AMD Confirms 32nm Yield Issues at Global Foundries

HopJokey

Platinum Member
May 6, 2005
2,110
0
0
http://www.anandtech.com/show/4894/amd-confirms-32nm-yield-issues-at-global-foundries

AMD just announced revised revenue projections for Q3. Revenue is up compared to Q2 by 4 - 6%, but AMD had originally expected an increase of 10%. The reason for the revised projections? Llano supply is limited by apparently poor yields on Global Foundries' 32nm process. We had heard rumors to this effect for a while, but now they're officially confirmed by AMD.

The official statement is below:

The less-than-forecasted preliminary third quarter 2011 revenue results are primarily due to 32 nanometer (nm) yield, ramp and manufacturing issues at GLOBALFOUNDRIES in its Dresden, Germany factory that limited supply of "Llano". Additionally, 45nm supply was less than expected due to complexities related to the use of common tools across both technology nodes. AMD continues to work closely with its key partner GLOBALFOUNDRIES to improve 32nm yield performance in order to satisfy strong demand for AMD products.

The bigger concern in the near term is the impact this will have on the ramp of Bulldozer. Llano wasn't a huge chip, Bulldozer is.

Source: AMD
 
Last edited:
Aug 11, 2008
10,451
642
126
Just another problem for AMD. Intel has been at 32nm for what now, 9 months??
Seems like I am always criticizing AMD, and I really would like them to do well, but they are just such a frustrating company. Llano is attractive in the mobile space to me but meh on the desktop, and bulldozer has had delay after delay. And once they get a popular product, they cant meet demand.
 

lifeblood

Senior member
Oct 17, 2001
999
88
91
I wonder if they can contract with Intel to make their CPU's?

Ok, I'm being sarcastic.
 

HopJokey

Platinum Member
May 6, 2005
2,110
0
0
Just another problem for AMD. Intel has been at 32nm for what now, 9 months??
Seems like I am always criticizing AMD, and I really would like them to do well, but they are just such a frustrating company. Llano is attractive in the mobile space to me but meh on the desktop, and bulldozer has had delay after delay. And once they get a popular product, they cant meet demand.


Intel has had a released 32nm CPU since Jan 2010 (Westmere). That would make it almost 21 months.

http://www.computerworld.com/s/article/9143031/Intel_gets_ready_to_launch_32nm_Westmere_chips
 

BlueBlazer

Senior member
Nov 25, 2008
555
0
76
I wonder if they can contract with Intel to make their CPU's?

Ok, I'm being sarcastic.
If that were to happen then AMD's stock would shoot upwards. Supply issues solved overnight. But at the moment, I don't think Intel has decided to cater for other third party chips yet (not even Apple). :D
 

OCGuy

Lifer
Jul 12, 2000
27,224
37
91
And the process lead grows and grows for Intel...

AMD is having problems with 45nm still, and Intel is deciding when to ramp up 22nm..
 

RobertPters77

Senior member
Feb 11, 2011
480
0
0
If that were to happen then AMD's stock would shoot upwards. Supply issues solved overnight. But at the moment, I don't think Intel has decided to cater for other third party chips yet (not even Apple). :D

Tis' a shame. Intel could be the invisible hand in the semi conductor industry with their fabrication capability. Theoretically they could give TSMC, GloFo, and Others, a run for their money. I don't know about you but how many OEMs would kill for access to intel's fabs? Alot. A hell of a lot.
 

trollolo

Senior member
Aug 30, 2011
266
0
0
so what does yield mean in relation to cpu's? i know its English definition
 

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
Additionally, 45nm supply was less than expected due to complexities related to the use of common tools across both technology nodes.

OMG that is not good, not good at all.

As a process development guy I can tell you exactly how this reads to me: they've got a serious contamination (metals, etc) issue from cross-over of their 32nm HK/MG wafers on the shared tools used for 45nm capacity.

That's the kind of stuff that can snowball into having an entire fab shutdown to eliminate the contamination.

We had a fab in singapore once that was shutdown for nearly 2 months because of a metals contamination issue in the front-end. Also our houston fab was shutdown for nearly 2 weeks for a metals contamination issue.

That AMD press piece read to me exactly how I would expect them to communicate a metals (or other from the HK portion) contamination issue in the fab which is bleeding over into impacting the yields/reliability of their 45nm chips.

I hope I'm misreading that because being dependent on a single interconnected fab complex like AMD is can be disastrous in such a situation.

It kinda sounds like they caught it in time to limit the impact to just a few of the tools, thus impacting capacity but not taking down the entire line. That's a much rosier outcome. I'll have to check around on this one.
 

theeedude

Lifer
Feb 5, 2006
35,787
6,197
126
Is it a specific process step where things get cross contaminated? Is the metal gate deposition equipment shared with some poly-gate process step?
 

cebalrai

Senior member
May 18, 2011
250
0
0
Is it a specific process step where things get cross contaminated? Is the metal gate deposition equipment shared with some poly-gate process step?

This is in Germany dude. Bits of weiner dog fur and schnitzel grease get into the fabs and next thing you know we're up to our ears in tri-core CPUs.
 

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
Is it a specific process step where things get cross contaminated? Is the metal gate deposition equipment shared with some poly-gate process step?

It won't be the metal deposition equipment per se, that will be for sure dedicated and segregated.

It will be the other process-materials non-specific process steps. The litho tools, the implanters, the cleans tools, the etch tools, as well as the other deposition tools (dielectrics for sidewall nitride/oxide/cap/etc).
 

dbigers

Junior Member
Jun 28, 2004
21
0
0
It won't be the metal deposition equipment per se, that will be for sure dedicated and segregated.

It will be the other process-materials non-specific process steps. The litho tools, the implanters, the cleans tools, the etch tools, as well as the other deposition tools (dielectrics for sidewall nitride/oxide/cap/etc).

Although I know you don't have specific knowledge of AMD's problems at the moment. I do appreciate your insight to the specific problems that can arise with this complex process.

When you mentioned the Houston fab, if you don't mind if I ask, what company do you or did you work for?

Thanks again for the insight and for helping me to understand what this means. Personally I don't have much vested in BD at this point. I may upgrade my rendering machines if I see gains and performance in Cinebench, which I use. (Cinema 4D). Unlike some, I did not go and guy MB's and other stuff specific to either platform. Otherwise, if BD doesn't do so well, a few 2600k's will go great with the 2500k gaming rig I just built.
 
Last edited:

Mopetar

Diamond Member
Jan 31, 2011
8,491
7,747
136
so what does yield mean in relation to cpu's? i know its English definition

In general, it's similar to the English definition in that it refers to the amount produced. It can also be used to refer to the percentage of dies per wafer that are good.
 

dbigers

Junior Member
Jun 28, 2004
21
0
0
Surely AMD could find something to sue Apple over. Keeping Apple busy would keep them from suing everyone else.
 

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
When you mentioned the Houston fab, if you don't mind if I ask, what company do you or did you work for?

TI...although I just realized I may have been thinking of the Sherman fab :confused: that was '96 when it got hit.

Dealing with metals contamination is fairly routine in the industry owing to the introduction of Copper processing 10+ yrs ago. That required stringent/robust controls to be implemented for segregating the FEOL from the BEOL (not that there weren't stringent cross-contamination controls before then as well) to prevent copper contamination of the FEOL (front-end of the line, where the xtors are built).

Gate-last MG integration has many advantages over gate-first, one being the easier segregation of the MG contamination from the FEOL toolsets.

It is but one complication that GloFo has to deal with that Intel doesn't.

That said, it would be really surprising if they are getting hit by a cross-contamination issue, the expertise exists to deftly manage the risks...that said, there it is in black-and-white from AMD themselves and there are only so many ways to interpret the "on the ground" reality of what those words are communicating.

It could be x-contamination, it could also be a capacity constraint in the truest sense. What that would mean is that they are capacity constrained on the very tools that are shared, which would imply a planning fiasco of some sort.

This would be something like not ordering enough litho steppers which both 32nm and 45nm use, and as such GloFo finds itself in the position of having to reduce 45nm WIP in order to process 32nm WIP.

That scenario would be far less dire and much more your garden variety "capacity planning" type management gap.

The more I think about the nature of the vernacular in the press release the more I am inclined to suspect this is really nothing more than a pure capacity limitation issue rather than a x-contamination issue. My initial gut-reaction response is probably wrong. I won't know anything IRL until tomorrow at the earliest though, if ever.