Supercomputer Explained
How can you become a part of the world’s supercomputer
What is a Supercomputer?
A supercomputer is exactly what it sounds like: an immensely powerful computer. It performs at the highest operational rate of the time, millions of times faster than a regular desktop PC. These machines have the super ability to deal with the most complicated calculations, especially in science and engineering. They can handle enormous databases and perform a vast amount of computation. Tasks like weather forecasting simulating nuclear weapons explosions or reconstructing billions of the universe’s history are just a few areas of the supercomputers’ usage.
The supercomputers invariably have some specific features that cannot be compared to that of a desktop PC. Although, they both have hard drives, memory spaces, and processors, they are quite different in a number of ways. A supercomputer has thousands of processors, hundreds of RAM gigabytes ,and hard drives that allows for up to thousands of gigabytes of storage space. Moreover, a supercomputer has multiple central processing units (CPUs), which enable it to process faster circuit switching and accomplish more tasks at once.
Apart from being able to work faster, a supercomputer has the unique feature of operating differently. Instead of the serial processing system that common computer uses, a super computer uses the parallel processing. This means that, it splits problems into pieces and works on these pieces at once. The most powerful supercomputers are in fact multiple computers connected to perform parallel processing.
As for the software, most of the supercomputers run quite ordinary operating systems, just like desktop PCs. The most common one is the open-source Linux. A good number of applications for these machines are written in traditional scientific programming languages such as Fortran, as well as popular, more modern languages C and C++.
You can even make a supercomputer of your own by connecting off-the-shelf computers with a high-speed local area network (LAN), so they work in a broadly similar way. That kind of supercomputer is called a cluster. A grid is another type of supercomputer. It is also made up of separate computers. However, the machines are in different places and connected through the Internet. A grid is an example of a distributed computing with the computing power spread across multiple locations.
Top Five Supercomputers
Twice a year, the TOP500 organization publishes a list of the world’s fastest supercomputers. The machines are ranked by floating-point operations per second (FLOPS) — a measurement of how fast the computer can do math equations. The most recent version is from November 2017.
- The Sunway TaihuLight of China’s National Research Center of Parallel Computer Engineering & Technology (NRCPC) tops the list with 93 petaflops which equals 93 quadrillion flops. This machine installed at the National Supercomputing Center in Wuxi. It is ranked first the fourth time in a row.
- Tianhe-2 (Milky Way-2) is another Chinese machine. The system is developed by China’s National University of Defense Technology (NUDT) and deployed at the National Supercomputer Center in Guangzho, China. With 33.86 petaflops, it is still the world’s number two supercomputer.
- Piz Daint is a system installed at the Swiss National Supercomputing Centre (CSCS) in Lugano, Switzerland. Last year it was upgraded with NVIDIA Tesla P100 GPUs, and now, with the performance of 19.59 petaflops. It is the most powerful machine in Europe.
- Gyoukou is deployed at Japan’s Agency for Marine-Earth Science and Technology. It can achieve a result of 19.14 petaflops, and peak performance of 28.19 petaflops. The system’s 19,860,000 cores represent the highest level of concurrency ever recorded on the TOP500 rankings of supercomputers.
- Titan. The fastest US machine is ranked as the fifth with 17 petaflops performance. Titan is a five-year-old Cray XK7 system installed at the Department of Energy’s (DOE) Oak Ridge National Laboratory. It is equipped with NVIDIA K20x GPU accelerators. However, US researchers are keen on reclaiming the top of the ranking. The same laboratory plans to finish the new supercomputer called Summit until the end of 2018. Its theoretical maximum performance is around 200 petaflops.
Distributed supercomputers
Supercomputers ideate the image of massive, overheating machines that occupies the whole warehouse and consumes as much energy as a small town. That is the truth when we speak about machines that only governments and global corporations can afford. But with the block chain technology, massively distributed cloud calculations becomes more and more popular as a way to make ultrahigh-performance computing pervasive and affordable for everyone.
SETI@home was one of the first successful examples of massively distributed supercomputing. It is a crowd sourced science project launched in 1999 to find signals from an alien intelligence. The system sends participants millions of chunks of data to be analyzed by their home computers. And thereafter, the software processes the data on the participants’ computers and sends out the analysis results.
The computers of SETI@home volunteers are parts of an informal, ad-hoc supercomputer grid. The block chain technology allows you make money from this by renting your computer’s power to those who need to perform massive calculations on one computer. A few exciting projects were launched recently to embody this idea. For example, Golem is going to be an “Airbnb for computers” with an Ethereum-based transaction system.
Another example of distributed supercomputing idea is the Altumea — a new decentralized platform connecting GPU owners with those in need of complex performance calculations. This type of technology allows the execution of complex computations which can process massive amount of data. The project will attract individual owners, willing to monetize their hardware and contribute to some cool research as a side-bonus.
The potential of distributed ledger technology is to introduce a new era of massively distributed computing. The next-generation platforms will enable participants to lend and borrow computing resources — and make money in the process. Moreover, distributed computing can genuinely fuel the rapid progress in areas such as the Internet of things, AI and deep learning. That means more and more exciting innovations are on the way. Thanks to shared effort.
