But for people with anger management issues, scientists at UCL have recently built a computer that will never crash or freeze. This ‘systemic’ computer, modelled on natural phenomena such as a swarm of bees or the neurons in the human brain, is able to repair itself when there is a glitch.
There are many reasons why conventional computers crash or freeze but the central issue is that they work sequentially, that is, carrying out one instruction at a time. If the processor stumbles on one instruction then the whole system comes to a standstill – when a piece of software freezes, it gets trapped into one computation that, if left, will go on forever. The systemic computer divides the various instructions into systems that interact with each other randomly and carry out several tasks simultaneously.
It is just like the brain with lots of different neurons all chatting away to each other
Humans Invent spoke to Dr. Peter Bentley, a computer scientist who has built the systemic computer at UCL, to find out how it works and what impact this breakthrough could have on traditional computing.
Bentley says, “It goes against traditional computation as nothing is centralized, everything is now distributed. It is just like the brain with lots of different neurons all chatting away to each other on random occasions.”
A traditional computer uses a programme counter which makes the processor go through each instruction individually. In fact, the belief that computers are multitasking when you have more than one programme running is a fallacy.
Bentley says, “You might have 5 programmes running on your computer at once but actually the computer is switching its attention between each of those five in a split second. So that’s why, if one of them locks up or freezes, your whole computer can crash because it’s not actually doing things in parallel.”
Even with the advent of multi-core processers a lot of the processing remains sequential, as Bentley explains, “The problem is that they all access the same bit of memory and the memory is still centralized. If something goes wrong with the memory you are still screwed, the computers going to go down.”
While it is frustrating when your PC or laptop crashes , this type of parallel computing could be most beneficial in situations where a computer protects us from danger.
It’s got to work hand in hand with traditional computers
Bentley explains,“In an aircraft the normal approach to making it robust against damage is to have an awful lot of on-board computers. They might have seven duplicates up there in case one of them goes wrong. But it seems a little bit silly to do that. You don’t carry seven brains in your head, it’s just that redundancy happens on a much lower level.”
Interestingly, though this type of computing is much more advantageous than traditional fare, Bentley doesn’t believe it will entirely replace traditional processing.
Bentley says, “I think it’s got to work hand in hand with traditional computers. What we’ve already had to do in the systemic computer is include a conventional processor so anything that is sequential we pass off to the conventional computer as it is quicker at it and we leave all the parallel, stochastic fun stuff for the systemic computer to do.
He continues, “The other reason we kind of need to have both is because the whole world uses conventional Von Neumann architecture and the only way our systemic computer can talk to any other computer is by having a little friend who can send messages through and that friend is the conventional code processor.”
So it seems a calmer world awaits us. Who knows, I might even miss screaming expletives at my belligerent old codger of a computer. Actually, no I wont.