«Tornado SUSU» Supercomputer
|
Complex solution of company «RSC» – one of the leading Russian developers and integrators of supercomputers, http://www.rscgroup.ru. «Tornado SUSU» supercomputer is a physical part of the «SKIF-Aurora SUSU» supercomputer, however designed as a standalone item for its reliability.
It consists of 480 compact and powerful computing blade-modules with liquid cooling.
«Tornado SUSU» supercomputer have direct liquid cooling and high energy efficiency (50th position in Green500).
|
|
|
   
|
|
TOP500 Rating
|
TOP500 is the list of the most powerful supercomputers in the world. The project started in 1993 and publishes an updated list of supercomputers twice a year. This project aims to provide a sound basis for identifying and tracking trends in high performance computing.
The supercomputer «Tornado SUSU» took the following positions in the TOP500:
| Editorial Rating |
Position |
| November 2012 |
170 |
| June 2013 |
249 |
| June 2013 |
127 |
| June 2014 |
156 |
| November 2014 |
190 |
| June 2015 |
244 |
| November 2015 |
348 |
| June 2016 |
498 |
|
Green500 Rating
|
The Green500 provides rankings of the most energy-efficient supercomputers in the world. Typically energy-efficiency is measured as FLOPS per watt.
Supercomputer «Tornado SUSU» took the following positions in Green500:
| Editorial Rating |
Position |
| June 2013 |
50 |
| November 2013 |
64 |
| November 2014 |
107 |
| June 2015 |
150 |
| November 2015 |
174 |
| June 2016 |
189 |
|
TOP50 Rating
|
To help correctly navigate the world of high-performance computing systems and be able to quickly track the development trends of this area, the Interagency Supercomputer Center of the Russian Academy of Sciences and the Scientific Research Computing Center of the Moscow State University named after MV Lomonosov in May 2004, with the support of the Russian company T-Platforms, launched a joint project on the formation of the list of 50 most powerful CIS computers TOP50. The list includes 50 computer systems installed in the CIS and showed the highest productivity on the Linpack test by the time the list was published.
Supercomputer «Tornado SUSU» took the following positions in TOP50:
| Editorial Rating |
Position |
| 02.04.2013 |
4 |
| 24.09.2013 |
4 |
| 01.04.2014 |
3 |
| 23.09.2014 |
5 |
| 31.03.2015 |
6 |
| 28.09.2015 |
7 |
| 29.03.2016 |
8 |
| 26.09.2016 |
8 |
| 04.04.2017 |
8 |
| 25.09.2017 |
8 |
| 03.04.2018 |
8 |
| 24.09.2018 |
9 |
| 02.04.2019 |
11 |
|
HPCG Rating
South Ural State University with its supercomputer «Tornado SUSU» has taken 54th place in a HPCG rating. The HPCG (High Performance Conjugate Gradients) Benchmark project is an effort to create a more relevant metric for ranking HPC systems. The computational and data access patterns of HPL are no longer driving computer system designs in directions that are beneficial to many important scalable applications. HPCG is designed to exercise computational and data access patterns that more closely match a broad set of important applications, and to give incentive to computer system designers to invest in capabilities that will have impact on the collective performance of these applications.
The supercomputer «Tornado SUSU» took the following positions in HPCG:
| Editorial Rating |
Position |
| November 2015 |
54 |
| June 2016 |
65 |
| November 2016 |
83 |
| June 2017 |
94 |
| November 2017 |
97 |
| June 2018 |
108 |
Technical features:
| Quantity of computing units/processors/cores: |
480/960/384/29184 |
| Type of processor: |
Intel Xeon X5680 (Gulftown, 6 cores with frequency 3.33 GHz) — 960 pcs. |
| Type of coprocessor: |
Intel Xeon Phi SE10X (61 cores with frequency 1.1 GHz) — pcs. |
| RAM: |
16.9 TB |
| Disk memory: |
300 TB, Solid-state drives Intel SSD, Parallel storage system Panasas ActiveStor 11 |
| Type of communicative network: |
InfiniBand QDR (40 Gbit/s) |
| System network: |
InfiniBand QDR (40 Gbit/s) |
| Service network: |
Gigabit Ethernet |
| Peak performance SKIF Avrora + Tornado SUSU: |
473.6 TFlops |
| Performance SKIF Avrora + Tornado SUSU on LINPACK test: |
288.2 TFlops |
| Operating system: |
CentOS 6.2 |
|
Features
|
Processing Performance
- High peak processing performance is up to 47,5 TFLOPS per one rack based on the Intel x86 architecture with the AVX instructions set.
- High package density — 74 TFLOPS per square m.
- High scalability — up to the level of several PFLOPS (ten racks).
- Compactness — decrease of the data center area in a few times.
- Ability to use accelerators.
Power efficiency
- Record energy efficiency — the PUE (Power Usage Effectiveness) index of the energy efficiency reaches the record value in the HPC industry that is 1.06 (the ratio of the power consumption of the entire system to the power consumption of the electronic components). This means that the system cooling does not exceed the 5.7% barrier of the whole power consumption.
- Effective cooling: direct liquid cooling on level of motherboards provides much more efficient heat removal in comparison to air cooling.
- Low price of possession: up to 60% of economy of electric power.
Compatibility
- Full compatibility with architecture x86.
Reliability
- Total lack of moving parts in processor units (absence of noise and vibration).
- Three-level system of monitoring and control.
Fail-safe feature
- Backup of all subsystems.
- Mode of «hot» change of equipment.
|
|
|
