Intel Explains 45-nm Chip Fabrication Technology

Understanding chip-fabrication technology is like getting the facts on that other thing you became curious about when you were 12. Most people don't have anyone who really sits them down and explains it; they just kind of pick it up. Well, I can help you with the chip-making stuff. Listen to my podcast with Intel Senior Fellow Mark Bohr, who has been leading the development of the industry's first 45-nm processors, which will appear later this year when the quad-core Penryn ships.Think of it as a 15-minute course that'll get you up to speed on the critical issues of modern chip manufacturing. Mark is unusually articulate; I ask a few questions, but mostly stay out of the way.

Listen to the podcast by clicking here.

(Alternatively, you can right-click and save the file to your desktop. To do this in Internet Explorer, after you right click, move your mouse over "Save Target As...," click, and save to your desktop. In Firefox, after you right click, select "Save Link As..." After the file download is complete, open it and listen to it in Windows Media Player.)

In the podcast, Bohr makes the point that 45 nm allows Intel to pack more transistors onto a given slice of silicon than was possible at 65 nm, and to run the resulting chips at lower power. Here are some highlights of the chat:

"As you make transistors smaller, they do get faster, and you can use that in one of two ways. You can design single-cores that use ever-more transistors and burn more power, or you can use those extra transistors instead to make two smaller, simpler cores that each use less power. So for many applications, having dual core is a more power-efficient way to realize improved computing performance." Intel is the first to go into high volume manufacturing on 45-nm. Intel's been pretty consistent in delivering new technologies on a two-year cadence. We first introduced our 65-nm products in 2005. Here in 2007, we're first introducing our 45-nm products. And we're on track with our next-generation 32-nm process technology, to have that process ready in the second half of 2009. "We are running into limits that some people view as fundamental, but we are also inventing solutions around them." "As we scale transistors and make them smaller, they get faster, but they tend to get leakier." (Bohr goes on to explain what this means.) "If you have hundreds of millions of transistors on your chip, and each of them is leaking a tiny amount of current, it does add up. They way it may up in an actual product [is], maybe your laptop computer would have a shorter battery life because it's leaking power." P.S. The IEEE Spectrum article referred to in the podcast is available here. The article discusses the challenges facing the developers of extreme ultraviolet (EUV) photolithography, an advanced chip-making technique originally targeted for use beginning in 2011. EUV is supposed to operate at 13.5 nm, which is beyond ultraviolet and obviously much smaller than the 45 nm we're discussing in the podcast. Also, see the Quad-Core Processor Buyer's Guide 2007, which includes a chart showing the current pricing of available quad-core chips.