GlobalFoundries: Only a process tech licensee?

[Fjodor2001]

Hi,

So we all know Global Foundries (GF) will license Samsung 14 nm process technology. But does this mean that GF has given up on developing their own process tech going forward too? Or will we see 10, 7, 5 nm GF nodes?

At least we know this prediction did not turn out as expected:

 

[Jimzz]

Depends what/who they get in the IBM deal would be my guess.

 

[sm625]

lol @ that roadmap.

 

[Arachnotronic]

lol @ that roadmap.

This

 

[ShintaiDK]

It only shows what a joke GloFo is and how misleading foundry roadmaps can be.

 

[Fjodor2001]

Depends what/who they get in the IBM deal would be my guess.

Hmm.... could you please clarify that?

 

[DrMrLordX]

Does Globalfoundries have any customer right now that is not AMD?

 

[krumme]

Does Globalfoundries have any customer right now that is not AMD?

The high Lords

 

[Burpo]

With emphasis on "High"

 

[krumme]

Carrizo density is that only hdl and design? perhaps gf is going in reverse pr. New strategy instead of the old fairytales. Lol. Promote it as 28nm while its more like 22nm. New style in town.

 

[seitur]

lol @ that roadmap.

Indeed.

 

[NostaSeronx]

Fab 8 is of the mainstream bulk development. (copy Samsung etc)
Fab 1 is of the advanced bulk/(+mainstream)SOI development. (copy STM/IBM + improvements)
Fab 9 will follow Fab 8, while the acquired IBM Fab will follow Fab 1.

Samsung's LPE/LPP FinFETs are and still will be better than GlobalFoundries XM FinFETs.

LPM/XM series are cancelled replaced by the Advanced series. (GF28A/GF20AM/GF14AM/etc)

 

[Jimzz]

Hmm.... could you please clarify that?

GF was paid to take IBMs foundries and also got many engineers as well. If IBM was working on any process designs then GF could use those and come up with their own.

 

[ShintaiDK]

Carrizo density is that only hdl and design? perhaps gf is going in reverse pr. New strategy instead of the old fairytales. Lol. Promote it as 28nm while its more like 22nm. New style in town.

Its just HDL. They could use UHD for even smaller.

But performance would be even lower than Carizzo.

 

[Fjodor2001]

Its just HDL. They could use UHD for even smaller.

But performance would be even lower than Carizzo.

I suppose you mean Carrizo performance would be even lower than Kaveri. But then how come AMD is still considering desktop Carrizo? See this.

 

[ShintaiDK]

I suppose you mean Carrizo performance would be even lower than Kaveri. But then how come AMD is still considering desktop Carrizo? See this.

Thats a Kaveri Refresh. Not Carrizo.

 

[Enigmoid]

I suppose you mean Carrizo performance would be even lower than Kaveri. But then how come AMD is still considering desktop Carrizo? See this.

Carrizio implements a number of improvements. There is no reason not to use the architecture.

I read that as "If we can get over the high density libaries problem or push the crossover point between high performance and high density libraries to higher performance levels then there may be a desktop Carrizio."

 

[ShintaiDK]

http://www.dvhardware.net/news/2015/amd_godavari_cpus.jpg

 

[Fjodor2001]

I read that as "If we can get over the high density libaries problem or push the crossover point between high performance and high density libraries to higher performance levels then there may be a desktop Carrizio."

Nothing in the article says that. It says that desktop Carrizo was down prioritized because of budget reasons:

So the first new piece of information we got was the fact that AMD actually had a desktop Carrizo planned, something that was later canceled due to management’s decision. It was originally designed to be put onto the FM2+ socket and FP4 packaged and relieve the Kaveri, being a true successor to the older APU.

However, to cut costs and following the profit motive (something rather understandable) they decided to reiterate Kaveri with a refresh named Godavri.

 

[Fjodor2001]

Thats a Kaveri Refresh. Not Carrizo.

Huh? Not sure what you mean. Please clarify. The article I linked to says:

Interestingly he also mentioned that Carrizo still might end up on the desktop segment and AMD is currently debating this very question.

 

[Enigmoid]

Nothing in the article says that. It says that desktop Carrizo was down prioritized because of budget reasons:

Precisely. If the intended performance benefits are not large enough then there is little point in using it.

To get Carrizio to a point where the ROI was beneficial would require higher clocks, something not possible with high density libraries. AMD does not have the budget to make a high performance carrizio as well as a power optimized version.

Another point on the HDL. AMD sells the same chip as desktop and mobile. The cost difference between selling a die as mobile only and desktop and mobile is minimal; AMD has nothing to lose by selling carrizio on the desktop as well as mobile (Selling on desktop actually helps mobile performance because you can send the more power-hungry bins to the desktop instead of having to use them for a mobile SKU). However, Carrizio is not suitable on the desktop because of the clockspeed reductions due to HDL. It would not be a compelling sale and money would be lost (like you said everything is a financial decision but its why that decision was made that is important). Quite likely, if midway through Carrizio's lifetime a lot of fab improvements are made (a la trinity -> richland), allowing for higher speeds, carrizio may make it to the desktop.

You are saying budget reason. I agree. But the reason those budget reasons are there is because HDL makes this product undesirable for the desktop.

 

[monstercameron]

Does Globalfoundries have any customer right now that is not AMD?

rockchip afaik

 

[Idontcare]

GlobalFoundries: Only a process tech licensee?

So we all know Global Foundries (GF) will license Samsung 14 nm process technology. But does this mean that GF has given up on developing their own process tech going forward too? Or will we see 10, 7, 5 nm GF nodes?

The way the fab club works is you basically are paying an upfront membership fee in lieu of paying a backside licensing fee, but in terms of how the IP (process node) itself is created, developed, and rolled out it is very much a "licensee" type of business model.

In that regard, Globalfoundries has since its birth been a process tech licensee. IBM's R&D engineers develop the nodes, taking feedback from a handful of GloFo R&D engineers who are stationed in Fishkill but pretty much required to sit on their hands on the sidelines and watch from afar, and then handing off the node to the fab club members so they can internalize it within their fabs and attempt to ramp the yield from near zero to something manufacturing worthy.

The model itself was rather broken, not anything close to the image of collaborative partnership that made it into the PR, but IBM recently had to give GF $1.5B just to convince GF to take their fabs off their hands because even IBM wasn't making enough money off its fab club partners to support the business model. (ergo the abrupt swing towards licensing 14nm from Samsung)

We'll see what happens down the road, I bet Samsung will license 10nm to GF as well. The timing is too soon for GF to absorb the IBM M&A to get their own 10nm stuff off the ground in any better shape than 14XM got off the ground.

But 7nm may see GF roll out their own true internally developed node. If they don't, then it is hard to conceive of what kind of future they will have as a foundry that relies on other foundries for their process tech.

 

[DrMrLordX]

The high Lords

Whoa, check out the gate pitch on that . . . wait, what was I saying?

rockchip afaik

Huh. Is that going to continue now that Intel is involved with Rockchip?

Nothing in the article says that. It says that desktop Carrizo was down prioritized because of budget reasons:

It actually says that AMD is kinda sorta maybe thinking about desktop Carrizo in H1 2016. Whether that means FM2+, an early FM3 implementation, or just FP4 BGA repackaged is anyone's guess.

 

[Fjodor2001]

Precisely. If the intended performance benefits are not large enough then there is little point in using it.

Nothing in the article says that it's due to performance reasons, only due to budget reasons.

AMD is in a tight financial position so there could be a number of reasons why they had to down-prioritize Carrizo on desktop. They can't run too many projects in parallel, and I guess their focus is on K12 and Zen on 14 nm right now.