Do you know what’s better when it’s more floppy? Supercomputers. Hi everyone, in today’s article we are going to discuss is supercomputers can predict future?.
China has just announced it has created the most powerful supercomputer yet. The machine, named Sunway TaihuLight, is theoretically capable of almost 125 petaflops, or 125 million billion flops. I don’t mean 125 million billion football players, I mean FLOPS, a metric for computer speed. It’s an acronym that stands of Floating-point Operations Per Second, which translates to how many calculations they can do with gigantic or infinitesimally small numbers in one second. Because supercomputers can do a hundred million billion calculations every second, and because the range of numbers it can use is so incredibly large, supercomputers are perfect for running models that deal with things of all sizes, from how galaxies form to how atoms interact. In other words, they’re ideal for science. Every time we mention on PCGUIDE that an experiment used a computer model, they did it on a supercomputer. These computers can be privately owned, as is the case with some pharmaceutical companies that use them to model how a drug will behave on a molecular level. Supercomputers can be ludicrously expensive though. Sunway TaihuLight cost $270 million to research, develop, and build. And that’s before the costs of operation are factored in. Imagine how much electricity it takes to run a computer as big as your house. And how much it would cost to keep it cool so said the house didn’t melt itself. For that reason more often they’re paid for by a government and then that government decides what experiments are awarded access. Sometimes supercomputers are built with just one purpose in mind, like IBM’s Deep Blue which was made to beat one guy at chess in 1997. Two years later they made Blue Gene study how proteins folded and to make a lame pun. Usually, though supercomputers can be general-purpose. They’ve been used to test nuclear weapons, not just the explosion themselves, but everything that could affect how they work from when they’re pulled out of storage to when they go off. Even things like how much a transport truck rattles could affect a nuclear missile’s guidance system, so it’s factored into the calculations. And all this can be done without actually setting off a nuke, so they’re making our deadly weapons safer… I guess?
Supercomputers are invaluable to astronomy and cosmology, modeling the first few moments after the big bang on an atomic level, or what happens during a supernova. They’ve been instrumental in simulating how a protein can fold itself because they can track how every single atom would interact with those around it. This can also be used to figure out why proteins sometimes go wrong and cause diseases. And for our day to day lives, they have been most noticeably useful for predicting the weather, taking in information from hundreds of thousands of weather stations around the world and comparing it to the data from yesterday to predict what will happen tomorrow. Extending that much farther and using averages instead of specifics, supercomputers can be used to simulate climate change. They’re the only beasts powerful enough to take in the many many variables that go into predicting the climate far into the future. And there’s still room for improvement. When they crack the next milestone in speed, the exaflop, or a billion flops, scientists think they’ll be able to make a fully coupled earth system model that is accurate down to a kilometer. Every variable a scientist could want to include, of the entire earth, down to a kilometer resolution, all calculated before I die of old age. That’s what I call a supermodel. Today’s supercomputers build on the foundation laid down by the earliest computer scientists. With that in mind, we salute Ada Lovelace, the OG computer nerd. If you had access to a supercomputer, what would you use it for?