Richland & Kabini rumours

[erunion]

Yes that's exactly it. If you offer a PC builder an i5 or i7 at a price that is much closer to Pentium prices, what do you think they'll do?

Drop prices on the quads and ULV's...and omg suddenly pentiums and celerons show weakness in sales? Is this a surprise to you, I really hope not.

Very creative, but not supported by any facts.

A more likely scenario is that tablet sales have reduced demand for budget laptops the most, while higher end PCs are still safe.

 

[SiliconWars]

Very creative, but not supported by any facts.

A more likely scenario is that tablet sales have reduced demand for budget laptops the most, while higher end PCs are still safe.

It's not creative, it's business. I'm going to go out on a limb and suggest that's why I own mine and you probably don't own yours.

Just as a clue, if Intel suddenly gave my distributors a better deal in SB Core cpu's so that they were lower priced than the equivalent Trinity...I'd probably buy into that too and AMD could kiss my rosy red ass this quarter.

 

[erunion]

It's not creative, it's business. I'm going to go out on a limb and suggest that's why I own mine and you probably don't own yours.

I'll take that as an admission that its not supported by any facts and call it a day.

I was originally under the impression that you weren't familiar with Intel's Q4 report. Turns out you are just reading into it what you want to see. Good luck with that.

 

[pablo87]

Short-term, there's little AMD Can do with $300MM of wafers because PC demand and AMD MSS are more or less price inelastic. Mid term they'd have to come up with some big *** die and hope it sells...something like 315mm^2 perhaps?

Btw, employees and mgmt usually do very well coming out of Chapter 11, its one of the side effects. I still remember when WDC's market cap was $135MM coming out of Chapter 11, everybody and his brother loaded up and did very well. So the problem is not the mgmt reluctance, it's the bond holders who apply pressure not to. This bond business is very lucrative until the wheels come off, which it does with Chapter 11, then all bets are off.

 

[SiliconWars]

I'll take that as an admission that its not supported by any facts and call it a day.

I was originally under the impression that you weren't familiar with Intel's Q4 report. Turns out you are just reading into it what you want to see. Good luck with that.

If you want to ignore the obvious signs then go ahead. ASP up, inventory down, "Pentiums and Celerons weak", AMD basically forced into lowering production on Trinity.

Intel is dumping inventory as fast as it can. I'm willing to wager that ASP won't be up next quarter now that the better SB chips have been burned through. In fact, what is happening to intel right now reminds me a lot of what happens to AMD regularly. It's not pretty.

 

[Vesku]

Even with other people's money it seems weird that they wouldn't rather use the WSA forced wafers to roll out some stock ARM+GraphicsCoreNext chips or some such. They have alraedy announced plans to have an untweaked ARM out in 2014 so it's not as if they are going to be spending unplanned R&D money to work on it. Paying to not use it does seem, from my outside perspective, like they are phoning it in.

Just, imo, AMD needs to stop acting like an old school IBM. Walking in Intel's shadow hasn't really produced a profitable business.

 

[Ajay]

I'll second that motion that you will eat crow

Seriously though, AMD doesn't need chapter11 for economic reasons, not anytime within the next 2yrs at least.

I think all the discussion surrounding chapter 11 is based on the notion of taking advantage of the re-organization aspects that chapter11 legally affords and in doing so AMD could wrestle itself free from the existing protracted 2024 WSA + Take-or-Pay contract with GloFo.

Emerging from such a bankruptcy re-org, free to actually choose their foundry as a fabless entity, would be priceless at this point in time. GloFo is bleeding off what precious little resources AMD has...sucking away their future by starving their R&D pipeline today.

Where would rather see AMD's $360m Q4 write-off go? AMD R&D or GloFo as a cash transfer? At least if it disappeared into the vacuum of R&D there might be something to show for it in 3-4 yrs time.

That is why I'm suggesting that it needs to happen. And like you I'm not too worried about AMD's agreement with Intel for the reason you state. The WSA is as much an Albatross around AMD's neck as the union pay&rules structure was around GM's neck.

 

[krumme]

That is why I'm suggesting that it needs to happen. And like you I'm not too worried about AMD's agreement with Intel for the reason you state. The WSA is as much an Albatross around AMD's neck as the union pay&rules structure was around GM's neck.

Intel must laugh all the way to the bank. They have always used much money keeping AMD barely alive, at a time even Otellini had to call his good friend Michael from Dell, and now they have Mubala doing the work for them. I guess they are very happy the way it played out.

After the settlement with AMD witch allowed AMD to get x86 produces outside AMD, quickly followed by bobcat, two years later, they are in a situation with an AMD starved to death.

RR is probably stucked unless Mubadala gets tired of burning money. (read: someone cares to look at the real value of the GF assets).

One thing for sure, it cant continue as it does now. Either the new kabini/temash is going to be a smash hit beyond anything seen before, followed by a kaveri that just makes anand happy, or Mubadala is going to change something imho.

 

[Idontcare]

RR is probably stucked unless Mubadala gets tired of burning money. (read: someone cares to look at the real value of the GF assets).

Mubadala needs AMD to legitimize GloFo as a foundry. Without AMD they have no leading edge customers to showcase and entice others to GloFo.

As soon as GloFo signs on another high profile customer, say Qualcomm or Apple or Nvidia, then you can kiss Mubadala's interests in AMD goodbye.

GloFo is the national strategic interest there, not AMD. AMD is just a convenient customer for the moment because they are basically the only high profile customer for the moment. That will change, one way or the other.

 

[Abwx]

Glofo is owned by ATIC , an entity iyself owned by Abu Dhabi and whose
aim is to create an world class leading foundry , as such ATIC has an industrial
project as goal.

On the other side Mubadala is a separate entity that is also owned by Abu Dhabi and whose raison d etre is to make investments in stock markets.

 

[mrmt]

Glofo is owned by ATIC , an entity iyself owned by Abu Dhabi and whose
aim is to create an world class leading foundry , as such ATIC has an industrial
project as goal.

On the other side Mubadala is a separate entity that is also owned by Abu Dhabi and whose raison d etre is to make investments in stock markets.

ATIC is controlled by Mubadala. They are not independent entities.

 

[Abwx]

Then it fares better for AMD since Mubadala is the money castle.
I would be surprised if they favoured Glofo over AMD , both are
strategical assets.

 

[ShintaiDK]

Then it fares better for AMD since Mubadala is the money castle.
I would be surprised if they favoured Glofo over AMD , both are
strategical assets.

The asset is GloFo. They stock value in AMD is what, 300mio$? Their take in Glofo is billions. AMD is nothing but a cow for them. Keeping AMD alive for the while have payed those 300mio$ back several times.

 

[Abwx]

Mubadala has no interest seeing AMD stock returning to double digit values ,
rather to keep the stock at bottom levels since it allow them to progressively
dilute the other shareholders stakes thanks to timely fund pourring that allow
them to gradually increase their stake for peanuts.

 

[Idontcare]

Then it fares better for AMD since Mubadala is the money castle.
I would be surprised if they favoured Glofo over AMD , both are
strategical assets.

The sole motivation for Mubadala and ATIC to be involved with GloFo is to create jobs in United Arab Emirates.

Specifically to create a native semiconductor industry and leave future generations a legacy that transcends oil and fossil fuels because they have no other natural resources to speak of in terms of supporting industries that could possibly support the existing population at its current rate of economic activity.

More than 85% of the UAE's economy was based on the exports of natural resources in 2009. The UAE has tried to reduce its dependency on oil exports by diversifying the economy, particularly in the financial, tourism and construction sectors. While Abu Dhabi remained relatively conservative in its approach, Dubai, which has far smaller oil reserves, was bolder in its diversification policy.

Unless AMD has plans of establishing an IC design center in the UAE, they are of no strategic importance to the UAE, and by extension they are of no particular strategic importance to Mubadala or ATIC.

For now AMD are simply a means to an end and that is why you see GloFo having no qualms with sticking it to AMD in terms of the take-or-pay contracts and so forth. If AMD was strategic then you'd see a far more favorable contract arranged between the two businesses.

 

[Abwx]

Agree , IDC , but if i was Mubadala i would apply the strategy i hinted
in my previous post.

 

[ShintaiDK]

AMD is only on for the ride until other customers fill their place(capacity).

 

[Ajay]

The sole motivation for Mubadala and ATIC to be involved with GloFo is to create jobs in United Arab Emirates.

Specifically to create a native semiconductor industry and leave future generations a legacy that transcends oil and fossil fuels because they have no other natural resources to speak of in terms of supporting industries that could possibly support the existing population at its current rate of economic activity.



Unless AMD has plans of establishing an IC design center in the UAE, they are of no strategic importance to the UAE, and by extension they are of no particular strategic importance to Mubadala or ATIC.

For now AMD are simply a means to an end and that is why you see GloFo having no qualms with sticking it to AMD in terms of the take-or-pay contracts and so forth. If AMD was strategic then you'd see a far more favorable contract arranged between the two businesses.

And the one thing about this, that struck me as very odd, is the new NY Fab for GloFo. I can't figure out why ATIC did this, since it doesn't meet Mubadala's goals. The fact that there was available talent in NY and NY gave GloFo allot of incentives seems to be the rationale, but it doesn't advance Mubadala's vision.

The whole AMD, GF, ATIC relationship seems, at least to me, to be dysfunctional.

Since AMD is having enough problems of it's own, it would appear sensible for them walk (actually run) away from ATIC & GF. The only reason I can see AMD not doing this is that the BoD being more interested in their own success rather than AMD's or that there is some agreement between ATIC and AMD that has not been disclosed (given the deal was done with Ruiz at the wheel, I wouldn't be surprised) - once only need look at how AMD treated investors WRT the WSA. What a freaking mess.

 

[ShintaiDK]

And the one thing about this, that struck me as very odd, is the new NY Fab for GloFo. I can't figure out why ATIC did this, since it doesn't meet Mubadala's goals.

Might simply be a learning project so to say. Plus to prepare the homefront in terms of labour. The first factory there was supposed to start construction in 2012, but got delayed.

 

[piesquared]

So as per the conference call, AMD will be 6-9 months ahead of the competition with a quad core x86 SoC. Pretty significant window of opportunity there.

 

[inf64]

I hope all those console deals they have won will bring some good to their financial record when the year ends.

 

[Abwx]

Back on topic , i saw an AMD slide at Hardware.fr that show that Piledriver
cores have one of the two integer execution unit enhanced with a hardware
divider , a feature that is explicitly claimed for Kabini but so far , apart from
the said slide i ve seen nothing about it in tech forums.

Curiously , it is not mentionned in all AMD Piledriver related slides
even the ones pointing the enhancements vs BD.

Possibility is that it s present in Trinity but is still not used , wich could
explain the slight improvement expected for Richland vs Trinity assuming
there are other slight improvements.

http://forum.hardware.fr/hfr/Hardware/Processeur/unique-processeurs-steamroller-sujet_924911_1.htm

 

[CTho9305]

Back on topic , i saw an AMD slide at Hardware.fr that show that Piledriver
cores have one of the two integer execution unit enhanced with a hardware
divider , a feature that is explicitly claimed for Kabini but so far , apart from
the said slide i ve seen nothing about it in tech forums.

Curiously , it is not mentionned in all AMD Piledriver related slides
even the ones pointing the enhancements vs BD.

Possibility is that it s present in Trinity but is still not used , wich could
explain the slight improvement expected for Richland vs Trinity assuming
there are other slight improvements.

http://forum.hardware.fr/hfr/Hardware/Processeur/unique-processeurs-steamroller-sujet_924911_1.htm

Generally, they only talk about things that are new to a chip family. In the case of Trinity, Llano already had a hardware integer divider, so it isn't a new feature for mainstream APUs. On the other hand, Jaguar's predecessor, Bobcat, did not have a hardware integer divider, so Jaguar's inclusion of a hardware divider is interesting to people watching the low-power APUs. (I have no idea whether or not BD had the integer divider too, or whether it was "new" again in PD...but it was there in Llano and therefore existing APUs).

 

[Idontcare]

And the one thing about this, that struck me as very odd, is the new NY Fab for GloFo. I can't figure out why ATIC did this, since it doesn't meet Mubadala's goals. The fact that there was available talent in NY and NY gave GloFo allot of incentives seems to be the rationale, but it doesn't advance Mubadala's vision.

The whole AMD, GF, ATIC relationship seems, at least to me, to be dysfunctional.

Since AMD is having enough problems of it's own, it would appear sensible for them walk (actually run) away from ATIC & GF. The only reason I can see AMD not doing this is that the BoD being more interested in their own success rather than AMD's or that there is some agreement between ATIC and AMD that has not been disclosed (given the deal was done with Ruiz at the wheel, I wouldn't be surprised) - once only need look at how AMD treated investors WRT the WSA. What a freaking mess.

The fab in NY makes perfect sense, but you'd probably have to be an industry insider to see what is happening at GloFo to understand why they are on the path they are on (and how the NY fab factors into it).

GloFo is unacceptably dependent on IBM. That is not viable going forward. The 32nm and 28nm issues at GloFo are squarely due to IBM.

The day AMD spun-off their fabs and formed GloFo was the day a plan was set in motion to bring process node R&D back internally to GloFo and remove themselves from the fab club.

They can't do that without seriously disrupting ongoing node development if the built an R&D fab in the UAE.

But if you park it just up the road (figuratively) from where your existing fledgling R&D team is currently located then you can bring up your internal R&D efforts on node N+2 in parallel to maintaining the necessary R&D commitments at IBM on node N+1.

Then once node N+1 begins to be transferred to production you formally end the fab club contract and you move over your Fishkill contingent to the Luther Forest site to augment your N+2 node development efforts as well as to fuel N+3 development.

I was not at all surprised when the NY fab was announced, it was predictable. I am not at all surprised that they dropped the other shoe and announced the their plans to build the TDC. Once 14nm-XM has transferred to production and out of the hands of the Fishkill development team do not be surprised to see GloFo formally announce they are leaving the IBM fab club.

If you knew what is currently going on, and what has been going on, you'd know they simply have no choice. GloFo will perish if they remain dependent on IBM and the R&D team knows this, as does the management. If you think the relationship between GloFo and AMD is frosty then you should see what it is like when you put GloFo engineers in the same room as IBM engineers. :shudders:

 

[Abwx]

Generally, they only talk about things that are new to a chip family. In the case of Trinity, Llano already had a hardware integer divider, so it isn't a new feature for mainstream APUs.

On the slide they point the hardware divider as being an enhancement
in respect of bulldozer , yet it is made no mention of this in a the more
accurate slide below.

http://forum.hardware.fr/hfr/Hardware/Processeur/unique-processeurs-steamroller-sujet_924911_1.htm