diary of a mad
househusband

20 april 2000

 

 too narrow, too linear

The End of Moore's Law? wonders Charles C. Mann in the current Technology Review.

The current economic boom is likely due to increases in computing speed and decreases in price. Now there are some good reasons to think that the party may be ending.

I don't think so, for reasons I'm about to state, and neither do most of the people whose thought-provoking responses I urge you to read.

Even if the power of a single CPU is approaching a limit, the number of usable CPU cycles will continue to grow exponentially. Most existing computers are idle much of the time, due to limitations in hardware, software, and imagination. Those limitations are being overcome.

Revolutions in RAM and ROM will soon ease the hardware bottleneck.

The inexpensive twin-CPU computer I'm using to write this is but the harbinger of cheap systems capable of symmetric multiprocessing, which will stimulate the development of software capable of utilizing SMP much more effectively.

Beowulf clusters, SETI@home, and RC5 (and other distributed computing projects) are also beginning to affect software development-- and to awaken corporations and governments to the untapped potential of an army of desktops.

And there's no way to predict the results of software like Gnutella.

DNA computing opens the door to the solution of problems beyond the practical capability of silicon. (It wasn't until after I had written this that I discovered Tech Review's own DNA computing article.)

Moore's law is really too narrow and too linear to have much meaning except to economists, day traders, and others thinking within very short periods of time. When one thinks in historical terms, it appears to me that the curve graphing the ability of humans to Do Things may be hyperbolic. For most of history, the curve was nearly parallel to time's abscissa, with the speed of data interchange capped at the pace of a walking man or horse, or a wind-driven boat. Less than two hundred years ago, steam put us into the knee of the curve. Are we still in the knee, or are we now in the portion asymptotic to the ordinate? Sometimes, it seems to me it must be the latter. Just since my birth, the speed at which we can move our favorite class of data packet, a physical human presence, has increased from subsonic to supersonic to escape velocity.

previous | next 

 

contents
current
about
exit

exit
about
current
contents