Moore's Law. The end is near says Gordon Moore.
There actually is no Moore's Law in the strict sense of the word. It
doesn't attempt to describe an immutable fact about the progression of
technology. Nevertheless, it has achieved a level of fame tantamount to
invincible prophecy by axiomatically predicting that the number of circuits on a
silicon chip would double every 12 - 18 months. As the number of circuits
increases so does the functionality of the chip while the cost remains
approximately constant. And that is what has been happening for the last
40+ years.
It all started back in 1965 during the transition from single transistors on a
chip to whole circuits and then multiple circuits. Gordon Moore, then
working for Fairchild Semiconductors believed that a major improvement in the
manufacturing processes would take place about every 18 months and this would be
accompanied by the ability to place more devices on the chip. On a more
frequent level other design improvements would permit additional incremental
improvements yielding a 12 month generational cycle. He was so convinced
that the future lay on this path that a few years later he and Robert Noyce
joined forces to create a company dedicated to this concept.
It is the Intel we know today.
We can follow the trail of the improvements by looking at the microprocessor
generations that have been so critical for the PC market. First the
4004 in 1071 with 2,250 transistors for use in hand calculators, then the 8080
in 1975 which became the engine for the first PC with 4,500 circuits to the 286,
386, 486, Pentium and now the Duo 2 and latest Itanium. The Duo 2 with
about 700,000,000 and the Itanium with 1.9 billion transistors. Each
generation was achieved by shrinking the size of the transistor and
interconnecting lines. Intel has just started shipping the latest
technology level known as the 45 nm level. That is the resolution of the
features on the chip are 45 billionths of a meter wide. Already in the lab
work is going on at the 32 nm level with research under way at the 18 nm level.
So doesn't it keep on going? The answer is no and that is what prompted
Gordon Moore to declare that the end of this particular technology journey is
coming to an end. Already the technologists are dealing with thicknesses
of just a few atoms and the material is becoming very difficult to
control.
So where will we go from here? We are probably good for another 10 to 15
years before another technology will be required to continue the parade.
That technology will most certainly be quite different from what we are using
today. That doesn't mean everything comes to a screeching halt. The
factories will be producing the chips we have today and derivatives thereof for
decades. It means that we will need other solutions in order to make
faster and more powerful computers. We have, in fact, already started down
this path with the dual and quad processors on a chip that are in the latest
machines. Our hand was forced because simply trying to run the circuits
faster as had been the case for so long was just not working any more.
Faster meant more heat and the heat was growing faster than our ability to
remove it. This is deadly because heat is the enemy of transistor devices.
Some believe that the Nano technology now in the research laboratories will
provide the answer. We shall see...