IBM, GlobalFoundries move to 28nm process tech

...we have seen anything good coming from the IBM alliance???? I guess talk is free!

...would TSMC not be part of the fab 5? Seems like the past few years has been TSMC this and that, their hand in everything. Does it go by another name that was listed above?

Actually IBM R&D lab has developed the 32 and 45nm processas that AMD is using as well.
What else, umm yeah the Cell Processor, the Xbox360 processor...
currently developing photon based processor... i think there are several other venues tehy are engaged in.
Keep your fanboi comments to yourself.
Thanks for coming out, don't call us we'll call you.

Intel has already started manufacturing 32nm chips, which AMD hasn't yet. 28nm isn't a big step from 32nm. It'll only provide around 30% transistor density increase. Moving from 32nm to 22nm (which is the next node after 32nm in the ITRS roadmap), however, will provide 100% increase in transistor density (just like 65 to 45 or 45 to 32 provide). Even if there comes a time when AMD has 28nm processors while Intel is still selling 32nm, AMD won't have a big advantage. Intel are already developing their Haswell architecture, which will be launched on 22nm. So maybe they're already up to something.

they need two more

28nm is more or less an optical shrink of 32nm... not sure what all the press is about as it is typical for foundries to do half node shrinks (though this is close to a 3/5 node shrink). "2nd gen highK" means absolutely nothing when the first gen isn't out yet and it is not clear what the benefit of the "2nd gen" is without establishing a 1st gen baseline.

This is typical IBM talking about the future - and unfortunately you have to take a wait and see approach. I have no doubt high K will help on 32nm (but wouldn't expect much more on 28nm)... but just look back to 65nm where AMD was claiming 40% improvements over 90nm which never materialized (on a 'fab club' based process). The improvements might come, but there is not a recent track record to just blindly accept things at face value.

And be careful with the dates... by AMD (I mean GF's) definition, Intel is starting to ramp 32nm right now. What you don't hear is actual products out dates... that is the telling info (but is missing).

I might add the Power CPU line. That's not trash either.

AMD Hired Out Tsmc, Grew wearry of FABS Taking 20 or More Years To Iron out. So Global(College)Foundries Has struck Up/swing from its' Previous 135 Nm Limit Into Todays Edge, As AMD Needs Something.

With BoToxEta Getting Hoist Job & Better Offers from Intel, Inventor of CELL, now Lashes Out, Corez Galoree'. So Ibm Stays fresh out in Front W/ Chartered Snivieling thru Pockets of High grade MK Ultra cableline&wireless Mfgs Worldwide & AMD Gets it ALL as HardLine. Medal. Gamers With Smiley Faces, Probably O'bama relative. IPTV Finally Tells Truth To Mass Audiance & Mobile, My Gosh & Slosh, Already I can tell You Infinetron 100% increase in Gain ani't Total Recall, especailllll.... if you cann't see it, yet Mobile, huuummmm, you might not think just applying 20% more Strain on archetecture isn't much, yet 28 Nm says it all, Its' Too Close to Call. One Chip At Time. This is Perfected Archetcture in tighness, Yet most scream for 4 inch focus blurr Screen of High Speed thrill Lives & Even more Know Heavens.However, as those views do work,Wallah...Perfecto Ultee' drashek

Atty Adds: Getting 1 Perfecto In Third round 45/32/28 Leaves heavy Blow to Oldster TSMC Previouly Rapidly expanding Horizons. Until its found Out To Be Front for North Korean Gangsters.DAGOLD.

47TM 18C 127 PatPEND next....

As others have pointed out, this is an "improvement" over a non-existing 32nm HKMG process. And yes, this is a "low power" a.k.a. "low performance" process. Doing high-k in a low-power implementation is relatively straight-forward, TI was probably the first to do it quire a while ago. The really difficult thing is the high-performance implementation, where Vt targeting and device reliability are much more difficult to attain. By those measures Intel is still ahead. Yes, Raa Yee is correct, talk is cheap...

IBM alliance rallying cry should be "ours is smaller!!"

any demo?

POWER6 based Power 520 Express has price starting belo 6k$. It's running at 4.2GHz and is combination of 90nm and 65nm. The Thing is intended for market where reliability matters. Its SPEC-ratings are fantastic.

You don't propably need to be Einstein to guess that it will be lot faster when process shrinks?

IBM has been actually very good in delivering performance chips. They are likely to have most advanced Fabs and technology. Intel gets more publicity, since they are mainstream, but publicity and technological superiority happen to be different things.

Every member of alliance is going to benefit and that's the purpose of alliances.

Power6 runs circles around Nehalems and Itanics. And it doesn't cost fortunes. 28nm makes it even better.

Where to start...."28nm makes it better"
This is a LOW POWER/LOW PERFORMANCE process - it is not for CPU's.

"They are likely to have the most advance fabs and technology" Based on 'your' assessment? First to strain silicon, high K, metal gate, 45nm, 32nm... that would be IBM right? Ummm, no. Cu interconnect was the last major breakthrough that IBM was first to that had a measurable improvement. Then of course doesn't AMD also claim they have the most advanced fab? I'm just so confused...

"It's running at 4.2GHz and is a combination of 90nm and 65nm" Looking at JUST GHz is a ridiculous way to assess things - look at the architecture and the # of stages in the pipeline - the 4.2GHz is not due to the process, but the way the chip was intended. Intel did this with P4, and you know what, a "SLOWER" Core 2 (in terms of GHz) actually is much faster.

Thanks for the misinformation.

Don't mock SyBa, Mr. Ed.!