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

254

Technical DataSheet MVS-10P MP

Technical Info MVS-10P MP

RSC Group has deployed a new MVS-10P MP supercomputer based on RSC PetaStream™ massively parallel architecture at the Joint Supercomputer Center of the Russian Academy of Sciences (JSCC RAS). It is the first project in Russia and CIS based on the highest-performing Intel® Xeon Phi™ 7120D coprocessors. The new MVS-10P MP system has expanded JSCC RAS computing resources that are still based on MVS-10P supercomputer build on RSC Tornado cluster architecture and has been deployed by RSC in 2012. The total performance of liquid-cooled RSC supercomputers at JSCC RAS now exceeds 600 Teraflops.
The new MVS-10P MP supercomputer based on 64x the latest highest-performing Intel® Xeon Phi™ 7120D coprocessors provided RAS scientists with 3,904 x86 computing cores with total performance of 77.33 Teraflops in addition to resources of MVS-10P supercomputer. The com-puting part takes only 0.18 m3, so its specific performance-per-volume ratio is 560 Teraflops/m3, which is currently a record for HPC industry. 
 

CHALLENGES

  • Demand for higher performance and small footprint. 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. However, the most important requirement to supercomputer solutions is not only high performance but also minimization of space usage, power consumption, high energy efficiency and low total cost of ownership (TCO).
  • Protection of software investments. Need to use existing applications and programming models in the process of transition to future multi-core platforms.
  • Readiness to future technologies. Need to start mastering skills and problem solving approaches using future multi-threaded systems that will inevitably appear in the coming Exascale era.

SOLUTIONSЕ

  • Achieving high performance and small footprint. Responding to these challenges, JSCC RAS has selected a computing system based on revolutionary ultra high-density high performance RSC PetaStream™ massively parallel architecture with RSC's direct liquid cooling. Solutions based on this architecture enable record-breaking computing density of 1.2 Petaflops and execution of 250,000 threads on 1024 x86 computing nodes in a single cabinet taking only 1 m2 space. Besides, RSC PetaStream™ has set the world record of power density – over 400 kW per cabinet.
  • High energy efficiency and record-breaking PUE. Energy-efficient direct liquid cooling technology developed by RSC is tested with time and has been used by customers in numerous projects since 2009. It provides very high energy efficiency level with average power usage efficiency (PUE) of 1.06. It means that no more than 6% of power is used for system cooling. PUE is a critical factor for exascale computers with power consumption level of several dozen megawatts.
  • Protection of software investments. RSC PetaStream™ solution based on x86 architecture that is a de-facto industry standard throughout the world provides protection of investments in software development and optimization. The developers will be able to use existing programming model in new software for future computing systems thus getting the opportunity to run existing applications on Exascale level supercomputers. At the same time, flexibility of RSC PetaStream™ architecture enables testing of innovative programming models that might be more efficient for Exascale computing. RSC PetaStream™ solution is supplied with Intel® Cluster Studio XE 2013 tool package for Linux and a set of optimized libraries.
  • TCO reduction and reliability. In addition to highest application performance provided by RSC PetaStream™, this solution enables double reduction of power consumption compared to traditional cluster systems in modeling tasks with high level of parallelism. Therefore, users can significantly reduce TCO while obtaining leading performance with small solution footprint. RSC PetaStream™ implements an innovative subsystem based on the leading industry standard of 400V DC power developed in cooperation with Emerson Electric. This subsystem enables power distribution efficiency over 90% improving overall energy efficiency of the system, reduction of operating costs and overall reliability of the complex.
  • Monitoring and control. RSC PetaStream™ provides extended hardware and software system monitoring and control features using integrated RSC BasIS software stack for user convenience. The system uses Intel® Node Manager Technology for control and power consumption monitoring of every node enabling supercomputer centers to establish flexible policies of power optimization reducing power costs and costs of computing system operation.

RESULTS

RSC Group has developed and deployed the MVS-10P MP supercomputing system based on RSC PetaStream™ massively parallel architecture. The new MVS-10P MP system has expanded JSCC RAS computing resources that are still based on MVS-10P supercomputer with RSC Tornado cluster architecture that was deployed by RSC in 2012. The total performance of liquid-cooled RSC supercomputers at JSCC RAS now exceeds 600 Teraflops.

MVS-10P MP solution is an implementation of massively parallel architecture using best available industrial components of the highest integration level and unique RSC PetaStream™ system architecture for high-performance computing developed by RSC specialists. Every node of MVS-10P MP massively parallel system is independent, based on 61-core Intel® Xeon Phi7120D coprocessor with 16GB high-speed GDDR5 memory and working under embedded Linux OS. MVS-10P MP also uses the following Intel® server products — Intel® Xeon® E5-2600 family processors, Intel® server boards and Intel® SSD DC S3500 solid state drives for enterprise data centers. All 64 computing nodes are joined with high-speed interconnects based on Infiniband FDR technology providing small-footprint high-performance solution with 15.6 thousand threads per 1 cabinet based on x86 architecture. With massively parallel computing system MVS-10P MP based on advanced RSC PetaStream™ architecture, specialists and scientists of numerous research institutes and organizations of the Russian Academy of Sciences now can develop and optimize applications for future systems with massively parallel architectures and new multi-core processors.

   

New MVS-10P MP computing system based on RSC PetaStream™ massively parallel architecture will enable RAS scientific teams to develop the latest applications for breakthrough R&D in such fields as medicine, bioengineering, astrophysics, chemistry, radioelectronics, control systems and others.

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:
July 2014 - December 2014
Project Location:
Moscow, Russia



Share Page