Who wouldn’t want to be able to choose one from a bunch of supercomputers to take home with them? Or am I wrong? Of course, prepare the electricity bill, because the cost is going to be astronomical.
According to its name, many users may misunderstand the supercomputer, like a personal computer with the highest performance, for example. However, this is not at all accurate.
Do you want to know what a supercomputer is? We are going to see it here today, knowing all its history, its main characteristics and, of course, the uses that are given to it. Also, to finish surprising you, we are going to leave you a top of the most powerful supercomputers that currently exist, so you can know their amazing real performance.
What are Supercomputers in computing? Definition
Supercomputer is that computer equipment that has a very high computing capacity, is designed with a specific purpose and has a design that is really great.
To achieve this, what is done is to train a system of several very powerful computers, which are part of the same structure and, therefore, are linked together, so that both power and performance are multiplied. At present, these systems are also designated as computers or high-performance computing environments.
Although they are large and complex devices, the truth is that what is sought from them is that they work with a good number of floating operations per second, for which there is a concept, the FLOPS. This measures, as we say, the number of operations that the machines are capable of doing and corresponds to a unit of measurement of performance.
This type of computers are introduced “on the market” in the 70s. Most of them showed up thanks Seymour Cray, a dominant company at the time that was responsible for its design and manufacture. Initially they were nothing more than full-speed scalar processors.
The next step is taken by competitiveness with minicomputers. This consisted of the inclusion of some vector processors working in parallel, between 4 and 16. This is how they began to lower their price and improve, expanding the number and type of users who used them.
Seymour Cray and the super computers
It is from here that we begin to see changes to consider in the teams. From vector vectors to massive parallel processor systems with common CPUs counted by the thousands.
Later, when the microprocessor was properly developed and tested, it was moved to replace the processor with this, considerably reducing the number of parts to be used in the design of each supercomputer. During the last decades they have opted for the microprocessor-servisor.
Thus, we have that at present the concept of supercomputer is maintained in terms of function and power, but both its appearance and the type and quantity of parts used have been considerably modified, giving clusters of fine-tuned computing devices running on high-capacity conventional processors, of thousands and even millions of cores, that combine, yes, with interconnections of a special nature.
What are supercomputers for and what are their uses?
Since their birth, supercomputers have had very specific goals. Sand designed to perform calculation operations in massive volume in very short periods of time (millions per second).
That is the basic function that all of them have, although, the purpose can be very different. For example, we would talk about weather predictions, quantum physics problems, physical, space and nuclear simulations, research on climate change, research on the biological processes of the human body and its implications in diseases and to work on problems that have yet to be solved. …
So, The main users of these machines are large corporations and agencies such as academics, government, military and, in some cases, business for huge multinational corporations.
What main characteristics does this type of super computers present?
Supercomputers, as we’ve mentioned, come with all kinds of bits and pieces. Although, what characterizes them the most would be:
- Processor. Processing speed is key to achieving millions and millions of operations per second. For this, in addition to a particularly efficient model, a very large number of cores is required.
- RAM. In line with the above, you need to have a spectacular RAM memory, of several Tebibytes.
- Cooling system. Cooling systems are essential to combat the very high temperatures that occur in the system. To get an idea, some processors are capable of dissipating 10 times the heat of a stove disk.
- Interconnection capacity. It is necessary that, in large environments, thousands of users can be connected at the same time.
Other characteristics that we can define of these machines is that lack the capacity for social penetration (no interest is given), that they are very difficult to use, so much so that it can only be done by specialized personnel, that their size is enormous, that they have achieved the achievement of many milestones of a social nature, especially in research and that there is less thousand around the world and do not require peripherals.
The 10 Most Powerful Supercomputers in the World
After having told you about the incredible specifications of this type of equipment, we could not finish without telling you which have been the most powerful supercomputers that have been manufactured.
It was built by Cry.Inc in conjunction with King Abdullah University of Science and Technology, which it now owns. It has been in use since 2015 in these facilities.
With a Linux OS, the Cray, it is impossible not to be surprised to learn that it has almost 200,000 cores in its processor, which is, by the way, the Xeon E5-2698v3 16C. This provides a throughput of 5537 TFLOP / s. More surprising is knowing its power requirement, which is “only” 2800 KW.
Cray.Inc is in charge of the design and manufacture, in 2015, of Hazel Hen, a 41-cabin supercomputer with the Xeon E5-2680v3 12C 185088 cores. Its performance is 7.42 PetaFLOPS / s. The operating system is the same as in the previous case. In this case we have that, although the average power is unknown, it is quite efficient, with a known maximum figure of 1512 KW.
Cray.Inc is also in charge of this manufacture that goes directly to Switzerland, where it has been located since 2013, specifically at the National Supercomputing Center. Today is the best supercomputer on the continent.
It achieves it thanks to its processor Xeon E5-2670 8C that works at 2.6 GHz and 115984 cores, its 28 cabins and a theoretical performance of 7.8 petaFLOPS that we have seen that, in operation, double, having already reached the 19.6 petaFLOPS. We repeat SO.
It is used by both the US Department of Energy, the National Nuclear Safety Administration and the Los Alamos National Laboratory, where it is currently located.
We repeat manufacturer and operating system and we find a Xeon E5-2698v3 16C 2.3GHz including 380,000+ cores. The processing it achieves is superior to 27000 TFLOPS / s theoretical. Its memory is 365056 GB.
We found it in China, in the Supercomputing Center. Its manufacturer was NUDT and began working in Guangzhou in 2013.
Its OS is Kylin Linux, the processor is a Intel Xeon E5-2692v2 12C 2.2 GHz 3.12 million cores. It is capable, thus, of surrendering to 33.86 TFlop / s, working at 17808 kW.
We finish the list with another supercomputer located in China, at the Supercomputing Center located in Wuxi. Manufacturing, in this case, came from the hand of NRCPC and began operating very recently, in 2016.
Its operating system is exclusive, Sunway RaiseOS 2.0.5, which is based on Linux. It has more than 10 million cores in its Sunway SW26010 260C processor at 1.45 GHz. We obtain a performance of 93 TFlops to 15371 kW.
Another great machine, this time Japanese, that works with a SPARC64 VIIIfx at 2 GHz that has 700000 cores. The performance it offers reaches 10.5 TFLOP per second. In return, the energy power amounts to 12659 KW. The OS is Linux. We have it at the RIKEN Advanced Computer Science Center.
IBM was in charge of the manufacture of this device in 2012, specific and exclusive for the National Nuclear Safety Administration of the United States, being located in the Lawrence Livermore National Laboratory.
Your processor is a Power BQC 16C 1.5 million cores as well as a performance of more than 17 TFlop / s. Works with Linux. Its power consumption is 7890 kW.
Performs at 17.59 petaflops thanks to its Opteron 6274 16C at 2.2 GHz with more than half a million cores. It has Cray Linux Enviroment and an Nvidia GPU. It is manufactured by Cray.Inc and found at the Oak Ridgem National Laboratory in the United States.
It was manufactured by IBM in 2012 and since then it has been located at the Argonne National Laboratory. The processor, in this case, is a Power BQC 16C with a staggering almost 800,000 cores. Performance exceeds every second 8586 TFLOPS / s. It has Linux and works at 3945 KW.
If you have any questions, leave them in the comments, we will answer you as soon as possible, and it will also be of great help to more members of the community. Thank you! 😉