News & Analysis
Comment
yalanand
@Attoman I agree with you that there is real challenge in process development. I ...
SiliconAsia
I guess EEtimes also likes to have an oil money so can't say anything bad!!
GlobalFoundries charts road to 14 nm
Rick Merritt
8/30/2011 11:00 AM EDT
SANTA CLARA, Calif. – In the latest look down its road map, GlobalFoundries claims it is shipping thousands of 32-nm wafers a week, gearing up multiple 28-nm offerings for 2012 and planning first customer tape outs using a planar 20-nm process at the end of next year.
The company's recent partnership with Amkor Technology will help pave a path to new kinds of 3-D ICs, potentially at 28 and 20 nm. In addition it has ordered an extreme ultraviolet lithography system to be installed late next year that may be tested out in a 20-nm process and applied for work at 14 nm.
GlobalFoundries will sketch out its road map at its annual Global Technology Conference here Tuesday (Aug. 30).
"We are producing 65-, 45-, 32- and 28-nm chips and shipping to more than one customer from our Dresden fab," said Ajit Manocha, interim chief executive of the company in an interview with EE Times. The company has 150 customers including its former owner Advanced Micro Devices and customers of Chartered Semiconductor it acquired in 2009.
GlobalFoundries will name a permanent chief executive by the end of the year. Meanwhile, it has no plans on the table for any additional acquisitions, but it is studying the feasibility of building a fab in Abu Dhabi, home of its majority shareholder, Advanced Technology Investment Co. (ATIC).
The Dresden fab is shipping thousands of 32-nm high-K, metal gate wafers per week "on traditional yield curves," said Gregg Bartlett, senior vice president of R&D at GlobalFoundries. "Thirty-two nanometer HKMG is very challenging technology, but that transition is now behind us," he said.
The next goal is bringing up a handful of 28-nm processes sometime in 2012. At the event, the company will demonstrate a 3 GHz version of a dual-core ARM Cortex A9 chip made in its high performance 28-nm process and a 2 GHz version made in a low power flavor as early proof points of the node.
It will also describe a new high performance, low power 28-nm process geared for smartphone, tablets and notebooks it will release in tandem with Samsung. Compared to a 45-nm node, the new process could cut active power 60 percent while maintaining chip frequency or boost performance 55 percent while maintaining power leakage levels, it claimed.
Four 300-mm fabs will qualify the technology. GlobalFoundries' Fab 1 in Dresden and Fab 8 in New York along with Samsung's S1 in Korea and S2 in Austin, Texas.
Next: To 20-nm and beyond
|
|
|
|
|
Robotics Developer
8/30/2011 1:00 PM EDT
It is nice to see the roadmap! I wonder what the future holds for the processes beyond 14nm? The concept of a 3Ghz dual core ARM 9 is really appealing. Just for now I will have to wait...
Sign in to Reply
Attoman
8/30/2011 4:33 PM EDT
There is a real challenge in process development and control of 3D and sub-22nm structures and patterns. Metrology roadmaps have identified the need for 3D microscopy of 40 to one and better depth to width measurement and imaging which is not avaiable today.
Perhaps ATIC should consider world leadership in this critical area.
Sign in to Reply
yalanand
9/5/2011 6:16 AM EDT
@Attoman I agree with you that there is real challenge in process development. I feel leakage will be biggest gating facotr. I am not sure if high-K dielectric will solve all the leakage problems at 14nm.
Sign in to Reply
gutiea
8/30/2011 8:38 PM EDT
Has the company presented a roadmap for making money or at least not bleeding their soul out? That would be roadmap that would shock the world; roadmaps based on dreams, delays and losses are hardly worth discussing.
Keep the losses WELL hidden from the Abu Dhabians, we might be able to rebalance energy trade on account of them!
Good job GloFo, make up those mammoth operating losses on wafer volume! :-(
Sign in to Reply
SiliconAsia
8/30/2011 11:52 PM EDT
You got the point!
Sign in to Reply
Archeologist
8/31/2011 5:15 AM EDT
It is not a bad deal if Abu Dabi keeps unemployment in semiconductor industry low in USA and Germany
Sign in to Reply
wilber_xbox
8/31/2011 10:30 AM EDT
yeah, its surprising to read only the future roadmap. Do you know how well they fair quarterly?
Sign in to Reply
SiliconAsia
9/4/2011 1:50 PM EDT
I presume they don't have much to say about how well they fair quarterly or entire year. We will soon know by AMD announcement how many 32nm is being shipped if it is being shipped at all.
Sign in to Reply
resistion
8/30/2011 8:40 PM EDT
If they already made the jump to DP at 20 nm, they will re-use that at 14 nm. EUV is already late for their 14 nm development.
Sign in to Reply
greenpattern
8/31/2011 5:02 AM EDT
If graphics and networking require different packaging from mobile, it's a totally different game from just silicon, and Global has virtually no control in this aspect.
Sign in to Reply
Archeologist
8/31/2011 5:10 AM EDT
AMD is better of counting on TSMC than GF roadmap on paper
Sign in to Reply
docdivakar
8/31/2011 6:51 PM EDT
@Rick Merritt: it was good to meet with you again at GloFlo's conference.
In the afternoon tracks, I sat in the manufacturing track and got some good info on their packaging and MEMS plans. I am not how ever sold on the fab-less model for large volume MEMS which is still an IDM play led by ST & TI.
@greenpattern: you are some what partially correct. The gray-shaded area is like this: if there is backend fab activity (like flipchip, RDL, Si interposers) in a product, then the options are wide open, a fab like GloFlo can get that action though packaging houses can also play. But being upstream in that flow gives fabs an advantage.
Dr. MP Divakar
Sign in to Reply
greenpattern
8/31/2011 8:37 PM EDT
@Dr. Divakar: what I mean is, the packaging houses are holding the cards, rather than the silicon foundries.
Sign in to Reply
docdivakar
9/1/2011 12:00 PM EDT
@greenpattern: I agree... how ever, in the evolving 3D stacked IC ecosystem, even though packaging houses hold the cards, the business model is not yet clear as to who 'owns' the yield issues! So far, 3DIC via stacking has been a niche play for a few companies but its real growth can come from fabless design houses who can productize the technology with applications (of which software is a big part of innovation). When that happens, who do they engage first? Foundries or the packaging houses, or both? The last one may be the obvious choice but my hunch is not many fabs or packaging houses are willing to give a minute of their work day to an ASIC startup or small companies.
Foundries like GloFo, if they want to spur product innovation, can certainly do a lot more by leveraging their partnerships they just announced with Amkor.
MP Divakar
Sign in to Reply
help.fulguy
8/31/2011 7:49 PM EDT
Hey Rick, Come On. Where are the tough questions? Did you ask about the yield on 32nm? Did you just take their answer that they are done with 32nm! Why cant you ask the tough questions like an honest journalist would do! EETimes is becoming more of a PR company than a Technical oriented company! What the heck is going on with Llano supply issue? Is it an issue with their process? Whats with Bulldozer delay? Come on.
Sign in to Reply
SiliconAsia
9/4/2011 1:53 PM EDT
I guess EEtimes also likes to have an oil money so can't say anything bad!!
Sign in to Reply