The fastest computer in the world. Modern supercomputers

People do not fly on Mars, the cancer has not yet cured, did not get rid of oil dependence. And yet there are areas where humanity has reached incredible progress over the past decades. Computational power of computers - just one of them.

Twice a year, experts from the National Laureren's Laboratory in Berkeley and the University of Tennessee publish the Top-500, which offer a list of the most productive supercomputers of the world.

A little running forward, we suggest you try to taste these numbers in advance: the performance of representatives of the first ten top is measured with tens of quadrillion flops. For comparison: Eniac, the first computer in history, possessed with a capacity of 500 flops; Now average personal Computer It has power in hundreds of gigaflops (billions of flops), iPhone 6 has a capacity of approximately 172 gigaflops, and the PS4 game console is 1,84 teraflops (Trillion flops).

Armed with the latest "Top 500" from November 2014, the editorial office of Naked Science decided to figure out that the 10 most powerful supercomputers of the world represent from themselves, and to solve which tasks it requires such a grandiose computing power.

10. CRAY CS-STORM

• Location: USA
• Productivity: 3.57 Petaflops
• Theoretical Maximum Performance: 6.13 Petaflops
• Power: 1.4 MW

Like almost all modern supercomputers, including each of those presented in this article, CS-STORM consists of a variety of processors united in a single computing network According to the principle of mass-parallel architecture. In reality, this system is a plurality of racks ("cabinets") with electronics (nodes consisting of multi-core processors) that form entire corridors.

CRAY CS-STORM is a whole series of supercomputer clusters, but one of them still stands out against the rest of the rest. In particular, it is a mysterious CS-STORM, which uses the US government for unknown purposes and in an unknown place.

It is only known that American officials bought extremely effective from the point of view of energy consumption (2386 megaflops per 1 watt) CS-STORM with a total number of nuclei of almost 79 thousand from the American company Cray.

On the manufacturer's website, however, it is said that CS-STORM clusters are suitable for high-performance computing in the field of cybersecurity, geospatial intelligence, image recognition, seismic data processing, rendering and machine learning. Somewhere in this series, probably the application of government CS-STORM has settled.

CRAY CS-STORM

9. VULCAN - BLUE GENE / Q

• Location: USA
• Productivity: 4.29 Petaflops
• Theoretical Maximum Performance: 5.03 Petaflops
• Power: 1.9 MW

The "volcano" was developed by the American company IBM, refers to the Blue Gene family and is located in the Livemore National Laboratory named after E. Lawrence. The supercomputer belonging to the US Department of Energy consists of 24 racks. Function cluster began in 2013.

In contrast to the already mentioned CS-STORM, the scope of the "volcano" is well known - these are various research, including in the field of energy, such as modeling natural phenomena and analyzing a large amount of data.

Various scientific groups and companies can access the supercomputer on the application that needs to be sent to the center of innovation in the field of high-performance computing (HPC Innovation Center), based in the same Livemore National Laboratory.

Supercomputer vulcan.

8. JUQUEEN - BLUE GENE / Q

• Location: Germany
• Productivity: 5 Petaflops
• Theoretical Maximum Performance: 5,87 Petaflops
• Power: 2.3 MW

From the moment of launching in 2012, Juqueen is the second power supercomputer in Europe and the first in Germany. Like "Volcano", this supercomputer cluster was developed by IBM as part of the Blue Gene project, and the same generation of Q.

There is a supercomputer in one of the largest research centers in Europe in Julikha. Used respectively - for high-performance computing in various scientific research.

Supercomputer juqueen

7. Stampede - Poweredge C8220

• Location: USA
• Performance: 5.16 Petaflops
• Theoretical Maximum Performance: 8.52 Petaflops
• Power: 4.5 MW

Stampede located in Texas is the only one in the top ten Top-500 cluster, which was developed by the American company Dell. The supercomputer consists of 160 racks.

This supercomputer is the most powerful among those that are used exclusively for research purposes. Access to Stampede's capacity is open to scientific groups. A cluster is used in the widest spectrum of scientific areas - from the exact imaging of the human brain and the prediction of earthquakes before identifying patterns in music and language structures.

Supercomputer stampede

6. PIZ DAINT - CRAY XC30

• Location: Switzerland
• Productivity: 6.27 Petaflops
• Theoretical Maximum Performance: 7.78 Petaflops
• Power: 2.3 MW

The Swiss National Supercomputer Center (CSCS) boasts the most powerful supercomputer in Europe. Piz Daint, named after the Alpine Mountain, was developed by Cray and belongs to the XC30 family, which is the most productive.

PIZ DAINT is used for various research purposes, such as computer simulation in the field of high-energy physics.

Supercomputer Piz Daint.

5. MIRA - BLUE GENE / Q

• Location: USA
• Performance: 8,56 Petaflops
• Theoretical Maximum Performance: 10.06 Petaflops
• Power: 3.9 MW

The supercomputer "World" was developed by IBM as part of the Blue Gene project in 2012. The separation of high-performance computing of the Argonian national laboratory in which the cluster is located, was created using public financing. It is believed that the growth of interest in supercomputer technologies from Washington in the late 2000s and early 2010s is due to rivalry in this area with China.

Located on 48 MIRA racks is used for scientific purposes. For example, the supercomputer is used for climatic and seismic modeling, which allows to obtain more accurate data on earthquake prediction and climate change.

Supercomputer Mira.

4. K Computer

• Location: Japan.
• Performance: 10,51 Petaflops
• Theoretical Maximum Performance: 11.28 Petaflops
• Power: 12.6 MW

Developed by Fujitsu and located at the Institute of Physico-Chemical Research in Kobe, K Computer is the only Japanese supercomputer present in the top ten Top-500.

At one time (June 2011), this cluster took the first position in the ranking, one year becomes the most productive computer in the world. And in November 2011, K Computer became the first in history, which was able to reach the power above 10 Petaflops.

The supercomputer is used in a number of research tasks. For example, to predict natural disasters (which is relevant for Japan due to the increased seismic activity of the region and the country's high vulnerability in the case of a tsunami) and computer simulation in the field of medicine.

Supercomputer K.

3. SEQUOIA - BLUE GENE / Q

• Location: USA
• Performance: 17,17 Petaflops
• Theoretical Maximum Performance: 20,13 Petaflops
• Power: 7.8 MW

The most powerful of the four supercomputers of the Blue Gene / Q family, which came to the top ten rankings, is located in the United States in the Livemore National Laboratory. IBM developed Sequoia for the National Nuclear Security Administration (NNSA), which was required to be a high-performance computer for quite a specific goal - modeling nuclear explosions.

It is worth mentioning that real nuclear tests are prohibited since 1963, and computer simulation is one of the most acceptable options for continuing research in this area.

However, the supercomputer power was used to solve other, much more noble tasks. For example, the cluster was able to put record records in cosmological modeling, as well as when creating an electrophysiological model of the human heart.

Supercomputer sequoia

2. Titan - Cray XK7

• Location: USA
• Productivity: 17,59 Petaflops
• Theoretical Maximum Performance: 27,11 Petaflops
• Power: 8.2 MW

The most productive from ever created in the West supercomputers, as well as the most powerful computer cluster under the brand of Cray, is located in the United States at the Oak Ridge National Laboratory. Despite the fact that the supercomputer at the disposal of the American Ministry of Energy is officially available for any scientific research, in October 2012, when Titan was launched, the number of applications exceeded any limits.

Because of this, a special commission was convened in the Ocridge laboratory, which of the 50 applications selected only the 6 most "advanced" projects. Among them, for example, modeling the behavior of neutrons in the heart of a nuclear reactor, as well as forecasting global climatic changes for the next 1-5 years.

Despite its computing power and impressive dimensions (404 square meters), Titan long lasted on a pedestal. Already six months after TRIUMPA in November 2012, the pride of Americans in the field of high-performance computing suddenly pushed the exhaust from the east, unprecedentedly overtaking the previous ranking leaders.

Supercomputer Titan.

1. TIANHE-2 / Milky Way-2

• Location: China
• Performance: 33.86 Petaflops
• Theoretical Maximum Performance: 54.9 Petaflops
• Power: 17.6 MW

Since its first launch of Tianhe-2, or "Milky-2", for about two years already is the Top-500 leader. This monster is almost two times in performance No. 2 in the ranking - the Titan supercomputer.

Developed by the Defense Scientific and Technical University of the People's Liberation Army of the People's Republic of China and Inspur, Tianhe-2 consists of 16 thousand nodes with a total number of nuclei of 3.12 million. RAM All this is a colossal design that occupies 720 square meters is 1.4 petabytes, and the storage device is 12.4 petabytes.

"Milky Wall-2" was constructed on the initiative of the Chinese government, so there is nothing surprising that its unprecedented power serves, apparently, the needs of the state. It was officially stated that the supercomputer is engaged in various modeling, analysis huge number data, as well as providing state security of China.

Given the secrecy inherent in military projects of the PRC, it remains only to guess which application from time to time receives "Milky Way-2" in the hands of the Chinese army.

Supercomputer tianhe-2

Time reading:7 min.

Until now, humanity has never achieved the heatcopes of Mars, did not invent the elixir of youth, the cars can not be overwhelmed over the ground, but there are several areas in which we all succeeded. Creating powerful supercomputers - just such a sphere. To estimate the power of the computer, you need to determine which key parameter is responsible for this characteristic. This parameter is flops - the value that shows how many operations can perform a PC in one second. It is, on the basis of this magnitude, our magazine is a large rating and put the most powerful computers in the world for 2017.

Supercomputer power - 8.1 Pflpl / s

This computer stores data that is responsible for the security of the United States military structure, it is also responsible for the state of the availability of the nuclear attack, if necessary. Two years ago, this car was one of the most powerful and expensive in the world, but today TRINITY has displaced newer devices. The system on which this supercomputer is working - Cray XC40, thanks to it, the device and can "issue" such a number of operations per second.

Mira.

Supercomputer power - 8.6 Pflpl / s

Cray has released another supercomputer - Mira. The US Department of Energy ordered the release of this machine to coordinate its work. The sphere in which Mira is working - the industry and the development of research capabilities. In a second, this supercomputer can calculate 8.6 Petaflops.

Supercomputer power - 10.5 pflpl / sec

The name of this device immediately describes the power, the Japanese word "kei" (K) means ten quadrillion. This figure almost accurately describe its productive power - 10.5 Petaflops. "Fishka" of this supercomputer is its cooling system. A water cooling is used, which reduces the consumption of energy reserves and reduces the layout speed indicators.

Supercomputer power - 13.6 Pflpl / sec

Fujitsu - the company from the country of the rising sun, did not stop at work, releasing the supercomputer K Computer, they immediately began to new project. This project was the OakForest-PACS supercomputer, which belong to the new generation machines (Generation Knights Landing). It was ordered by Tokyo and Tsukubi Universities. According to the initial plan, the memory of the device was to be 900 TB, and the performance of OakForest-PACS would be 25 quadrelones of operations per second. But with a lack of financing, many aspects were improved, therefore the supercomputer power was 13.6 Petaflops per second.

Cori.

Supercomputer power - 14 pflch / s

Last year, Cori was on the sixth line in the list of the most powerful supercomputers in the world, but with a crazy technology for the development of technologies, he gave way to one position. This supercomputer is located in the United States, in the National Laboratory named after Lawrence and Berkeley. Scientists from Switzerland, with the help of Cori were able to develop a 45-cubic quantum computing machine. The production capacity of this supercomputer is 14 Petaflops per second.

Supercomputer power - 17.2 Pflpl / s

Scientists from around the world long time converged that Sequoia is the fastest supercomputer on the planet. And this is not just so, because it is able to produce arithmetic calculations, for which people in the amount of 6.7 billion it would take 320 years, in one second. Truly, the dimensions of the car are striking - it takes more than 390 square meters and its composition includes 96 racks. Sixteen thousand trillion operations or other words 17.2 Petaflops - the production capacity of this supercomputer.

Titan

Supercomputer power - 17.6 Pflpl / sec

In addition, this supercomputer is one of the fastest on the planet, it is also very energy efficient. Energy efficiency indicator is 2142.77 megaflops per watt energy required for consumption. The reason for such low power consumption is NVIDIA accelerator, which provides up to 90% of the power required for computation. In addition, the NVIDIA accelerator significantly reduced the area that this supercomputer occupied, now he needs only 404 square meters.

Supercomputer Power - 19.6 Pflpl / Sec

The first launch of this apparatus took place in 2013, in Switzerland, in the city of Lugano. Now the geolocation of this supercomputer is the Swiss National Center for Supercomputers. PIZ DAINT is a combination of all better characteristics The above machines, it has a very high energy efficiency indicator and it is very fast in calculations. Only one characteristic leaves much to be desired - the dimensions of this supercomputer, it takes 28 huge racks. PIZ Daint is able to work with a computing power of 19.6 Petaflops per second.

Supercomputer power - 33.9 Pflpl / sec

This apparatus has a romantic Tianhe name, which from Chinese, translated, means "Milky Way". Tianhe-2 was the fastest computer in the list of 500 fastest and powerful supercomputers. It can calculate 2507 arithmetic operations, which translated into Petaflops will be 33.9 pflplops. Specialization in which this computer is used is construction, it calculates operations related to building and gasket. From the first launch in 2013, this computer does not lose its position in the lists, which proves that this is one of best cars in the world.

Supercomputer power - 93 Pflpl / sec

Sunway Taihulight is the fastest supercomputer in the world, except for its huge calculation speed, it is also famous for its huge dimensions - it covers an area of \u200b\u200bmore than 1000 square meters. international Conference 2016, which was held in Germany, recognized this supercomputer in the world in the world and he still does not have a serious competitor in this regard. Its speed is three times higher than the TIANHE-2 indicators, the supercomputer nearest to him in this regard!

Technical progress does not stand still, it develops with space speed, affects many aspects of human life, has many both positive and negative parties. For a person, the most affordable technique has become the most different types: Computers, robots and appliances. But the main goal of any equipment is to simplify the life of a person, the technique should not become meaningless entertainment, which will only spend your time.

The first ATLAS supercomputer appeared in the early 1960s and was installed in the University of Manchester. It was several times less powerful than modern home computers. In our review, the "dozen" is collected the most powerful in the history of supercomputers. True, in connection with the technologies rapidly developing in this field, these powerful cars are annoyed for 5 years.

The performance of modern supercomputers is measured in Petaflops - a unit of measurement showing how many floating-point operations per second performs a computer. Today we will talk about the ten most expensive modern supercomputers.

1. IBM Roadrunner (USA)

$ 130 million
Roadrunner was built by IBM in 2008 for the National Laboratory in Los Alamos (New Mexico, USA). He became the first computer in the world whose average working performance exceeded 1 Petaflops. At the same time, he was designed for maximum productivity In 1.7 Petaflops. According to the SUPERMICRO Green500 list, in 2008 Roadrunner was the fourth energy efficiency supercomputer in the world. ROADRUNNER was written off on March 31, 2013, after which it was replaced by a smaller and more energy efficient supercomputer called Cielo.

2. Vulcan Bluegene / Q (USA)

$ 100 million
Vulcan is a supercomputer consisting of 24 separate blocks, which was created by IBM for the Ministry of Energy and is installed in the National Laboratory of Louurens Livermor, California. It has peak performance in 5 poveflops and is currently ninth speed supercomputer in the world. Vulcan entered into operation in 2013 and is now used by the Livemore National Laboratory for Research in Biology, Plasma Physics, Climatic Names, Molecular Systems, etc.

3. SuperMUC (Germany)

$ 111 million
SuperMUC Currently is the 14th speed supercomputer in the world. In 2013, he was 10th, but the development of technology does not stand still. Nevertheless, he is in this moment It is the second speed supercomputer in Germany. Supermuc is managed by Leibnitsky Supercomputer Center at the Bavarian Academy of Sciences next to Munich.

The system was created by IBM, it works on the Linux shell, contains more than 19,000 Intel and Westmere-EC processors, and also has peak performance just over 3 PETAFLOPS. SuperMUC is used by European researchers in areas of medicine, astrophysics, quantum chromodynamics, computational hydrodynamics, computing chemistry, analysis of genome and earthquake modeling.

4. Trinity (USA)

$ 174 million
It would be possible to expect that a similar supercomputer (given what it is built for) should be insanely expensive, but thanks to the development of technologies, the Trinity price has become possible. The US government is going to use Trinity in order to maintain the efficiency and safety of America's nuclear arsenal.

Trinity, which is currently being built, will be a joint project of the Sandi National Laboratory and the Los Alamos National Laboratory as part of the program for the forecast modeling and computational processing of the National Nuclear Security Administration.

5. SEQUOIA BLUEGENE / Q (USA)

$ 250 million
SEQUOIA supercomputer class Bluegene / Q was developed by IBM for the National Nuclear Security Administration, within the framework of the program modeling and data computing program. It was commissioned in June 2012 in the Livemorm National Laboratory and became the fastest supercomputer in the world at that time. Now it ranks third in the world in terms of speed (theoretical peak of the performance of Sequoia - 20 poveflops or 20 trillion computing per second).

A stable computer operates at 10 podaflops. Sequoia is used to support various scientific applications, studying astronomy, energy, human genome, climate change and nuclear weapons.

6. ASC PURPLE AND BLUEGENE / L (USA)

$ 290 million
These two supercomputers worked together. They were built by IBM and installed in 2005 in the Livemore National Laboratory. From the operation they were bred in 2010. At the time of creation of ASC Purple, it took 66th place in speed in the top 500 supercomputers list, and Bluegene / L was the previous generation of the Bluegene / Q model.

ASCI PURPLE was built for the fifth stage of the forecast modeling and computing processing of the US Department of Energy, as well as the National Nuclear Security Administration. Its purpose was to simulate and replacing the real tests of weapons of mass destruction. Bluegene / L used to predict global climate change.

7. Sierra and Summit (USA)

$ 325 million
NVIDIA and IBM will soon help America to return leading positions in the field of super-speed supercomputer technologies, scientific research, as well as economic and national security. Both computers will be completed in 2017.

Currently, the fastest supercomputer in the world is Chinese Tianhe-2, which is capable of reaching power in 55 poveflops, which is twice as much as the device in second place in the list. Sierra will produce more than 100 poveflops, while Summit will be able to develop 300 petaflops.

Sierra, which will be installed in the Livemore National Laboratory, will ensure the safety and effectiveness of the country's nuclear program. Summit will replace the outdated Titan supercomputer at the Oak Ridge National Laboratory and will be intended for testing and supporting scientific applications around the world.

8. TIANHE-2 (China)

$ 390 million
Chinese Tianhe-2 (which translates as "Milky Way-2") is the fastest supercomputer in the world. The computer developed by the team of 1300 scientists and engineers is located in the National Supercomputer Center in Guangzhou. It was built by the Chinese defense scientific and technical university of China's People's Liberation Army. TIANHE-2 is able to perform 33,860 trillion computing per second. For example, one hour of the supercomputer calculations is equivalent to 1000 years of work 1.3 billion people. A machine for modeling and analyzing government security systems is used.

9. Earth Simulator (Japan)

$ 500 million
The "Earth Simulator" was developed by the Japanese government in 1997. The cost of the project is 60 billion yen or about $ 500 million. Earth Simulator was completed in 2002 for the Aerospace Research Agency of Japan, the Japanese Research Institute for Atomic Energy and the Japanese Center for Marine and Ground Studies and Technologies.

ES was the fastest supercomputer in the world from 2002 to 2004, and it also serves to work with global climate models, to assess the consequences of global warming and evaluating the problems of geophysics of the Earth's crust.

10. Fujitsu K (Japan)

$ 1.2 billion
The world's most expensive supercomputer is just a fourth in the world in the world (11 Petaflops). In 2011, he was the fastest supercomputer in the world. Fujitsu K, located at the Institute of Advanced Computing Technologies Riken, is about 60 times faster than the Earth Simulator. It takes about $ 10 million per year, and uses a supercomputer 9.89 MW (how many electricity use 10,000 country houses or one million personal computers).

It is worth noting that modern scientists stepped so far that they have already appeared.

Previously, the first-place supercomputer K Computer will move in third place. Its performance is 11.28 Pflpls (see Figure 1). Recall that the floating-point operations per second, flops is a unit of measurement of computers, which shows how many floating-point operations per second can perform this computing system.

K Computer is the joint development of the Institute of Physico-Chemical Research Rikagaka Kenkyo (Riken) and Fujitsu. It was created as part of the High-Performance Computing Infrastructure Initiative (high-performance computer computing infrastructure), headed by the Japanese Ministry of Education, Culture, Sports, Science and Technology (Mext). The supercomputer is installed on the territory of the Institute of Advanced Computing Sciences in the Japanese city of Kobe.

The supercomputer is based on a distributed memory architecture. The system consists of more than 80,000 computing nodes and is placed in 864 racks, each of which accommodates 96 computing nodes and 6 I / O nodes. Nodes containing by one processor and 16 GB of RAM are connected to each other in accordance with the six-dimensional loop / torus topology. In total, the system uses 88,128 eight-year SPARC64 VIIIFX processors (705 024 kernels) produced by Fujitsu using 45 nm technology.

This general purpose supercomputer provides a high level of performance and support for a wide range of applications. The system is used to conduct research in the field of climate change, prevent natural disasters and medicine.

A unique water cooling system reduces the likelihood of equipment failure and reduce overall energy consumption. Energy savings are achieved through the use of highly efficient equipment, heat and electricity cogeneration systems and array solar batteries. In addition, the reuse mechanism of the wastewater from the cooler reduces the negative impact on the environment.

The building in which K Computer is located is seismically resistant and able to withstand the earthquake of magnitude 6 or more points along the Japanese scale (0-7). For more efficient placement of racks with equipment and cables, the third floor of 50 × 60 m is completely freed from the carrier columns. Modern technologies The construction has allowed to provide a permissible level of load (up to 1 t / m 2) to install racks whose weight can reach 1.5 tons.

Supercomputer sequoia

SEQUOIA supercomputer installed in the Livemore National Laboratory. Lawrence, has a performance of 16.32 Pflplops and takes the second line of the rating (see Figure 2).

This Petaflop supercomputer, developed by IBM based on Blue Gene / Q, was created for the National Nuclear Security Administration (NNSA) as part of the Advanced Simulation and Computing Program (high-tech modeling and computer calculations).

The system consists of 96 racks and 98,304 computing nodes (1024 nodes per rack). Each node includes a 16-nuclear PowerPC A2 processor and 16 GB of DDR3 RAM. In general, 1,572,864 processor kernels and 1.6 PBITI in memory are used. Nodes are connected together in accordance with the five-dimensional torus topology. The system occupied area is 280 m2. General power consumption is 7.9 MW.

On the supercomputer sequoia for the first time in the world, scientific calculations were carried out for which the computational power was required more than 10 PFShpls. Thus, the HACC cosmological modeling system required about 14 PFShls at start-up in mode 3.6 trillion particles, and during the launch of the CardiOD project code for modeling the electrophysiology of the human heart, the performance reached almost 12 PFShpls.

Supercomputer Titan.

The most fast supercomputer was recognized by the Titan supercomputer, established in the US National Lab (ORNL) in the United States. In the test tests of LINPACK, its performance was 17.59 plpls.

Titan implements a CPU-GPU hybrid architecture (see Figure 3). The system consists of 18,688 knots, each of which is equipped with a 16-nuclear aMD processor Opteron and NVIDIA TESLA K20X graphics accelerator. A total of 560,640 processors are used. Titan is an update previously operated in the Ornl supercomputer Jaguar and occupies the same server cabinets (with a total area of \u200b\u200b404 m 2).

Ability to use already existing systems Nutrition and cooling made it possible to save about 20 million dollars during the construction. The power consumption of the supercomputer is 8.2 MW, which is 1.2 MW more than Jaguar indicators, while its performance when performing floating-point operations above almost 10 times.

Titan will first be used to conduct research in the field of science on materials and nuclear energy, as well as studies relating to improving the efficiency of internal combustion engines. In addition, with its help, climate change modeling and analysis of potential strategies to eliminate the negative consequences associated with them will be performed.

The most "green" supercomputer

In addition to the Top500 rating, aimed at identifying the most high-performance system, there is a Green500 rating, where the most "green" supercomputers are marked. Here, the basis of energy efficiency (IMFLFC / W) is adopted. At the moment ( last edition rating - November 2012) The Green500 leader is the Beacon supercomputer (253rd in Top500). The indicator of its energy efficiency is 2499 MFLFS / W.

Beacon works on the basis of coprocessors Intel Xeon. PHI 5110P and Intel Xeon E5-2670 processors, so peak performance can reach 112,200 GFSHs with a total power consumption of 44.9 kW. Xeon PHI 5110P coprocessors provide high performance at low power consumption. Each coprocessor has a capacity of 1 TFLOPS (when performing double accuracy operations) and supports up to 8 GB of GDDR5 class memory with throughput 320 Gb / s.

The Passive Xeon Phi 5110P cooling system is designed for TDP 225 W, which is the ideal indicator for high density servers.

Supercomputer eurora

However, in February 2013, there were reports that Eurora Supercomputer, located in the city of Bologna (Italy), surpassed Beacon (3150 MFHHlops / Watt against 2499 IMFLFS / W).

EuroRa built by Eurotech and consists of 64 knots, each of which includes two intel processor Xeon E5-2687W, two NVIDIA TESLA C20 GPU accelerator and other equipment. The dimensions of such a node do not exceed the laptop dimensions, but their performance is 30 times higher, and energy consumption is 15 times lower.

High energy efficiency in EURORA is achieved by using multiple technologies. Water cooling contributes the greatest contribution. Thus, each supercomputer node is a kind of sandwich: central equipment from below, a water heat exchanger in the middle and another unit of electronics from above (see Figure 4).

Such high results are provided with the use of materials with good thermal conductivity, as well as an extensive network of cooling channels. When installing a new computing module, its channels are combined with the cooling system channels, which allows the supercomputer configuration depending on specific needs. According to the manufacturers, the risk of leaks is excluded.

The power supply of EURORA supercomputer elements is carried out by 48 volt sources direct current, the introduction of which made it possible to reduce the number of energy transformations. Finally, warm water allocated from computing equipment can be used for other purposes.

Conclusion

The supercomputer industry is actively developing and puts all new and new records of productivity and energy efficiency. It should be noted that it is in this industry as anywhere else, today the technologies of liquid cooling and 3D-modeling are widely used, since there is a task to compose a heavy-duty computing system that would be able to function in a limited volume with minimal energy losses.

Yuri Khomutsky - Chief Engineer of the projects of the company Ay-Teko. You can contact him at: [Email Protected] . The article uses the materials of the Internet portal about the data center centers "www.aboutdc.ru - solutions for the data center".

People do not fly on Mars, the cancer has not yet cured, did not get rid of oil dependence. And yet there are areas where humanity has reached incredible progress over the past decades. Computational power of computers - just one of them.

Twice a year, experts from the National Laureren's Laboratory in Berkeley and the University of Tennessee publish the Top-500, which offer a list of the most productive supercomputers of the world.

As a key criterion, this rating uses a characteristic that has long been considered one of the most objective in assessing the power of supercomputers - flops, or the number of floating point operations per second.

A little running forward, we suggest you try to taste these numbers in advance: the performance of representatives of the first ten top is measured with tens of quadrillion flops. For comparison: Eniac, the first computer in history, possessed with a capacity of 500 flops; Now the average personal computer has a capacity of hundreds of Gigaflops (billions of flops), iPhone 6 has a capacity of approximately 172 gigaflops, and the PS4 game console is 1,84 teraflops (Trillion flops).

Armed with the latest "Top 500" from November 2014, the editorial office of Naked Science decided to figure out that the 10 most powerful supercomputers of the world represent from themselves, and to solve which tasks it requires such a grandiose computing power.

10. CRAY CS-STORM

Location: USA
Productivity: 3.57 Petaflops
Theoretical Maximum Performance: 6.13 Petaflops
Power: 1.4 MW

Like almost all modern supercomputers, including each of those presented in this article, CS-STORM consists of a variety of processors united in a single computing network on the principle of mass-parallel architecture. In reality, this system is a plurality of racks ("cabinets") with electronics (nodes consisting of multi-core processors) that form entire corridors.

CRAY CS-STORM is a whole series of supercomputer clusters, but one of them still stands out against the rest of the rest. In particular, it is a mysterious CS-STORM, which uses the US government for unknown purposes and in an unknown place.

It is only known that American officials bought extremely effective from the point of view of energy consumption (2386 megaflops per 1 watt) CS-STORM with a total number of nuclei of almost 79 thousand from the American company Cray.

On the manufacturer's website, however, it is said that CS-STORM clusters are suitable for high-performance computing in the field of cybersecurity, geospatial intelligence, image recognition, seismic data processing, rendering and machine learning. Somewhere in this series, probably the application of government CS-STORM has settled.

Cray CS-STORM / © Cray

9. VULCAN - BLUE GENE / Q

Location: USA
Productivity: 4.29 Petaflops
Theoretical Maximum Performance: 5.03 Petaflops
Power: 1.9 MW

The "volcano" was developed by the American company IBM, refers to the Blue Gene family and is located in the Livemore National Laboratory named after E. Lawrence. The supercomputer belonging to the US Department of Energy consists of 24 racks. Function cluster began in 2013.

In contrast to the already mentioned CS-STORM, the scope of the "volcano" is well known - these are various research, including in the field of energy, such as modeling natural phenomena and analyzing a large amount of data.

Various scientific groups and companies can access the supercomputer on the application that needs to be sent to the center of innovation in the field of high-performance computing (HPC Innovation Center), based in the same Livemore National Laboratory.

Supercomputer Vulcan / © Laura Schulz and Meg Epperly / LLNL

8. JUQUEEN - BLUE GENE / Q

Location: Germany
Productivity: 5 Petaflops
Theoretical Maximum Performance: 5,87 Petaflops
Power: 2.3 MW

From the moment of launching in 2012, Juqueen is the second power supercomputer in Europe and the first in Germany. Like "Volcano", this supercomputer cluster was developed by IBM as part of the Blue Gene project, and the same generation of Q.

There is a supercomputer in one of the largest research centers in Europe in Julikha. Used respectively - for high-performance computing in various scientific research.

Supercomputer Juqueen / © Jülich SuperComputing Center (JSC)

7. Stampede - Poweredge C8220

Location: USA
Performance: 5.16 Petaflops
Theoretical Maximum Performance: 8.52 Petaflops
Power: 4.5 MW

Stampede located in Texas is the only one in the top ten Top-500 cluster, which was developed by the American company Dell. The supercomputer consists of 160 racks.

This supercomputer is the most powerful among those that are used exclusively for research purposes. Access to Stampede's capacity is open to scientific groups. A cluster is used in the widest spectrum of scientific areas - from the exact imaging of the human brain and the prediction of earthquakes before identifying patterns in music and language structures.

Stampede supercomputer / © Texas Advanced Computing Center

6. PIZ DAINT - CRAY XC30

Location: Switzerland
Productivity: 6.27 Petaflops
Theoretical Maximum Performance: 7.78 Petaflops
Power: 2.3 MW

The Swiss National Supercomputer Center (CSCS) boasts the most powerful supercomputer in Europe. Piz Daint, named after the Alpine Mountain, was developed by Cray and belongs to the XC30 family, which is the most productive.

PIZ DAINT is used for various research objectives like computer simulation in the field of high-energy physics.

Supercomputer piz daint / © blogs.nvidia.com

5. MIRA - BLUE GENE / Q

Location: USA
Performance: 8,56 Petaflops
Theoretical Maximum Performance: 10.06 Petaflops
Power: 3.9 MW

The supercomputer "World" was developed by IBM as part of the Blue Gene project in 2012. The separation of high-performance computing of the Argonian national laboratory in which the cluster is located, was created using public financing. It is believed that the growth of interest in supercomputer technologies from Washington in the late 2000s and early 2010s is due to rivalry in this area with China.

Located on 48 MIRA racks is used for scientific purposes. For example, the supercomputer is used for climatic and seismic modeling, which allows to obtain more accurate data on earthquake prediction and climate change.

Supercomputer mira / © Flickr

4. K Computer

Location: Japan.
Performance: 10,51 Petaflops
Theoretical Maximum Performance: 11.28 Petaflops
Power: 12.6 MW

Developed by Fujitsu and located at the Institute of Physico-Chemical Research in Cobe K Computer is the only Japanese supercomputer present in the top ten Top-500.

At one time (June 2011), this cluster took the first position in the ranking, one year becomes the most productive computer in the world. And in November 2011, K Computer became the first in history, which was able to reach the power above 10 Petaflops.

The supercomputer is used in a number of research tasks. For example, to predict natural disasters (which is relevant for Japan due to the increased seismic activity of the region and the country's high vulnerability in the case of a tsunami) and computer simulation in the field of medicine.

Supercomputer K / © Fujitsu

3. SEQUOIA - BLUE GENE / Q

Location: USA
Performance: 17,17 Petaflops
Theoretical Maximum Performance: 20,13 Petaflops
Power: 7.8 MW

The most powerful of the four supercomputers of the Blue Gene / Q family, which came to the top ten rankings, is located in the United States in the Livemore National Laboratory. IBM developed Sequoia for the National Nuclear Security Administration (NNSA), which was required to be a high-performance computer for quite a specific goal - modeling nuclear explosions.

It is worth mentioning that real nuclear tests are prohibited since 1963, and computer simulation is one of the most acceptable options for continuing research in this area.

However, the supercomputer power was used to solve other, much more noble tasks. For example, the cluster was able to put record records in cosmological modeling, as well as when creating an electrophysiological model of the human heart.

Supercomputer sequoia / © bob hirschfeld / llnl

2. Titan - Cray XK7

Location: USA
Productivity: 17,59 Petaflops
Theoretical Maximum Performance: 27,11 Petaflops
Power: 8.2 MW

The most productive from ever created in the West supercomputers, as well as the most powerful computer cluster under the brand Cray is located in the USA in the Oak Ridge National Laboratory. Despite the fact that the supercomputer at the disposal of the American Ministry of Energy is officially available for any scientific research, in October 2012, when Titan was launched, the number of applications exceeded any limits.

Because of this, a special commission was convened in the Ocridge laboratory, which of the 50 applications selected only the 6 most "advanced" projects. Among them, for example, modeling the behavior of neutrons in the heart of a nuclear reactor, as well as forecasting global climatic changes for the next 1-5 years.

Despite its computing power and impressive dimensions (404 square meters), Titan long lasted on a pedestal. Already six months after TRIUMPA in November 2012, the pride of Americans in the field of high-performance computing suddenly pushed the exhaust from the east, unprecedentedly overtaking the previous ranking leaders.

Supercomputer Titan / © olcf.ornl.gov

1. TIANHE-2 / Milky Way-2

Location: China
Performance: 33.86 Petaflops
Theoretical Maximum Performance: 54.9 Petaflops
Power: 17.6 MW

Since its first launch of Tianhe-2, or "Milky-2", for about two years already is the Top-500 leader. This monster is almost two times in performance No. 2 in the ranking - the Titan supercomputer.

The Tianhe-2, developed by the Defense Scientific and Technical University of the People's Liberation Army and Inspur "Tianhe-2" consists of 16 thousand nodes with a total of 3.12 million nuclei. The operational memory of all this is a colossal design, occupying 720 square meters, is 1.4 petabytes, and a storage device is 12.4 petabytes.

"Milky Wall-2" was constructed on the initiative of the Chinese government, so there is nothing surprising that its unprecedented power serves, apparently, the needs of the state. It was officially stated that the supercomputer is engaged in various modeling, an analysis of a huge amount of data, as well as ensuring the state security of China.

Given the secrecy inherent in military projects of the PRC, it remains only to guess which application from time to time receives "Milky Way-2" in the hands of the Chinese army.

Supercomputer tianhe-2 / © popsci.com