of 181 TFLOPS  per  80*80cm*42u  rack,  or 141 TFLOPS/m3, which is 3.8 times higher than the previous world record for х86 architectures. Such a high computing density is required for exaflops-range supercomputers, which will consist of hundreds of racks each consuming over 100kW of power. The unique power and performance density level achieved by MVS-10P prototype system based on RSC Tornado architecture proves technical feasibility of building high-performance systems with required characteristics for Exascale era. The high computing density requires advanced liquid cooling methods for standard server boards, which were developed by RSC Group for the first time in the world.
• High energy efficiency and record PUE. the use of modern intel® Xeon® E5-2690 processors and latest Intel® Xeon Phi™ coprocessors together with the advanced liquid cooling system ensures energy cost savings of up to 60%. Energy efficiency of new JSCC supercomputer has reached Russia’s and CIS record level of 1,949.3 MFLOPS/W, which is a 5.5 times increase compared to the previous energy efficiency record for the region (also achieved by an RSC Tornado-based system with liquid cooling). The industry's record power usage effectiveness (PUE) level less of 1.1. The new JSCC supercomputer with current peak performance 523.8 TFLOPS consumes just 222 kW of power.
• Energy efficiency increasing and re- ducing of TCO. The new supercomputer at JSCC is also equipped with a flexible power and energy management RSC ecosystem technology   which additionally decreases the total cost of ownership (TCO) by 25%. the RSC’s ecosystem technology was first dem- onstrated by RSC Group at the leading European supercomputing show ISC’12 and deployed in the MVS-10P supercomputer for the first time.
• Increased reliability. Liquid cooling ensures precision heat removal, which extends service life of electronic components and increases fault tolerance of the whole solution. The absence of moving parts, such as fans and traditional hard disk drives (Intel Solid State Drives (SSD) are used instead) makes the supercomputer virtually noiseless and prevents vibration, which significantly increases the system reliability.

RESULTS
RSC Group has developed and introduced one of the best-performing supercomputing systems in CIS at Joint Supercomputer Center (JSCC) of the Russian Academy of Science (RAS).
the JSCC's supercomputer currently consists of 3,328 Intel® Xeon® E5-2690 processor cores and over 25,000 Intel Xeon Phi coprocessor SE10X cores. Each node supports a large capacity of energy efficient RAM – 64GB (DDR3-1600, low voltage green memory). A high-performance interconnect is based on advanced Infiniband FDR technology (56Gbps). At that time, the supercomputer’s maximum peak performance is 523.8 TFLOPS with 375.7 TFLOPS performance on LiNpacK test. JSCC system is the most energy efficient supercomputer in Russia and CIS region while it based on industry standard x86 architecture with no GPU accelerators. Such high performance is ensured by the advanced direct liquid cooling system, which also significantly increases fault tolerance and energy efficiency.

 

The new powerful supercomputer MVS-10P based on RSC Tornado architecture triples the performance of JSCC’s computing complex, allows achieving some unique characteristics in performance, energy efficiency and manageability, supports the research and solving of cuttingedge supercomputing tasks and will help solve urgent and fundamental scientific problems at the Russian Academy of Science with the goal to further develop country’s scientific and economic potential.

Project information

Joint supercomputer center (JSCC)

Joint supercomputer center (JSCC) is a state scientific organization founded in 1996 under a joint resolution by the presidium of the Russian academy of science (RAS), the ministry of science and technology of the Russian federation, the ministry of education of the Russian federation and the Russian fund of fundamental Research (RFFR)

Project Dates:
October 2012 - February 2013
Project Location:
Moscow, Russia




Share Page