AMD GLOBALFOUNDRIES

it means AMD is now free to approach other foundries for microprocessor production
 
it means AMD is now free to approach other foundries for microprocessor production

Yes i thought of that as well.But i think at the same time it puts pressure to GF to do better with current cpus at GF.
If they want to keep AMD as a customer.
 
AMD still has a deal with GF so its not like AMD is ditching them, but this removes AMD from any bad press caused by poor manufacturing. but like empoy said it also leaves AMD with the option to go to a different company if GF completely screws up or can't produce want AMD wants.

and bigston made a good point, it does put pressure on them now because they no longer have the AMD money to back them up if shit hits the fan. give it a year or two and we'll see where GF ends up. only time will tell if it was a good idea or a bad idea GF. either way AMD comes out on the winning side of this deal.
 
Contract negotiations...reminds me of the Dr. Dre situation where he feels like everyone before Emenim that he was involved with treated him shady... Obviously AMD - GF don't see eye 2 eye at this time exactly.
 
I would not be surprised if more news comes out of this subject in the next couple of days or months.
Another deal perhaps with another fab.
 
AMD still has a deal with GF so its not like AMD is ditching them, but this removes AMD from any bad press caused by poor manufacturing. but like empoy said it also leaves AMD with the option to go to a different company if GF completely screws up or can't produce want AMD wants.

and bigston made a good point, it does put pressure on them now because they no longer have the AMD money to back them up if shit hits the fan. give it a year or two and we'll see where GF ends up. only time will tell if it was a good idea or a bad idea GF. either way AMD comes out on the winning side of this deal.

huh? AMD had an 8% responsibility on the company, why would they rely on AMD for cash, seems like the company backing GF has more money than AMD, plus AMD is paying to get out of it :) and the article made it seem that this would void AMDs deal on exclusive 28NM APU production @ GF.

GLOBALFOUNDRIES waived the exclusivity arrangement for AMD to manufacture certain 28nm APU products at GLOBALFOUNDRIES for a specified period. As a result, AMD agreed to pay GLOBALFOUNDRIES a cash payment of $425 million, of which $150 million is being paid on March 5, 2012, $50 million will be paid by July 2, 2012, $50 million by October 2, 2012 and $175 million in the first quarter of 2013. In addition, AMD issued a promissory note to GLOBALFOUNDRIES for the payments due by October 2, 2012 and first quarter 2013.
 
So now what TSMC for everything?
If they plan for 28nm or even 22nm for 2013 where would they go? :p
 
why not both?

http://www.brightsideofnews.com/new...lk-at-gf2c-more-details-from-the-new-wsa.aspx

Bingo.

Ditching SOI-based chip designs allows AMD to use TSMC for manufacturing as well so they'll rely far less on GloFo. TSMC doesn't use nor planned on using SOI in its fab process so to attain that flexibility they had to ditch it. Considering GloFo/IBM are really the only chip manufacturers that use SOI it's probably a smart move. I don't think there are any plans for SOI on 20nm anyway, or if there are I haven't read about it.

What is going to be interesting is that GloFo/IBM uses gate-first where TSMC uses gate-last which is also favored by Intel.
 
http://www.brightsideofnews.com/new...lk-at-gf2c-more-details-from-the-new-wsa.aspx

Bingo.

Ditching SOI-based chip designs allows AMD to use TSMC for manufacturing as well so they'll rely far less on GloFo. TSMC doesn't use nor planned on using SOI in its fab process so to attain that flexibility they had to ditch it. Considering GloFo/IBM are really the only chip manufacturers that use SOI it's probably a smart move. I don't think there are any plans for SOI on 20nm anyway, or if there are I haven't read about it.

What is going to be interesting is that GloFo/IBM uses gate-first where TSMC uses gate-last which is also favored by Intel.

I don't remember exactly where I read this, but, a few years ago, I read some speculation that SOI doesn't benefit from process shrinks as much as Bulk SI designs, and that the nature of SOI would limit how small the gate sizes can be.

I think I read it on RWT back when AMD was considering using SOI for Barton.
 
After they release of 28nm they should go to 20nm directly in my opinion,they have to close some gap from intel if they want to remain competitive.
And TSMC would be able to provide that i think for AMD.
TSMC is already testing 20nm with a company but i don't remember witch one.
 
Last edited:
I don't remember exactly where I read this, but, a few years ago, I read some speculation that SOI doesn't benefit from process shrinks as much as Bulk SI designs, and that the nature of SOI would limit how small the gate sizes can be.

I think I read it on RWT back when AMD was considering using SOI for Barton.

I think 20nm will mean FinFETS for everybody.

shark-fin.jpg


http://www.xbitlabs.com/news/other/...lfoundries_Tapes_Out_Its_First_20nm_Test.html
 
What is going to be interesting is that GloFo/IBM uses gate-first where TSMC uses gate-last which is also favored by Intel.

Not sure if this of special note; the quality of silicon produced depends on a lot more than the order the gate is put down. Also there are good arguments for both gate first vs gate last, many of which vary as a function of node size.
 
Yea, but the order has to do with price and feasibility.

Not only does high-k sharply reduce gate leakage, the gate capacitance scales with a thinner equivalent oxide thickness (EOT). Though mobility may be not quite as high as in a chip using a native oxide, cutting the EOT with high-k enables a shorter gate length and improves the drive current.

However, pressure is building on the gate-first approach. Some high-k experts argue that the high-temperature steps following the high-k dielectric and metal gate deposition cause the Vt to shift, affecting PMOS performance in particular. Others, including John Pellerin, director of technology at GlobalFoundries, argue that the gate-first approach requires less-restrictive layout design rules, provides for a smaller die size, and eases IP porting, while meeting the performance needs of customers at the 32/28 nm node.

http://scsong.wordpress.com/2010/03...010-03-10-154115-semiconductor-international/

It's a bet that Intel nor TSMC wanted to make but the Fishkill Alliance thought the benefits outweighed potential cons. And now it seems that IBM/GloFo both resorted to gate-last for the 20nm/22nm half node and node.

http://www.bit-tech.net/news/hardware/2011/01/20/ibm-and-globalfoundries-go-gate-last-for-20/1

Despite their previous insistence that gate-first was the correct approach to manufacturing future CPUs - including upcoming 32nm/28nm models - GlobalFoundries and IBM have now changed their minds, announcing a gate-last strategy for the fabrication of their 20nm and 22nm chips.

There are definitely arguments for both, but it seems GloFo and IBM both agree that it gets more difficult as you keep shrinking and the cons outweigh the pros. The transition also provides some unity, at least as far as the gate goes and that's likely a good thing.
 
Yea, but the order has to do with price and feasibility.
Well it is feasible; GF *was* able to get gate-first working at 28nm.

How do you know it negatively impacts price? Certainly from a customer perspective it offers a lot which would reduce price (better density, easier DRC, doesn't necessarily require a layout redo). GF has had yield problems, but I have no idea how internet-commentators would be able to attribute that to one specific step in the dozens involved in semiconductor fabrication.

The fact that GF is switching to gate-last doesn't really speak to whether it was a benefit or a detriment to their process at 28nm.
 
The disadvantage with Gate-first is overall lower yields, less working chips per wafer. When GlobalFoundries gets a question about this it replies that Gate-first is the right alternative for GlobalFoundries, and there are currently no problems. The saved space will offer a superior value over Gate-last.

Which is the way I've understood it as well. The cost would depend on the WSA, but it would be reflected regardless of what the agreement was because the cost would hurt GloFo either way and in turn their customers as well.

http://www.nordichardware.com/news/...rst-at-28nm-says-its-28nm-is-much-better.html

Iirc, their yield issues revolved around Llano and this was likely due to the transistor density in the graphical portion of the APU. It would explain why we got athlons on the FM1 platform :p
 
The article you linked cites no source for the claim that gate-first negatively affected yields. I have to wonder if the author just made some sort of causal assumption on this point.

In fact the article that it does cite indicates regarding the move to gate-last:
This was forced, he explained, by the fact that the ‘density and scaling benefits of gate-first high-K metal gate no longer apply because of patterning-dominated lithography restrictions.’

This point is also reiterated here.

The takeaway being that the benefits (at least in terms of DRC) are largely gone at 20nm due to lithography constraints, which is the motivatating factor in moving to gate-last rather than an impact to yield or cost.
 
We made significant progress last year to strengthen our relationship, and we’re pleased with GLOBALFOUNDRIES’ recent performance in meeting our delivery requirements across our product line.

That's quite the diplomatic statement.
 
That's quite the diplomatic statement.

Well they cannot say anything bad since they still have cpus at GF that are ready to come out.:p
Its either globalfoundries wakes up or amd will chose another fab for some of their products.
That is how things stand for now.
 
http://www.brightsideofnews.com/new...lk-at-gf2c-more-details-from-the-new-wsa.aspx

Bingo.

Ditching SOI-based chip designs allows AMD to use TSMC for manufacturing as well so they'll rely far less on GloFo. TSMC doesn't use nor planned on using SOI in its fab process so to attain that flexibility they had to ditch it. Considering GloFo/IBM are really the only chip manufacturers that use SOI it's probably a smart move. I don't think there are any plans for SOI on 20nm anyway, or if there are I haven't read about it.

What is going to be interesting is that GloFo/IBM uses gate-first where TSMC uses gate-last which is also favored by Intel.

it´s quite funny, while GloFo abandoning SOI for 28nm, TSMC seems go opposite route ;) GloFo is movin from Gate First to the Gate Last, TSMC goes for Gate First instead. Mad world :D

ibm-and-globalfoundries-go-gate-last-for-20nm

FD SOI TSMC

TSMC 28nm FDSOI
 
Back
Top