Whether you call them HPCs or supercomputers, they are FAST! All computers are based on large numbers of tiny switches called transistors, each flicking between on and off and storing and manipulating numbers far faster than human beings can. The rate at which the switching happens is dictated by clock speed – the gigahertz or GHz you hear about on adverts and in computer shops. But mainly that’s just a marketing trick. Not all processors work the same way, and some of the newest processors in laptops, desktops and mobile devices like tablet computers don’t have clock speeds as high as those from a few years ago. That’s because they work smarter – not harder. The best measurement has always been the number of instructions or operations a chip can carry out in a given time. For home machines, this is usually measured in Millions of Instructions Per Second (MIPS), but for scientific computation the measure is Millions of Floating Point Operations Per Second (Megaflops, MFLOPS).
The very first commercially available microprocessor, the Intel 4004, could carry out just under 0.1MIPS. A reasonably priced laptop PC bought in 2012 could rate at over 100,000 MIPS – more than a million times faster. Rating the modern chip for scientific computation comes up with a figure of 50,000 MFLOPS – not bad! But one of the fastest supercomputers around today – Japan’s K Computer – can do over 10,000,000,000 MFLOPS. That’s the equivalent of 200,000 laptops’ power in a single computer.