Columns

So what are we going to do with all of this multicore floppiness? Reinders told me that one of the ideas he's most excited about is the day when you can enter the details of a virus and simply ask the computer to come up with the cure based on simulating its effect on the body. It's a big ask, too: given the best that current

ADVERTISEMENT

computing power can offer is a simulation of one or two simple cells, even zettaflops may not be enough and we'll enter yottaflop territory (that's a trillion trillion operations per second).

There's certainly no foreseeable point at which computers will be "powerful enough". To round off our conversation, I asked Reinders how much memory and power would be needed to build and run a complete molecular model of the human body. He just chortled in a reassuringly far-off sort of way. We did come to the conclusion, though, that it isn't beyond the bounds of possibility. The reasoning is simple: we're living proof that all the information to make a human can be stored in a volume roughly five or six feet high and a foot or two across. If the physical universe can simulate a human in the space of a human, there's every chance a computer could, too. But it would have to be a very, very floppy computer indeed.