View previous topic | View next topic

A Misconception on Moore's Law (Series M Ep. 16)

Page 1 of 1

1199222.  Tue Jul 19, 2016 3:28 pm Reply with quote

I adore QI, and as both a programmer and a hardware fanatic it pained me to hear the common misconception on Moore's Law getting repeated as the gospel truth.

Moore's Law did not, and has not, ever been associated with the doubling of computing power. Gordon Moore's full quote, as it pertains to semiconductors, is reproduced here:

The complexity for minimum component costs has increased at a rate of roughly a factor of two per year. Certainly over the short term this rate can be expected to continue, if not to increase. Over the longer term, the rate of increase is a bit more uncertain, although there is no reason to believe it will not remain nearly constant for at least 10 years.

Moore's observation pertained only to an observed increase in transistor density in the same surface area for the same cost. For reasons which will not be enumerated here, as to do so would be both voluminous and of tremendous breadth, the number of transistors in a given integrated circuit package does not have and has never had a direct relation to the overall computing power produced by said integrated circuit.

The idea that this doubling of transistor density correlates to an increase in computing power was put forth by David House, one of Moore's colleagues at Intel, and the period was specified by him as 18 months, not two years. Moore himself denies that his original quote was related to computing performance.

Furthermore, in the past two to four years, the semiconductor industry has begun to reach the limits of Moore's Law. Intel, for example, has been forced to push back several "die shrinks", or process improvements that result in a physically smaller transistor on the silicon die, due to myriad reasons. It should be obvious that an object's volume cannot be halved infinitely without eventually becoming as small or smaller than an individual atom, and semiconductor manufacturers are already beginning to reach the point where electromigration and quantum effects pose a real problem around which they must work, and it will only get more difficult from here. Thus, the QI Elves repeating this classic misconception - on an episode about misconceptions, no less - is even more heinous due to how recently the episode aired with respect to the past half-decade of Moore's Law being increasingly invalidated.

1199676.  Sat Jul 23, 2016 1:06 pm Reply with quote

Excellent quibble! Thank you TheMogMiner and welcome to the forums :-)

Mr Red
1215666.  Fri Dec 09, 2016 8:24 am Reply with quote

When I was an Electronic Engineer purchasing Fairchild (remember them?) and Intel products Mr Moore was quoted as stating the time constant (for your doubling) was 18 months. He may have revised that in the light of more data and more money for research etc. Certainly everyone else has revised it! Re-purposed it!
And on current showing it is good enough to warn tech companies not to stand still.


Page 1 of 1

All times are GMT - 5 Hours

Display posts from previous:   

Search Search Forums

Powered by phpBB © 2001, 2002 phpBB Group