Databus Issue: 2010 3 07/31/2010
CUE View
David Thornburg, Ph.D. Director, Global Operations, Thornburg Center for Professional DevelopmentMoore's Law Dies Again
PDF
Most technology folks know about Moore's Law, named after its developer, Intel's Gordon Moore, over forty years ago. In the early days of integrated circuitry and computer chips, he observed that the complexity of these circuits was doubling every two years, usually measured by the number of transistors that could be crammed onto a silicon wafer. He also noted that, as complexity increased, cost went down and speed improved.
Over the intervening years, this “law” has been modified to the point where the “doubling time” has been reduced to one year. The present state of development is giving us processors with over 750 million transistors with a throughput equivalent of 12 gigahertz. Gollyorkins!
As chip complexity increased, naysayers pointed out that, sooner or later, Moore's Law was going to run out of steam. Transistor size can only be reduced so much, heat dissipation has its limits, and the architecture of microprocessors will finally have bumped into a stone wall. There is even a report that suggests the cost of fabrication facilities has grown so large that they have to be used for a long time just to recover the investment, and this will provide a death blow to Moore's Law. This pronouncement, due to take hold in 2014, was reported in a recent article in EE Times (www.eetimes.com/showArticle.jhtml?articleID=217900102&cid=NL_eet).
I have a different theory.
Advances based on Moore's Law are going to end because it no longer matters.
That's right. It simply doesn't matter how much faster processors are kids' hands when it comes to educational computing.
Processor speed and complexity have reached the point where virtually no ordinary person cares how fast a processor is. For those who do care, cluster computing is the way to go. Video games (which use the most computing power of any home or business application) have gotten to the point that you can strap eight Sony PS3s together and make one of the fastest supercomputers on the planet (www.ps3cluster.umassd.edu). This site shows you how to build your own supercomputer that outperforms the 200 node IBM Blue Gene computer.
What big iron (very large, even room-sized computers, servers, etc.) cares about today is reliability, and this means redundancy. Basically, if you have a parallel machine running at a few teraflops and a few processors croak, then all that happens is your throughput drops a bit until you swap in fresh boards.
As for home machines, there is a reason that netbooks are growing in popularity. For us mortals, we have entered the world where bandwidth matters more than processor speed, and with 3G networks, etc., we will see computer-intensive applications move to big machines on the web, leaving our laptops for simple programs. We forget that the Apple II and its ilk ran wonderful software with less than a one mHz clock. Five-gigabit processors make money for Redmond and the electric utilities. But, seriously folks, how FAST do you need to boot Solitaire?
Acer now has a netbook with an 11-inch screen and a mostly regular-sized keyboard. With a six-cell battery, you are talking six hours of actual use between charges. Costco has
the computer for about $350 (with a regular battery). When this machine gets Google's Android version of Linux in the Fall, the price will likely drop below $300. Plus it tips the scale at about one kilogram. No more sore shoulders.
I've been carrying a small netbook with me for some time, and it is doing most everything I need when I'm on the road. Yes, I'd like a larger screen (coming) and a standard-sized keyboard (coming), but these do not impact the overall utility of the system; and, after a little while, my fingers adjust pretty well. Aside from the cost (I paid less than $300), the advantage is overall size and reduced weight. And (I almost forgot) battery life. These machines are pretty green, which is important when you realize that our global computers use about 20 percent of all electricity generated on the planet! See Kevin Kelly talk about this in his video about the 5000 days of the web on www.ted.com.
For kids of all ages, netbooks can be pretty cool devices. Built-in wi-fi, decent storage (I have 160 GB of hard drive space), and the ability to run good applications all favor these computers. And, let's not forget that we are living in stressful times when it comes to budgets. Look in your kid's backpack. (Bend your legs when lifting the backpack so you don't strain yourself). Each of those obsolete textbooks costs your school system $75 or more. Four books equals a netbook at full retail. And, you've traded several kilos of passive paper for a kilogram of dynamic computing power with which kids not only can read what others have done, but also can create and build their own understandings of the things they will need to know about in their future.
For several generations, we have celebrated Moore's Law without realizing that we have also been enslaved by it. Our quest for faster, better computing has cost us a fortune. And, the day we installed our new computers, even faster machines were introduced to the market. Now we have finally said, “Enough!” If there is a silver lining to the economic crisis in education, it may be that it will finally get us to provide every child with the kind of powerful technology some of us have been predicting for many years.
How cool is that?
---------------------------------------------------------
David D. Thornburg, Ph.D., Director, Global Operations, Thornburg Center for Professional Development (
