June 2000 Archives

Hardware and software are usually very different things. Sometimes you can hardwire the maths of software in silicon, and the results could be hundreds of times faster. Computers are usually given graphics hardware accelerator cards to make the displays more complex and faster than software alone could make them. You can also emulate hardware in software, but it works very much more slowly than the hardware version. For example you can get a macintosh to pretend its an IBM-compatible PC, but its will run windows very slowly. Garage software start-up companies that grow into international successes have long been real-life legends in the computing scene. The reason this has been possible is that developing and reproducing software doesn't require the expense of hundreds of components per prototype, or entire factories to mass produce your end product. All you need to do is to copy the product onto a cheap disk, or make it available to download over the internet. Computer rebels can make a social and political statement by refusing the Microsoft or Macintosh operating system and software for their computers, and selecting instead the free GNU Linux operating system, that you can legally copy from a friend or download off the internet for free. The giant megacorporation owned human resources have been replaced by thousands of skilled and dedicated volunteers all over the world, who have produced an operating system that doesn't crash every few days like Windows, and doesn't need to be rebooted more than once a year. However the Central processing unit, or CPU, still takes a giant multinational corporation's resources to physically prototype and to mass produce in factories. Its just hasn't been possible for the computer user with a social conscience to beat the corporations at their own game, with hardware, its simply too expensive. All of this has been changed with the invention of a new reconfigurable hardware called a Field Programmable Gate Array - FPGA's. These new chips are the hacker's dream that has spawned the Freedom CPU project that is leading to cheap upgradable alternatives to brand-name CPUs. If you download the design for say, a Pentium 3 to this chip, it morphs its insides to match the circuits of a Pentium 3. If you change your mind and want the chip to work as a mobile phone, you can load the circuit design of a mobile phone to the chip, and it will softwire its insides to work as a mobile phone. If you have invented a new electronic gadget, lets call it a tricorder, in your garage, you can download the design to the chip, and it works as your tricorder. If you're happy with your design, and want to start producing them for sale, you just copy your new design over to more FPGA chips, and you've manufactured more tricorders. Alternatively, you may make your circuit design available for download on the internet and sell your tricorders all over the world, without the need to physically send a single unit out of your garage. Just as you might upgrade your software with the latest improvements over the internet, you will be able to upgrade your hardware gadgets over the internet as well. If you don't need a particular gadget anymore, you could replace its softwiring with something completely different and more immediately useful. Clever programs have been written so that circuit diagrams that you might find in electronics magazines, textbooks and on the internet can be read with "optical symbol recognition software" that can read electronic circuits in the same way that optical character recognition programs can read printed text. Instead of getting an electronics magazine and then making a trip to the electronics shop to buy wire and transistors, capacitors and chips, you can just scan the project into your computer, and then tell your FPGA chip to softwire itself into the project. In the future, you may have libraries full of gadget designs you can access, and the have a pull-down menu on your screen to select what gadget you want your hardware to be today. Now this is all fine and dandy for the electronics experts, but what about everyone else? Adrian Thompson in the UK has looked to biology to harness the creativity of evolution to design new circuits. Using "Genetic algorithms" they bred programs to find a working circuit in the same way you might cross-breed apple trees to get a certain size and colour of fruit, by selecting those that are closest to what you want, having them breed, and then selecting again from the offspring those that are just a little closer to what you want, and repeating the cycle, over and over. In nature mutations creep in that sometimes add new characteristics, so genetic algorithms include mutations over the generations, also. Dr Thompson ended up getting his computer to design a working circuit, over thousands of generations of breeding and mutating programs, that could turn a green light on when he said "go" and turn it off when he said "stop". This circuit uses fewer than one-tenth the number of components that a human designer would have used, and he has no idea, as yet, how it works. There's simply no human designer to ask. In fact when he tried to copy just those parts of the circuit that were wired together over to a fresh chip, the design stopped working. It turned out that some components merely NEAR the wired up circuit were necessary to make it work, possibly by some sort of field effect that the fittest programs took advantage of in their designs. All of this means that in the future, you could own some sort of substrate for a universal gadget, complete with a menu of hardware that it could morph into. If you wanted it to do something new that wasn't on your menu, then you could select computer mode, run a genetic algorithm circuit designer program, and describe to the program what you would like it to design, perhaps a timer to record your favourite radio program. It would quietly churn through generations of programs better and better suited to the task, until it found a design that did exactly what you asked. At that point you could choose to donate the design to a public library of gadgets on the internet, and share with your friends or set up shop and sell your idea on the web. We're approaching a time when if you can think of a gadget thats physically possible with current technology - then you can have it.

Recent Comments

  • Adam A. Ford: After a summer of speaking to potential presenters and collaborating read more
  • Ian: I'll check the data and get back to you. From read more
  • Matthew Hall: Energy density? Number of charge cycles? Safety issues? Sounds like read more
  • Raz: Wow, it's Sterling's "German nanotech" (from Heavy Weather I think). read more
  • Matthew Hall: This looks like a great find Ian, I'll be viewing read more
Notify me



January 2013

Sun Mon Tue Wed Thu Fri Sat
    1 2 3 4 5
6 7 8 9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29 30 31    

Archives