Distributed computing

Distributed computing (also Decentralized Computing, Distributed DV, english distributed computing ) is a technique of application programming, in which the individual processes of a distributed application calculate a common result.

Idea

The background is the idea that the main processors of many computers are temporarily underutilized because the user usually works with only a few programs that require only a portion of the total CPU power. These unused resources you would like to make it usable for distributed computing. A corresponding client software is installed on the affected computer, which usually largely performs these tasks in the background.

Principle

Distributed computing has to be organized. For this purpose, a software that is made ​​available, which must be running on the clients to solve their specific task. Furthermore, the tasks that are processed must be precisely machined, or have yet to be distributed, to be managed.

If one wants to now engage in problem solving, that is, set the unused computing power of a computer available, it first loads the client software on the computer, install it and test the installation by means of predetermined test data. After that logs on to the website and can assign the data to be processed. After calculating the data packet, which can take up to several weeks of computing time to complete a few hours, the result is reported back to the website, and you can let yourself enter new data.

Areas of application

Distributed computing is used in many areas of research, especially at very compute-intensive applications (such as docking simulations for the design of future drugs, the calculation of protein folding processes, the search for primes or refute mathematical conjectures ), for their processing the performance of conventional supercomputers are not sufficient or are insufficient funds available for the. Distributed computing projects can be therefore very common in universities, foundations or small, or medium-sized firms completed projects.

Concrete projects

One of the first projects, which used the technique of distributed computing, was the SETI @ home project of the University of California, Berkeley, the thus achieved the computing power of an expensive supercomputer.

Many projects followed, such as

• Biophysical projects as Predictor @ home, Rosetta @ home and POEM @ home for the simulation of protein folding
• Projects on the search for gravitational waves ( Einstein @ home)
• Find-a - Drug for the treatment of various diseases (now finished)
• Modeling of climate change in the 21st century with ClimatePrediction.net
• Projects that have the solution of mathematical problems is prescribed ( for GIMPS Mersenne primes, Fermat primes for GFPS )

Today there are almost in all scientific fields, distributed computing projects, even the industry already uses the technology.

A new level of proliferation reached the project Folding @ home, which uses the capabilities of the gaming console PlayStation 3 as well as from ATI and Nvidia graphics processors (GPU) and as an output of up to 4 quadrillion calculations per second (4 petaflops ) achieved. With up to 700,000 registered PS3 - participants and about 50,000 regularly active consoles this project stands as a most powerful distributed computing network of all time in the Guinness Book of World Records.