Justin Ratner brings out the babes

That said, there was a lot of good stuff. By far the biggest thing Justin said was that Gesher is now code named Sandy Bridge. Sadly he didn't mention things like video on die or PCIe on die if they don't do that first by greenlighting Jasper.

The other interesting bit is the naming of the new Chinese fab in Dalian. It is called Fab 68, a departure from the usual Intel naming scheme. This is because of numerology, six = Smooth Sailing, eight = Prosperity.

Call me overly cynical, but when you see a peace park, it is usually a good place to get shot while buying drugs, 'Best Steak Houses' are places to get dysentery, and Smooth Sailing Prosperity Fab says death by fire. This will be one to watch, anyone want to start a pool?

Then he went on to talk about phase change memory, a successor to flash. The first product is called Alverstone, a 128Mb flash replacement. It is a bit addressable memory type with many of the benefits of flash like a 10 year non-volatile life and a million write cycles before it dies.

Terascale was the next order of business, and we told you about it a while ago, but the one part missing was stacking memory on the die. This is done face to face, Polaris and Freya, the RAM, are bumped with copper bumps, and mated. Think peanut butter and jelly sandwich with each slice of bread being a chip, and when you mash them together, you get the stacked chips.

The Freya chip is thinned down to am 70 microns, and through-silicon copper via pass power and signals up to Polaris on top. It may sound simple, but you try doing it. Note: The part that says Polaris Die above is actually the air gap between the chips. Polaris is above the slide.

Last up, they dragged Pat Gelsinger on stage to crank up the frequency of Polaris to hit the 2TFlop performance number. Call me cynical again, but you can never forget the first teraflop, the second just seems like another... wait for it... flop.

Other than that, the keynote was mainly a wrapup of the near past announcements. Nothing hugely new, nothing groundbreaking, just steady progress toward all the goals you want in a fast efficient processor. µ