Photo modified by Greg Yeric from original by Eric Fischer.
I was recently invited to add some thoughts on technology scaling over at Ed Sperling's excellent semiengineering site. Here is what I came up with:
What Goes Around Comes Around: Moore's Law at 10nm and Beyond
The gist of my article is captured in the (slightly altered) photo above: For an incredibly long time, chip designs have benefited from the amazing progress that is Moore's Law, and then by association Dennard scaling. It's been kind of like 49 years of being able to use an escalator, but now we find that convenience we've come to expect isn't working and now we have to use the stairs. And by "we" I mean designers in concert with semiconductor technologists, in an exercise often termed Design-Technology Co-Optimization. DTCO is one of the areas of our R&D group here at ARM, and I talk about some specific technology scaling issues in the article.
I was running a little tight on space for the article, and given the readership over at Semiengineering, I took the liberty to use a number of terms without defining them very well. Give it a read, and send me a question if something is not clear. As this is my first post on the ARM Connected Community, I'll be endeavoring from here on out to keep providing information along these lines, and about semiconductor technology trends in general, but here I'll take the time to explain things a bit better. Just like systems engineers have to slow down when they talk to me .
"cheers" (that's in the article)
You got it! Actually I was referring to the barely-legible plaque on the escalator. "Dennard & Strutt Escalators, Inc.". Dennard being the one who set the rules for transistor power and performance scaling, and John William Strutt, who became the 3rd Baron Rayleigh. Rayliegh set down the rules for the feature sizes you can print given a certain wavelength of light. Unfortunately, we stopped being able to use smaller wavelengths of light many moons ago. We've been stuck with 193nm wavelength light, and with some amazing ingenuity and raw computing power, we somehow print features that are almost 10 times smaller than that now! An interesting piece of trivia: Lord Rayleigh was awarded the 1904 Nobel Prize in Physics, not for his reasearch into light but for the discovery of the element Argon. The trivia being that the 193nm light we depend on for all modern computer chips uses an Argon Fluoride laser.