03 July 2013

RSC supercomputers continue to lead in energy efficiency among Russian HPC systems in Green500

Moscow, July 3, 2013. — Upon publication of the new edition of Green500 list of world most energy-efficient supercomputers (http://green500.org/lists/green201306), RSC Group, the leading Russian developer and full-cycle builder of an innovative HPC and data center solutions, maintained presence of three of its systems in this rating (since previous edition published in November 2012). For the second time, RSC computing clusters became the most energy-efficient Russian HPC systems in Green500 list. The same RSC supercomputers are also in Top500 rating of world the best-performing systems being listed there twice (November 2012, June 2013) and represent about 40% of all Russian supercomputers in Top500 and Green500.

MVS-10P supercomputer based on RSC Tornado architecture as a prototype of 10 PFLOPS computing system has been developed by RSC specialists and deployed at the Joint Supercomputer Center of the Russian Academy of Sciences (JSCC RAS). It keeps 30th place in Green500 again with 1949.3 MFLOPS/W, which is the best energy-efficiency level for all Russian systems. This is the record value for Russia and CIS, and it is 5.5 times higher than the previous energy-efficiency record for this region (which was delivered by RSC Tornado architecture based liquid cooled system also).

The prototype of powerful MVS-10P system in JSCC RAS is based on innovative RSC Tornado architecture with liquid cooling of all components placed on Intel® S2600JF server boards, including Intel® Xeon® E5-2690 processors, Intel® Xeon Phi SE10X coprocessors and Intel® SSD 330 solid-state drives. It has peak performance of 523.8 TFLOPS with LINPACK results of 375.7 TFLOPS. It is the largest European supercomputer with Intel Xeon Phi coprocessors out of two first such projects being deployed outside the US by RSC specialists in the end of 2012.

RSC Tornado SUSU supercomputer deployed by RSC specialists in Supercomputer Center of South Ural State University (Chelyabinsk, Russia) last year holds 50th position in the new edition of Green500 list (previously 40th place) with energy-efficiency of 999.5 MFLOPS/W. It is de facto the most energy-efficient university computing cluster in Russia and CIS and the largest European university HPC system with Intel Xeon Phi coprocessors. This powerful system consisting of 192 computing nodes based on innovative RSC Tornado architecture with liquid cooling, Intel® Xeon® processors and Intel Xeon Phi coprocessors has 236.8 TFLOPS peak performance and 146.8 TFLOPS LINPACK performance.

SKIF-Aurora SUSU, another supercomputer based on RSC Tornado architecture with liquid cooling and Intel® Xeon® processors, has been deployed at SUSU Supercomputer Center by RSC specialists four years ago. In current edition of Green500 rating it holds 181st place with 349.7 MFLOPS/W of energy-efficiency value. This system is included in the list of world the most energy-efficient computers for the fourth time (since November, 2011). This supercomputer has 117 TFLOPS peak and 100.4 TFLOPS performance by LINPACK test.

“Once again we managed to get such good result in terms of the number of Russian supercomputers at high positions in Green500 thanks to unique experience of RSC specialists in development and deployment of innovative liquid cooled solutions, specifically the high-density energy-efficient HPC systems based on RSC Tornado architecture using standard Intel server boards, high-performance Intel Xeon processors and Intel Xeon Phi coprocessors.”, said Alexey Shmelev, Chief Operational Officer, RSC Group.

Innovative liquid cooling system and RSC know-how used in RSC Tornado architecture based solutions helped to achieve the highest energy-efficiency value with record PUE level of 1.06 and unique in the industry computing density of 200 TFLOPS per 80 cm x 80 cm x 42U rack or 156 TFLOPS/m3 being de facto a new world record for х86 architectures. For example, with such high density it was possible to install all computing equipment of MVS-10P prototype system in JSCC RAS with 0.5 PFLOPS peak performance in three racks only, as well as a 10 PFLOPS supercomputer will require just 50 standard server cabinets. Such a high computing density is necessary for ExaFLOPS level supercomputers containing hundreds of server racks with power consumption over 100 kW per rack. Unique level of computing and power density in RSC™ Tornadoarchitecture proves that required results can be achieved when creating HPC systems for ExaScale era.

 

Unique features of RSC Tornado based solutions

 

The new stage in development of innovative RSC Tornado architecture for energy efficient data centers and supercomputing systems allowed RSC Group’s specialists to implement world's first direct liquid cooling for widely available Intel® Xeon® based standard server boards (from various manufacturers), initially designed for traditional systems with air cooling of electronic components, with the latest Intel Xeon Phi coprocessors. This is the third generation of RSC’s energy-efficient solutions for High Performance Computing, cloud environments and data centers.

Powerful solutions with high computing density based on RSC Tornado architecture with direct liquid cooling are designed to solve various customers’ needs. The product line includes RSC DCmicro (16 to 64 nodes), RSC DCmini (64 to 256 nodes) and RSC Data Center (from 2 high dense racks up to dozens of PFLOPS).

Among unique features of RSC Tornado architecture and solutions there are the following:

         Up to 128 х86 servers in a standard 42U rack 80х80х200 cm;

         High-density blade server design based on widely available boards;

         Record-breaking energy efficiency. The system allows achieving an industry-leading Power Usage Effectiveness (PUE) of less than 1.06 (the ratio of total amount of power used by a computer data center facility to the power delivered to computing equipment.). This means that maximum 5.7% of the consumed energy by a datacenter will be used for the cooling of the system;

         Record-breaking computing effectiveness ratio — more than 96% for Intel® Xeon® E5-2690 processors in LINPACK benchmark (Intel® Turbo Boost Technology is always on, which provides for clock speed gain of up to 400 MHz in LINPACK);

         Removal of over 100 kW of power from each rack using RSC’s unique liquid cooling design;

         Possibility to use Intel's top-performing server processors with heat dissipation of 135W, e.g. Intel Xeon E5-2690 processor (2.9 GHz, 8 cores) and the newest highest performing Intel Xeon Phi coprocessors;  

         High peak performance – over 47 TFLOPS per rack based on Intel x86 architecture with AVX Instruction set and up to 200 TFLOPS using Intel Xeon Phi coprocessors;

         High density – 74 TFLOPS per square meter (using Intel Xeon only) and 312 TFLOPS per sq. m (with Intel Xeon Phi coprocessors);

         High scalability – up to PFLOPS level (dozens of racks);

         Cost effectiveness – operating cost reduction by 60 percent (energy cost saving due to RSC solution operation);

         Compact size – multifold datacenter floor area reduction comparing to traditional air-cooled installations;

         Ability to use accelerators and coprocessors (e.g. Intel Xeon Phi);

         Fully integrated RSC BasIS software stack is optimized for high performance computing;     

         Performance and scalability of RSC Tornado based solutions are confirmed by Intel® Cluster Ready certificate (http://software.intel.com/en-us/cluster-ready).