The New York Times has an article from May 4, 2016, proclaiming that Moore’s Law may have run out of steam:
http://www.nytimes.com/2016/05/05/technology/moores-law-running-out-of-room-tech-looks-for-a-successor.html
But then they take it back by saying: “Predicting the end of Moore’s Law has for years been a parlor game in Silicon Valley, and not everyone in the industry believes that what it has come to represent is nearly over.”
Gordon Moore published a paper in 1965 stating that the transistor density on semiconductor chips would continue to grow. https://www.cs.utexas.edu/~fussell/courses/cs352h/papers/moore.pdf
His observation was that the cost per transistor goes down as the density increases, but the manufacturing yield of the chips also decreases. So, there is an optimum point. Moore said that density would increase towards a lower cost per transistor as manufacturing processes improved over time.
The technology Moore was describing became obsolete years ago. But, the idea that there was a fairly short doubling time for computer speed took hold as a business imperative for the electronics industry. “Moore’s Law”, that the capacity for technical computing would double every 18 months, has continued for over 50 years, albeit enabled by various technologies. The results are astounding.
In 1970 I was an undergraduate sophomore. I got my first computer user account with one hour of CPU time on an IBM 360, the supercomputer of its day. Today, that hour of computing can be done in 2 microseconds!
Over the years since Moore published his paper, computing speed has increased by a factor of over 17 billion. That is the number of seconds in 450 years. In other words, we can now compute in one second what would have taken the Apollo space program engineers nearly a half-millenium!
The cost reductions are similarly incredible. In the early 1980s the US national labs were bidding against each other to get the latest Cray-1 supercomputer, often paying $40 million or more for a machine that weighed thousands of pounds and required a huge refrigeration system to cool it. Today, a Raspberry-Pi is the size of a credit card and costs $39. It can run with only a heatsink (no fan) to cool it and has about 10 times the capability of a Cray-1.
In essence we live in an era of infinitely fast and vanishingly cheap computing capability.
So, what of the future for Moore’s Law? I believe computing capability will continue to increase at the established rate. It’s not really a technology problem: It is a business imperative for the processor makers. Computers will become faster maybe not because of the speed of the CPU, but because of factors like SSD storage, increasing integration (motherboard on a chip), and decreasing costs. CPU not fast enough? Well, here’s a board with 1,024 of them. Network not fast enough? Well, just put the microprocessor where the data is, and only put the results on the network. (Today’s automobiles contain about 50 microprocessors, including about 5 in the driver’s door alone.)
I can remember 10 years ago, when laser pointers showed up in the local dollar store. Heck! Forget the laser! You can’t buy the batteries for a dollar! In less than 10 years, your $40 million Cray-1 will be a commodity capability costing less than a dollar. It is a simply astonishing story.
Keith
