backtop


Print 20 comment(s) - last by IntelUser2000.. on Sep 21 at 8:30 PM


Intel's Paul Otellini holding up a 32nm test shuttle with SRAM lithography
The company is on track to bring its 32nm process in 2009

Intel today showed off its first test shuttle manufactured on a 32nm node process. Intel CEO Paul Otellini showed off the working 32nm chips during his keynote at the Intel Developer Forum. Although the test shuttles are working samples, the 32nm wafer contains no logic and just a test run of 32nm process technologies. Intel test shuttles are typically just SRAM.

The company plans to begin production of 32nm process processors in 2009 with the refresh of Nehalem Westmere. Westmere’s 32nm die shrink follows Intel’s tick-tock strategy, which alternates between process shrinks and new architectures. The company has parallel design teams all over the world to keep the tick-tock strategy on track.

"Tick-tock is the engine creating today's most advanced technologies and keeps them coming out at a rapid cadence,” Otellini said. “Our customers and computer users around the world can count on Intel's innovation engine and manufacturing capability to deliver state-of-the-art performance that rapidly becomes mainstream."

Following the tick-tock strategy, a mature 32nm process will be ready for Intel’s upcoming Sandy Bridge processor. Sandy Bridge is the successor to Nehalem, though Intel hasn’t said much about the architecture, which is due in 2010.


Comments     Threshold


This article is over a month old, voting and posting comments is disabled

Ornament
By Murst on 9/18/2007 5:46:58 PM , Rating: 2
I wonder if its possible to get one of those wafers anywhere (or even older, larger dye size ones). It'd make a great wall ornament in my living room.

I tried searching ebay, but nothing came up :-(

Not sure how the fiancee would react to that though...




RE: Ornament
By darkpaw on 9/18/2007 5:56:11 PM , Rating: 2
I wish I would have kept some of the wafers I got in a scrap lot a few years ago. Ended up giving them to my old semiconductor manufacturing professor though. They were pretty damn neat looking, but they were also much smaller 4" wafers I think.

Still, they would make quite a nice decoration.. geek interior design


RE: Ornament
By Polynikes on 9/18/2007 6:26:33 PM , Rating: 2
It beats turning your entertainment room into the bridge of the USS Enterprise.


RE: Ornament
By dever on 9/18/2007 6:56:56 PM , Rating: 2
When I left my last job with a semiconductor company I was given a framed wafter with a customized farewell message lithographed on it's surface by one of the people in our group. I've got it on my shelf... pretty cool, but no circuits.


RE: Ornament
By AmberClad on 9/18/2007 8:19:23 PM , Rating: 2
How much would one of those wafers that Otellini is holding up cost? I saw several of those IDF die shot pics, and every time, I couldn't help wondering how expensive it would be if he accidentally dropped it.

At least, the fully completed ones seem to me like they'd have quite a few dies on them, and would therefore be extremely expensive (multiply the cost of your average CPU by however many dies are on there).


RE: Ornament
By theapparition on 9/19/2007 8:51:53 AM , Rating: 2
If I told you, you'd probably go out and slash your wrists. It is ridiculously cheap. However, you are not paying for the materials, your also paying for the cost of the fab, labor, overhead, wire-bonding, packaging, distribution, sales, marketing, etc.

In the end, though, what sets a chips price is the perceived value in the marketplace.


RE: Ornament
By AmberClad on 9/19/2007 11:20:01 AM , Rating: 2
Eh, even after learning that, I guess I don't feel too terribly ripped off for paying hundreds of dollars for a piece of silicon on a PCB with an aluminum cover. I understand full well that the CD or DVD that a piece of software comes on only costs a few pennies at most. So I guess I shouldn't be surprised that a silicon wafer itself isn't worth much either.


RE: Ornament
By KristopherKubicki (blog) on 9/19/2007 12:21:19 PM , Rating: 2
quote:
How much would one of those wafers that Otellini is holding up cost?

Materials-wise? It's pennies. But obviously, that's not the cost.

A friend of mine works at a Taiwanese fab -- they're about 3 process nodes behind Intel. The test wafers go for about $400K each. Part of this guy's job includes physically hand delivering the test wafers to the clients in the US.

I'd have to assume the Intel ones go for considerably more :)


RE: Ornament
By 9nails on 9/18/2007 9:13:06 PM , Rating: 2
I have, on my wall at work, a post showing the 386 CPU all blown up with call-outs describing each section of the CPU and their function. It was something that I inherited from the previous person in the position. I've been proud to hang it.

On that poster is the actual 386 CPU for size reference, and it's still amazing to think of how far miniaturization had come in that design. It would be even more impressive with new CPU's. I would PAY for a new poster!

It would be a great way for Intel to get rid of failed silicone and for us enthusiasts to own a piece of history. There must be thousands of walls in computer labs just wishing they could update their pictures too.


RE: Ornament
By phaxmohdem on 9/18/2007 11:47:43 PM , Rating: 2
Its hard to find wafers of actual AMD/Intel cpu's but I have purchased a few wafers for wall decor off ebay before.

http://cgi.ebay.com/8-Tungsten-Microchip-Pattern-S...


RE: Ornament
By Hoeser on 9/19/2007 8:52:14 AM , Rating: 2
If you can find Intel's educational kits from the mid 90's they included a wafer of I believe Pentium-60 or Pentium-90 chips. Might have been something else, but it does look cool. I have one in my room. We had the kits in grade school and high school (Canada).

The kits were in a giant white box which folded out to look like a computer. It included a Wafer, some text material, a ridiculous VHS video titled "The Journey Inside the Computer" or something to that effect, and a bunch of other neat stuff.


The sleeping giant
By Adonlude on 9/18/2007 6:23:16 PM , Rating: 5
Intel has sure turned into a rapid progress machine. Core 2 blowing us away, Nehalem taped out, and the 32nm process on deck. Looks like AMD's pulling ahead in the beginning of this decade was much like Japan's bombing of Pearl Harbor... They've awoken a sleeping giant!




RE: The sleeping giant
By Expunged on 9/18/2007 10:11:57 PM , Rating: 1
Well said, lets hope that Intel will face strong opponents in the future to keep them on track. If AMD is unable to successfully battle Intel in the future, hopefully a company with outside the box ideas will step in and provide a new challenge. I would hate to have all this potential demonstrated and then go dormant if Intel progresses into a dominant position with no real threats.

Back to your comparison, it would be awful for Intel to crush their competition as the United States did and then go unchallenged for the next 70 years (3-5 years in the semiconductor world). This would most likely result in the deterioration of their R&D, ideas, foresight and spending in critical areas, much as the United States as stagnated with the common perception of any possible threats. (Lets not turn this into a political battle here though floks, just drawing some parallels.)


RE: The sleeping giant
By Master Kenobi (blog) on 9/19/2007 6:53:52 AM , Rating: 1
Well your right, but in the case of the US Military I think it has more to do with the public just doesn't seem to be able to stomach a long and drawn out fight regardless of the reason. I think there is an old phrase like "Let the military take the gloves off and do what they were trained to do."

In any case this does lead to some questions about AMD. Barcelona was a paper launch, we have still not seen this thing show up in the channel or OEM vendors. Intel is dropping Penryn in less than 2 months and I suspect it will be a full bore channel and OEM launch on release day with chips available immediately (Intel hasnt pulled a paper launch on new chips so far). Less than a year after Penryn they will drop Nehalem and the CSI interconnect. If I was AMD, I would be entrenching myself to weather the storm.


32 nm
By flipsu5 on 9/18/2007 11:47:18 PM , Rating: 3
I applaud Intel making a 32 nm commitment.

In 10 years, things will be very different.




RE: 32 nm
By ChipDude on 9/19/2007 12:40:11 AM , Rating: 2
What happens in 10 years? The grinch comes and steals Moore's law?

yeah its cool. I have a 1", 2", 3", 4", 5", 6", 8" and 12" product wafer. One of these days I'll frame it. I think I'll wait for that 450mm wafer.

By the way I don't think that 32nm 300nm is expensive in one sense. There is no product worth selling. I'm sure if there was anything valuable to be tested they already tested it before Paul got his greasy fingers on it.

Or its very expensive in the sense IBM and AMD would pay a handsome sum for that wafer.



RE: 32 nm
By Master Kenobi (blog) on 9/19/2007 6:55:32 AM , Rating: 2
Not really, the SRAM is easy and cheap to print on the wafer. Now if that wafer had say actual CPU logic circuits on it, yea then it would be worth a pretty penny.


Shrinking
By geddarkstorm on 9/19/2007 11:43:05 AM , Rating: 2
I wonder what they'll do when they run out of the ability to do any more meaningful shrinks. The leaps are already greatly slowing down. 130? to 90 to 65 to 45 to 32nm. I can't imagine how much farther they can go now; maybe two more rounds?

An atom is about one Angstrom, or 0.1nm, and a typical protein is about 1nm in diameter, so they are quickly approaching molecular size on these silicon wafers. Soon enough they'll have so much leaking and interference from the silicon that there will be signal to noise ratio problems, which I thought was always the reason silicon was perceived to be poor for nanotech. Obviously they'll switch to nanotech and probably carbon, but the die shrinking brick wall seems to be coming up soon--as in mere few years. So it seems to me anyways--and without die shrinking strategies, I wonder what other paradigms they'll switch their "tick-tock" philosophy to?




RE: Shrinking
By IntelUser2000 on 9/21/2007 8:30:18 PM , Rating: 2
quote:
I wonder what they'll do when they run out of the ability to do any more meaningful shrinks. The leaps are already greatly slowing down. 130? to 90 to 65 to 45 to 32nm. I can't imagine how much farther they can go now; maybe two more rounds?


I don't know how you consider it as slowing down. In absolute values yes, 130-90=40, 90-65=25, but look in terms of percentage. Have you actually caculated the numbers out:

90/130=0.692
65/90=0.722
45=65=0.692

About 70% reduction of size per generation in one dimension, meaning overall size reduction of 50%(0.7*0.7x=0.49).


By PAPutzback on 9/19/2007 9:27:57 AM , Rating: 2
I was getting ready to sit back and wait for another refresh before building a new system. And USB 3.0 to boot coming soon. Now where is the 1,000,000 DPI, blue laser, titanium mouse for that bus.




"When an individual makes a copy of a song for himself, I suppose we can say he stole a song." -- Sony BMG attorney Jennifer Pariser

Related Articles
"Nehalem" Taped-out and Running Windows
September 18, 2007, 12:04 PM













botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki