Topic 111

Processing power needed to create complex models by Raymon


Models are physical or graphical simulations and representations or what will happen in a certain situation for things that cannot be seen or observed by the naked eye easily, or cannot be really observed in real life at all. They also show scientists what will happen in a certain situation in theory, and are used to understand the deeper roots of physics and quantum mumbo-jumbo that layman will never understand in his lifetime.

In school, models are often little plastic or paper-and –ink pieces of work that show students how an atom is supposed to look, how the solar system is laid out, or show them what DNA looks like. In the laboratory, however, the hardcore research that must be represented in models is simply far too complex to show in a static, plastic model, or requires movement to show what is going on. This is where the powerful computers called supercomputers come in. In older times, supercomputers were about as powerful are our TI calculators we use today and did about the same stuff. They could do some simple simulation showing the speed of a molecule or what will happen when a ball bounces around a room. These simple models can be seen in games today, with the physics engines employed to make a grenade bounce around a room and blow somebody's leg off in an entirely convincing way. So, if this then-complex simulation can be seen in everyday life today, what do the calculations done by supercomputers look like today?

Blue Gene/L is currently the fastest supercomputer in the world. The fastest speed it has reached is 207.3 Teraflops – 207,300,000,000,000 floating point calculations per second; a floating-point calculation is something like π × 5.38194381. As you can see, the rate at which these machines can process information is mind-boggling. Blue Gene/L has a total of 131,084 processors so do this immense work. This immense power has allowed mathematicians and scientists to do calculations previously thought impossible, unable to be understood by the human mind, and completely useless to ordinary people; just recently, researchers completed modeling E8, an immense mathematical equation that, if put on paper, would apparently cover the city of Manhattan.

Perhaps, the power of these modern machines would be enough to solve any questions humans have. But if you thought that, you would be wrong, as a science magazine (Spectrum) was recently lamenting the inability to simulate the convergence of a galaxy and see how galaxies come together in such a fashion and spin in a certain direction. To even model a nebula coming together to form a new star, one would need so much processing power that one can wonder if it can really ever be done. A nebula is basically just a huge cloud of hydrogen – one that can contain, in theory, 1027 molecules of hydrogen. Each molecule of hydrogen has a gravitational pull to all 1027 - 1 other molecules in the nebula. Therefore, the machine would have to do 1054 calculations, something that would take even blue gene 5.583 × 1034 days.