GloFo and TSMC FinFet News

[mrmt]

We can hope, but have we even seen any 28nm chips from GF yet?

What if GLF wasn't AMD foundry, would you, piedsquared and others buy GLF claims at face value or even hope to GLF succeed in the same way you do now, or would you have the discernment to take GLF/IBM claims with the much needed huge mountain of salt?

In this node discussion I see the same psychological mechanism people had with AMD. Bulldozer had to succeed because AMD had to succeed, most people couldn't even fathom the possibility of Bulldozer being a failure, and those people stuck to the end with their hopes, even in face of legit leaks anticipating the failure it was, and now I see the same of GLF.

If it's not clear enough, no matter how people root for GLF, it won't make a difference in whether their 14nm process will be competitive or not, and there isn't anything pointing out for a success now. GLF 28nm chips will hit the market two years after TSMC 28nm process (!) and almost four years after Intel 32nm (!!!) process, and yet people expect GLF to field a 14nm process (which they won't, it's a half-baked upgrade of their 20nm process) competitive against Intel 2nd generation finfet process.

If GLF really had their 20/14nm process on track for 2014 production, there would be no point in selling 28nm as theya re doing now, much less committing themselves to a 28nm FDSOI now, as those processes would deliver chips some 6-9 months ahead of 14nm.

I really cannot understand GLF. A foundry business is all about credibility, is all about being credible enough to a multi-billion business to trust the manufacturing of their final product in your hands for years. When GLF promises something and doesn't deliver, and they are doing that for too long, they are in effect thoroughly trashing their reputation in a way that may not be repairable.

 

[SocketF]

One good thing in case of credibility and GF is the now often forgotten 20nm process. According to the roadmap they should be ready for production this year.

I am really sitting back and watching that 20nm-show. If they can deliver it, I'll believe their 14nm statements. But as IDC last time mentioned correctly in the FD-SOI thread ... they showed yield-numbers of 28nm at common platform, not of 20nm ...

Now everybody has to ask him-/herself: Why?

 

[Idontcare]

The aggressive timeline plans captured in their roadmap certainly defies logic given their resources. TSMC has 4x the resources and they aren't even willing to commit to the pie-in-the-sky type roadmap that GloFo is putting out there.

Still, would be nice if it lights a fire at TSMC and they throw money at making sub-20nm happen sooner instead of later.

 

[AtenRa]

.

If GLF really had their 20/14nm process on track for 2014 production, there would be no point in selling 28nm as theya re doing now, much less committing themselves to a 28nm FDSOI now, as those processes would deliver chips some 6-9 months ahead of 14nm.

Not everyone is using the latest process early on. That means there are customers still using 65/40nm process.
And lately it seams that chip makers will stay longer in the same process than before. That means that 28nm will be the preferred process for more than two(2) years.
There is the capacity availability to take in to consideration with GloFo because currently the majority of it its taken by AMD. Add that global 28nm availability is still very low because everyone still using it.
Also, People forgetting that from 2014 GloFo will have one more fab (New York) to help both in total wafer capacity (60 to 80 thousand wpm) and lower transition process times. 20nm will be available in Fab 8 in New York first and then transition to Dresden. That means that you dont take off 28nm capacity to transition to 20nm or 14XM.

14XM over 20nm bulk will be like 28nm FD-SOI is to 28nm bulk. The difference is that it may be cheaper than 20nm FD-SOI and it specifically targeting the LPH and not the high performance node. From my understanding, it seams that 14XM is targeting current 28nm chip makers that will be willing to transition to a smaller node by 2015.

So from my understanding, from 2014 and beyond it seams that GloFo will be in a much better way than it is now and that will also help AMD.

 

[AtenRa]

Just to add,

GLOBALFOUNDRIES To Offer Adapteva's Processor IP For 28nm SoC Designs

Santa Clara, Calif., February 5, 2013 — At today’s Common Platform Technology Forum, GLOBALFOUNDRIES announced a partnership with Adapteva to offer the company’s Epiphany IV microarchitecture to customers using GLOBALFOUNDRIES’ leading-edge 28nm-SLP process technology. The agreement will allow chip designers to integrate Adapteva’s massively parallel multicore technology into their next-generation system-on-chip (SoC) designs, enabling server-level performance in mobile devices such as smartphones and tablets.

Adapteva, a privately-held semiconductor technology startup, collaborated closely with GLOBALFOUNDRIES to optimize the Epiphany IV design based on the 28nm Super Low Power (SLP) process, allowing Adapteva to become the first company to demonstrate a processor operating at 50 GFLOPS/Watt. Adapteva’s partnership with GLOBALFOUNDRIES has allowed the company to dramatically improve energy efficiency over multiple generations, maintaining overall power consumption at just 2 Watts while moving from 16 cores at 65nm to 64 cores at 28nm.

Epiphany-IV 64-core 28nm Microprocessor (E64G401)

Features

64 High Performance RISC CPU Cores
800 MHz Operating Frequency
100 GFLOPS Peak Performance
1.6 TB/s Local Memory Bandwidth
102 GB/s Network-On-Chip Bisection Bandwidth
6.4 GB/s Off-Chip Bandwidth
2 MB On-Chip Distributed Shared Memory
2 Watt Maximum Chip Power Consumption
IEEE Floating Point Instruction Set
Fully-featured ANSI-C/C++ programmable
GNU/Eclipse based tool chain
Source synchronous LVDS off chip links for host or direct chip-to-chip interfacing.
Chip to chip links for integrating up to 64 chips on a single board
324-ball 15x15mm flip-chip BGA

also


GLOBALFOUNDRIES and Cyclos Semiconductor Partner to Develop High-Performance, Low-Power ARM Cortex-A15 Processors Using Resonant Clock Mesh Technology

Santa Clara, Calif., February 5, 2013 — At today’s Common Platform Technology Forum, GLOBALFOUNDRIES and Cyclos Semiconductor, the inventor and only supplier of resonant clock mesh (RCM) technology for commercial IC designs, announced plans to implement Cyclos’ low-power semiconductor intellectual property in ARM® CortexTM-A15 processors using 28nm High-K Metal Gate (HKMG) process technology from GLOBALFOUNDRIES.

The partnership will enable GLOBALFOUNDRIES’ customers to design next-generation SoCs with class-leading performance and power efficiency. It also reflects Cyclos’ commitment to ensure the broadest possible availability of its RCM IP technology for reducing power consumption.

 

[mrmt]

One good thing in case of credibility and GF is the now often forgotten 20nm process. According to the roadmap they should be ready for production this year.

I am really sitting back and watching that 20nm-show. If they can deliver it, I'll believe their 14nm statements. But as IDC last time mentioned correctly in the FD-SOI thread ... they showed yield-numbers of 28nm at common platform, not of 20nm ...

Now everybody has to ask him-/herself: Why?

If you are waiting for GLF 20nm to measure their credibility, you should write them off already, as the reason for not giving 20nm numbers is rather obvious: The numbers are not good.

There is no point in not showing the numbers if they were already good or at least in a trend that would put them in acceptable levels by the end of this year. The fact they didn't show means that GLF is far from that goal. Also for 20nm HVM this year means tape out last year, or at least Q113, and we didn't hear anything about this. So 20nm will be a flop, no doubt about it.

What remains to be seen is the level of their 20nm flop. Will it be a small delay that will put them in H114 or a big flop that will put them well into 2015?

 

[SocketF]

Yes you are probably right, but lets give them a (very) last chance, I have mercy with them ;-)

 

[Ajay]

I wonder what Mubadala exit strategy is?

 

[podspi]

What if GLF wasn't AMD foundry, would you, piedsquared and others buy GLF claims at face value or even hope to GLF succeed in the same way you do now, or would you have the discernment to take GLF/IBM claims with the much needed huge mountain of salt?

In this node discussion I see the same psychological mechanism people had with AMD.

Of course I would still hope for their success.

I agree we are seeing the same psychological mechanism people had with AMD -- GLF is AMD's foundry so many people here want them to fail. I want GLF to succeed because competition is good. Hey, I want TSMC to succeed as well (and I'd be very sad to see Intel go if it ever came to that -- which I doubt it will).

How you took my incredulous comment on GLF's roadmap as a pro-AMD fanboy comment I do not know.

 

[Idontcare]

I wonder what Mubadala exit strategy is?

GloFo IPO

 

[Abwx]

Mubadala exit strategy is to pour even more money at the problem , they have a lot of it and are not bound by quarterly results.

 

[Ajay]

GloFo IPO

Wow, that's scary!

Mubadala exit strategy is to pour even more money at the problem , they have a lot of it and are not bound by quarterly results.

Then should offer Intel's Bohr a ten year $200M paycheck. This guy is worth more than any sports figure I know of (and more than most CEOs for that matter).

 

[Abwx]

No guy is worth this price.

Bohr would be nothing without the thousands anonymous
engineers that are doing the actual work.

 

[Ajay]

No guy is worth this price.

Bohr would be nothing without the thousands anonymous
engineers that are doing the actual work.

Same can be said of CEOs, IMHO. But you kind of missed my point anyway.

 

[mrmt]

GloFo IPO

Why an IPO? The first thing you would want to do would be to get rid of the old management, or at least circumscribe their power to non-vital areas, and you can't really do that in an IPO environment. OTOH, with a private equity company buying a controlling stake, not only they would be able to change management, if they were successful Mubadala would recoup a lot of their investments.

 

[Idontcare]

Why an IPO?

No particular reason other than it is an exit strategy that gets them out of GloFo while putting money back in their coffers. If you don't like an asset then you sell it. Same reason GloFo came to exist in the first place.

 

[Phynaz]

Mubadala exit strategy is to pour even more money at the problem , they have a lot of it and are not bound by quarterly results.

Generally, the people with a lot of money understand its value more than most. Therefore they would understand that pouring more money into a failure is not a strategy.

Just because GloFo's investors have a lot of money doesn't mean they want to waste it anymore than you or I do.

 

[Ajay]

Why an IPO? The first thing you would want to do would be to get rid of the old management, or at least circumscribe their power to non-vital areas, and you can't really do that in an IPO environment. OTOH, with a private equity company buying a controlling stake, not only they would be able to change management, if they were successful Mubadala would recoup a lot of their investments.

Maybe Qualcomm should pony up some $$s to solve their capacity problems (not serious). Seriously, I agree, the only way to make GF work is with a more experienced team - but that would mean heavy involvement with a foundry or IDM, and I can't imagine anyone who would benefit.

 

[ShintaiDK]

Maybe Qualcomm should pony up some $$s to solve their capacity problems (not serious). Seriously, I agree, the only way to make GF work is with a more experienced team - but that would mean heavy involvement with a foundry or IDM, and I can't imagine anyone who would benefit.

Perhaps, Samsun is surely in headwind in terms of their component/foundry divisions.

http://www.neowin.net/news/nokia-may-drop-samsung-as-parts-supplier

First Apple, now Nokia also drops Samsung.

 

[SocketF]

I agree, the only way to make GF work is with a more experienced team - but that would mean heavy involvement with a foundry or IDM, and I can't imagine anyone who would benefit.

I remember the golden days when AMD made profit and the Dresden fab got prices like "Fab of the year" several times. Analysts all the time complained about the high labor costs in Germany but AMD told them all the time that they are well worth the money (and hey, CPUs were great, the fab got prices, seems that was correct).

Now after lots of trouble with several bosses in Dresden Chinese Management styles moved in. I guess the situation is still better than at some random picked sweat shop fab in mainland China ... but I guess that the good engineers looked for a better job in the mean time.

 

[SocketF]

GloFo IPO

I googled, there is alredy an arab-newspaper article about it, maybe in 2015:

GlobalFoundries, the microchip maker owned by the Abu Dhabi-based Advanced Technology Investment Company (Atic), will consider an initial public offering in 2015, when it expects to reach profitability.

http://www.thenational.ae/business/technology/abu-dhabi-microchip-maker-eyes-4-5bn-in-revenues

 

[ShintaiDK]

I remember the golden days when AMD made profit and the Dresden fab got prices like "Fab of the year" several times. Analysts all the time complained about the high labor costs in Germany but AMD told them all the time that they are well worth the money (and hey, CPUs were great, the fab got prices, seems that was correct).

Now after lots of trouble with several bosses in Dresden Chinese Management styles moved in. I guess the situation is still better than at some random picked sweat shop fab in mainland China ... but I guess that the good engineers looked for a better job in the mean time.

If AMD was run with Chinese management, they wouldnt where they are now.

AMD on purpose delayed 65nm, basicly pushed everything in front of them in their newfound glory and arrogance with the K8. Then they spend all their money on buying a GPU maker for way beyond its worth. Only today to be worth what, 1/10th of the original price?

So what did you expect? No money for fabs, no money for designs. All burned up on the casino.

 

[SocketF]

Probably also true .. the money is gone, without the Chartered-China-style there wouldnt be any of GF

 

[AtenRa]

Therefore they would understand that pouring more money into a failure is not a strategy.

Really ??? I guess they dont give a shit about 8+ Billions then.


GLOBALFOUNDRIES Extending Fab 8 to Meet Strong Customer Demand

Malta, NY, July 24, 2012 – GLOBALFOUNDRIES today announced it is moving forward with the final construction for the extension of Module 1 at the Fab 8 campus in New York. The project will add 90,000 square feet of manufacturing capacity, bringing the total capacity for Fab 8 Module 1 to 300,000 square feet. Construction activities are scheduled to begin in August and work is expected to be completed in December 2013.

Extending the Fab 8 cleanroom is expected to increase the Fab 8 capacity to approximately 60,000 wafers per month and increase the capital budget by approximately $2.3 billion, taking the total capital budget from $4.6 billion to approximately $6.9 billion, once tools and equipment are installed.

GLOBALFOUNDRIES To Build R&D Facility in New York to Accelerate Advanced Manufacturing Technologies for Global Customers

Milpitas, Calif., January 8, 2013 — GLOBALFOUNDRIES today announced plans to build a multi-billion dollar R&D facility at its Fab 8 campus in Saratoga County, N.Y. The new Technology Development Center (TDC) is expected to play a key role in the company's strategy to develop innovative semiconductor solutions allowing customers to compete at the leading edge of technology.

The TDC will feature more than a half million square feet of flexible space to support a range of technology development and manufacturing activities, including cleanroom and laboratory space. Representing an investment of nearly $2 billion, the facility will increase the total capital investment for the Fab 8 campus to more than $8 billion. Construction of the TDC is planned to begin in early 2013 with completion targeted for late 2014. Since breaking ground on Fab 8 in 2009, GLOBALFOUNDRIES has created approximately 2,000 new direct jobs and that number is expected to grow by another 1,000 employees for a total of about 3,000 new jobs by the end of 2014.

 

[piesquared]

Probably also true .. the money is gone, without the Chartered-China-style there wouldnt be any of GF

With AMD having their hardware in all next gen consoles, Samsung, Qualcomm, ARM, Sony, MediaTek, TI, Imagination, and 10-20 more members of the HSA foundation, worlds fastest GPU, SeaMicro, i'd say you AMD shorts are going to have to look elsewhere to get rich. Do you people all have access to Nemesis' account or something??