The new edition of the HPCG benchmark rating of actual performance of the world's most powerful supercomputers features 5 systems produced by RSC Group from Russia. RSC holds the leading position with 70% share of all Russian supercomputers in this list.
Austin (Texas, USA), SC15, November 20th, 2015 — The new edition of the HPCG benchmark rating (http://hpcg-benchmark.org/) created for more accurate assessment of actual performance of the world's most powerful supercomputers, features 5 systems produced by RSC Group from Russia. RSC holds the leading position with 70% share of all Russian supercomputers in this list, which currently contains 7 Russian systems. The current edition of HPCG rating includes 64 supercomputers from different regions of the world. RSC holds about 8% of this global list.
It is feasible to notice that the Top500 global rating of world's most powerful supercomputers created in 1993 has been developed on the basis of LINPACK (High Performance LINPACK, HPL) benchmark developed at the same time. However, in later years this rating and HPL benchmark became subject of heavy critics. Numerous experts and supercomputer users have stated that LINPACK benchmark does not conform to actual computing system workload profile. Besides, some organizations refuse to allocate the entire supercomputer for the time necessary to run the benchmark, and with increased cluster performance running LINPACK requires more and more time, up to several days. In 2013 makers of Top500 list proposed a new benchmark for supercomputers – High Performance Conjugate Gradient (HPCG), which is supposed to provide a better estimate of computing system performance on a broader range of actual applications. The first rating based on HPCG was published in June 2014.
The new edition of the HPCG rating includes the following RSC supercomputers:
- Polytechnic RSC Tornado (Saint Petersburg Polytechnic University named after Peter the Great, SPbPU) based on RSC Tornado cluster architecture with the result of 10.8 Teraflops (39th place);
- MVS-10P (Joint Supercomputer Center of the Russian Academy of Sciences, JSCC RAS) based on RSC Tornado cluster architecture with the result of 4.9 Teraflops (47th place);
- RSC Tornado SUSU (South Ural State University, SUSU) based on RSC Tornado cluster architecture with the result of 3.6 Teraflops (54th place);
- Polytechnic RSC PetaStream (Saint Petersburg Polytechnic University named after Peter the Great, SPbPU) based on RSC PetaStream massively parallel architecture with the result of 3.1 Teraflops (57th place);
- MVS-10P (Joint Supercomputer Center of the Russian Academy of Sciences, JSCC RAS) based on RSC PetaStream massively parallel architecture with the result of 1.2 Teraflops (47th place);
"We created HPCG benchmark to enable better and more relevant metric for HPC systems that closely mimic important application behaviour and data access patterns. It’s good to see that number of supercomputers being ranked by HPCG is constantly growing worldwide as well as the Russian share at this benchmark grew up so visibly thanks to five RSC systems.”- said Prof. Jack Dongarra, Director of the Innovative Computing Laboratory at the University of Tennessee and author of HPL and HPCG benchmarks.
"Now when 5 RSC systems are in the new HPCG performance rating, this is a clear proof of high efficiency of our solutions in running of actual applications and solving the important research and other tasks by Russian scientific and educational organizations," – said Alexey Shmelev, CEO of RSC Group.
Two RSC computing systems from HPCG rating have been developed and installed at one of the largest and most innovative HPC centers in Russia at the Saint Petersburg Polytechnic University named after Peter the Great with total peak performance over 1.1 Petaflops.
Polytechnic RSC Tornado cluster system has demonstrated performance of 658 Teraflops by LINPACK benchmark. This SPbPU supercomputing cluster based on the RSC Tornado architecture with direct liquid cooling has the peak performance of 829 Teraflops. It is the first project based on the latest Intel® Xeon® E5-2600 v3 server processors in Russia and CIS. Polytechnic RSC Tornado includes 712 dual-processor nodes containing 1424 high-performance Intel® Xeon® E5-2697 v3 server processors (14 cores, 2.6 GHz frequency each), Intel® S2600KP and Intel® S2600WT server boards for this processor generation, and the latest Intel® SSD DC S3500 solid-state drives for corporate data centers. These HPC resources are used not only to solve relevant scientific tasks but also for cloud, VDI (Virtual Desktop Infrastructure) and graphical services.
The second part of the hybrid HPC complex contains a unique high-density massive parallel RSC PetaStream system with direct liquid cooling and total performance of 295 Teraflops. Polytechnic RSC PetaStream supercomputer has demonstrated the performance of 170.5 Teraflops by LINPACK benchmark. This SPbPU system developed and produced by RSC specialists in Russia is based on 60-core Intel® Xeon Phi™ 5120D and Intel® Xeon® E5-2600 v2 family processors with Intel server boards and Intel SSD DC S3500 solid-state drives for enterprise data centers.
A unique feature of RSC PetaStream is the massively parallel architecture based on Intel® Xeon Phi™ enabling execution of up to 250,000 parallel threads on 1024 computing nodes based on x86 architecture in a single cabinet taking 1 sq. m. The RSC PetaStream massively parallel architecture holds the world records of computing and power density and compactness. 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. RSC PetaStream enables development of the latest applications to make breakthroughs in bioengineering, astrophysics, chemistry, radioelectronics and control systems.
Another new MVS-10P MP system based on RSC PetaStream massive parallel architecture with direct liquid cooling has been installed by RSC specialists at the Joint Supercomputer Center of the Russian Academy of Sciences (JSCC RAS). It is the first project based on high-performance Intel® Xeon Phi™ 7120D in Russia and CIS. MVS-10P MP is based on massive parallel RSC PetaStream architecture using the following Intel® server products: 61-core Intel® Xeon Phi™ 7120D processors, Intel® Xeon® E5-2600 family processors, Intel server boards and Intel SSD DC S3500 solid-state drives for corporate data centers.
The new MVS-10P MP system has expanded JSCC RAS resources that are still based on MVS-10P supercomputer based on RSC Tornado cluster architecture with direct liquid cooling that has been deployed by RSC in 2012. Total performance of RSC supercomputers with direct liquid cooling at JSCC RAS now exceeds 600 Teraflops.
RSC Tornado SUSU cluster with peak performance of 473 Teraflops runs at the South Ural State University in Chelyabinsk since 2009. This RSC system is based on RSC Tornado cluster architecture with direct liquid cooling, Intel Xeon server processors, Intel server boards and Intel solid-state drives.