Personal: , Recommended Reading: , Science & Technology:

Some trends that may affect the computer you can buy in 2012

For some perspective on the challenges of my job this page, which is part of a larger description of past trends and future evolution in microprocessor development, give a reasonably good description of the problems I am trying to overcome.

I recommend reading the entire article, which is part one of a series on "The Quest for More Processing Power".

What is interesting is that I didn't see really any note in the parts I skimmed regarding how in the next few years we will be undergoing a transition from where process technology (which is the way we actually make the microprocessors, using different processes to put films down and selectively remove them to make the transistors and the wires that connect them together) and the improvements in the process technology are paramount in the increase of capability of microprocessors to where the design of the microprocessor itself becomes the key to continued increase in processing power.

Most of the improvements in electronics (specifically processing and memory capacity) have been due to improvements in process technology. In my career, which started in 1991 when I was working on state of the art technology of the time, we have gone from the thinnest layer we make (which happens to be the heart of the transistor, and the part that I have consistently worked on) going from a thickness of roughly 9nm (a nanometer is 1E-9 meters, a meter is a wee bit over 3 feet long, and nano, or 1E-9 is 1/1,000,000,000, so a nanometer is 1/1,000,000,000 of 3 feet... atoms on average are about 0.5nm apart) to about 1.5nm. Atoms are roughly 0.5nm apart (not all are that distance apart, but it's close enough and it makes the math easier), so we have gone from the thinnest film being 18 atoms thick in 1991 for the absolute state-of-the-art, to about 3 (yes THREE) atoms thick for the current state-of-the-art microprocessors.

Think about it, 3 atoms. There's not much room left for this film, is there? Admittedly, there are some ways around it, notably changing the material used, but I've been working on that project for the last 5 years, and no one has the answer to which material to use yet.

Also, another key part of the transistor is the length of the gate. The gate is what is controls whether the transistor is "on" or "off", in other words, a "1" or a "0" in the binary signals used in MPUs (MicroProcessor Units). That gate length has gone down to where it is in the range of 100 atoms across.

Again, think about it, 100 atoms.

So, this change in primacy from process technology to design techniques for improvement in performance is not only expected but almost inevitable unless we come up with some breakthrough in either materials or fundamental structure of transistors.

I don't see this truly watershed transition really recognized widely.

As an FYI, the stuff I work on now will go into production around 2011 or 2012, which means I have a reasonable view of what will be available around 7 years out.

There are a lot of wide-ranging effects that will arise out of this transition. I may write on them later, but for now, to sleep, perchance to dream, but hopefully not and instead an all too brief visit with the bliss of an inactive brain.