BOINC OR Berkeley Open Infrastructure for Network Computing is a very, very strange phenomenon. It is the most famous name of a phenomenon called DC (Distributed Computing), or distributed calculations. His name Berkeley's open infrastructure for network calculations carries inside information about the place of his birth, about the University of Berkeley. And Berkeley is known to all of us. It is one of the world's leading centres for physics in all kinds of sectors, from quantum mechanics to astrophysics.
Today's scientific research uses increasingly modern technologies to monitor a phenomenon, bringing about an increasing number of data obtained for further analysis. And this quantity is now increasing almost exponentially. And here we have two possibilities of technology to process them. Let's go see the first one.
We have already outlined a huge amount of data waiting for further processing using various algorithms. Now let's imagine hardware requirements for a computer that would be designed only for their processing. The basic requirement is the number power. Today, early 2008 we know that one of the most successfully used Intel's new processors, namely among us calculators very popular Intel Core 2 Duo Quad @ 2,40GHz has a basic number output of about 5,500,000 operations per second. More clearly, the 5.5 Gflops power. However, the current demand for numerical output from the average project is ten to a hundred thousand times higher. So we can buy a huge hall, and we can put 10,000 computers fitted with this processor for example. And we're going to plug these computers into the network. I think a 20-story hall the size of a league football field could do just fine for a start.
We now have a structure that represents a kind of "ideal computer." We can be satisfied, we built it with a millions of dollars of investment. But we must not forget another cute feature of every computer. It needs a steady supply of electricity to operate. And since we now know that the average average consumption of this processor's average computer is about 180 W, and we have 10,000 of them in the lobby, we need a power supply of 1,800 kWh. Okay, we're gonna build our own power plant for a few more million. Here we go. Another mistake, folks. The 10,000 computers have another negative feature. Given that some of the electricity is turned into heat in their operation, and the loss heat will only take 50 W per one of them, we have an extra 500kW of heat output. And that would flood a smaller town. As we know, heat and electronic components are brought to a certain ratio. If we do not want to bake our investment for millions, we must effectively cool down = additional costs. And we haven't gotten to the lighting, the service, the maintenance, and the doorman yet. If you feel I have exhausted all the demands now, you are mistaken. The trend of data development for analysis is increasing exponentially, as we have already said. In order to make the project work in a year through this megacomb, it will probably be necessary in one year:
Upgrade hardware to the latest level, raise the hall by another ten floors and double the number of computers connected. And that's more and more investment. And no bank in the world will finance this, even if it has 100% certainty that at the end of this effort you will find answers to all the mysteries, discover immortality and a means to treat all the diseases.
Let us now look together at one interesting paradox, which is the possibility of processing the mentioned data with minimum costs and maximum flexibility.
Statistics predict that each of you nice readers who just arrived on these lines has some computer available. The size, performance and type, operating system and ownership form do not matter much. It depends more on what you use it for and when. Why? I will explain the model situation that can be used just as much for the computer at home or in your office.
You come to your PC and turn it on. It depends on his type and equipment, his operating system and your skills with passwords in Windows before he "fully booms up." You start an app, a gamesque or Excel, and you start to engage in interactive entertainment (or suffering, depending on the type of application). But what's going on inside the computer? In a processor? One minute, he's processing a set of instructions. Then nothing. And he's working again. Nothing. It's nothing seemingly subtle, but it is. And here's the paradox. You have a computer. Running computer. A computer consuming electrical power. But sometimes it's not doing anything. And these moments can be used to process the vast amount of data that is waiting for it. How?
We're going to connect all these computers into one giant structure. After a few small pieces, we will compile a respectable computing power only from the time of your lazy processors. And with our machines of all types and nationalities, we will be able to outdo the above megacomb many times. Well, admit it, isn't that genius? It is, isn't it?
Now you ask how to achieve that. Simple. Let's get some light on the principle of BOINC.
The BOINC platform is built on the idea of diluting huge data files into much smaller parts that are actually able to process a common computer in the order of several hours to days. In order to exclude force majeure from this system, it became the basis of a file called Boinc Manager, including Boinc Core, or the core. This software is freely disseminated, and after installing it into your computer after a short setup, with your participation, it is automatically communicate with your selected project so that it does not bother you and, at the same time, omitting the need for manual control eliminates the problem between the keyboard and the monitor. But we'll go back to the initial setup itself in the installation manual, and now, let's skip it.
The Boinc manager runs as a service, and he communicates with your chosen project for you. After the project is logged in, the application will be downloaded, especially the set of algorithms and data units, so-called WU or workunits. Boinc manager monitors the utilization of your processor. When they find out the "doing nothing" state, they plug one of the above units into the processor and they're fucking back to work. But once the manager finds out the request of your system or application for the processor's devices, the calculation disconnects and waits. And so it runs until the unit is ready, sent and replaced by another if you have ruled this option out in the original setting.
And you've certainly already thought about what it's gonna be worth if you help your computer save the above-mentioned costs. Nothing, my friends, nothing. I mean, almost nothing. For each unit counted and dispatched, you will receive a corresponding credit in your account. Not financial, not real account. Only virtual credit, whose function we will mention in chapter two. Your task as administrator and at the same time right is to decide which project and to what extent you will join. And your reward is knowing that you help in many scientific fields to find the answers and solutions that projects deal with.
If you are interested in the previous lines and want to learn more about individual scientific projects and their focus, you will be asked for further information Here.