EETimes: ST plans for Dresden FDSOI production

[Abwx]

Actually I think you have it backward.

Indeed.
My wording is quite confusing i reckon.

We count from the smaller to the biggest numbers
and the current usage to put the most significant
number backward in respect of how we write , from
left to right , is due to the decimal system notation
that was left unchanged when translating scientific
books from arabic in the meedle ages , yet the muslims
authors were rightfully logic since they write from right
to left , thus they did start from the least significant
to the most significant number , in the same direction
as the writings.

 

[mrmt]

LOL, me doing my job was a disaster Thankfully my employer didn't see it that way until all the checks were cashed. Me and Ruiz, twins separated at birth

What's your pet hedge fund for insider trading? And what did your wife say when you bought that *huge* business opportunities for 10 times your current income?

 

[SocketF]

How credible would be a "below 50%" yield for the Bulldozer CPUs@32nm SOI? Heard it from a friend of a friend who knows somebody ... not sure what I should make of it and no clue how much below it is.

 

[Abwx]

Unlikely , the chip has enormous built in redundancy circuitry
to iMPROVE yields , just thinbk that is 30% bigger than a Trinity
while having the same transistor budget.

 

[ShintaiDK]

How credible would be a "below 50%" yield for the Bulldozer CPUs@32nm SOI? Heard it from a friend of a friend who knows somebody ... not sure what I should make of it and no clue how much below it is.

Not credible at all.

 

[Phynaz]

Unlikely , the chip has enormous built in redundancy circuitry
to iMPROVE yields , just thinbk that is 30% bigger than a Trinity
while having the same transistor budget.

Redundancy uses transistor budget. Your argument is invalid.

 

[Abwx]

Redundancy uses transistor budget. Your argument is invalid.

I m talking of the functionnal transistors budget.

My argument is valid , it s your understanding that is short...

 

[SocketF]

Unlikely , the chip has enormous built in redundancy circuitry
to iMPROVE yields , just thinbk that is 30% bigger than a Trinity
while having the same transistor budget.

That's nothing special, the GPU-parts of Trinity are packed much denser than the CPU-part, hence an Orochi die is totally seen less dense packed than a Trinity die. If you compare both dies' Bulldozer modules however, there are no differences.

Not credible at all.

Ok, but why should you be more credible? Do you also claim to have an inside source?

 

[mrmt]

Ok, but why should you be more credible? Do you also claim to have an inside source?

GLF reached critical performance milestones since 2011 and because of that it was entitled to receive extra-payments from AMD and was able to charge per wafer, and not per good die as was the case when 32nm was introduced. AMD's problem with Bulldozer isn't availability, but inability to sell chips, to the point it had to pay GLF to not manufacture the thing.

All in all there is nothing pointing for bad yields.

 

[AtenRa]

GLF reached critical performance milestones since 2011 and because of that it was entitled to receive extra-payments from AMD and was able to charge per wafer, and not per good die as was the case when 32nm was introduced. AMD's problem with Bulldozer isn't availability, but inability to sell chips, to the point it had to pay GLF to not manufacture the thing.

All in all there is nothing pointing for bad yields.

Bulldozer/Piledriver sales went up from Q1 to Q2 to Q3 across all three segments (Quads, X6 and X8).

http://www.3dcenter.org/news/genauere-zahlen-zum-absatz-einzelnen-amd-prozessorenarchitekturen

 

[ShintaiDK]

Bulldozer/Piledriver sales went up from Q1 to Q2 to Q3 across all three segments (Quads, X6 and X8).

http://www.3dcenter.org/news/genauere-zahlen-zum-absatz-einzelnen-amd-prozessorenarchitekturen

Because they stopped selling old. AMDs shipment went from almost 19 million in Q1 to nearly 13 million in Q4.

And 127000 8 core chips in Q3.

Not to mention amount of Bobcat chip sales increased even as the total amount of CPUs shipped dropped. So where does that leave the BD/PD uarchs?

 

[AtenRa]

Because they stopped selling old. AMDs shipment went from almost 19 million in Q1 to nearly 13 million in Q4.

And 127000 8 core chips in Q3.

Not to mention amount of Bobcat chip sales increased even as the total amount of CPUs shipped dropped. So where does that leave the BD/PD uarchs?

What Bobcat has to do with the fact that BD/PD sales are rising across all quarters ??

 

[Tuna-Fish]

What Bobcat has to do with the fact that BD/PD sales are rising across all quarters ??

When you start from 0, any growth looks pretty nice. Based on your numbers, AMD replaced ~13M sold (non-bobcat) cpus, mostly phenoms, with ~9.4M sold cpus, mostly BD. BD is selling more than yesterday, but it's selling a lot, lot worse than what AMD was selling before they had BD.

Looking at the BD growth alone is misleading, because when you plan to replace all your sales with BD, it paints a picture of "BD grew by 7M, it must be awesome", when what really happened was that AMD lost 4M chip sales compared to what they had before.

The biggest problem AMD has right now that they are not able to sell chips.

 

[mrmt]

Bulldozer/Piledriver sales went up from Q1 to Q2 to Q3 across all three segments (Quads, X6 and X8).

It's easy for you to add up Bulldozer/Trinity sales numbers on that table and verify that BD/PD sales are going up, but you cannot add the total numbers because they are shrinking, much less see that Bulldozer isn't replacing K10.x sales 1:1. This info is all there, in the same table you quoted, and on top of that conveniently forgetting that AMD paid GLF to not manufacture chips for them on the following quarter.

This is your already habitué show of intellectual dishonesty in plain sight for everyone to see and quite frankly, I'm full of it, spare me if you can next time. But if the urge for another show is too much for you to resist, just try to be a little more competent when twisting the numbers.

 

[Olikan]

how this thread became a discussion from AMD financials ?

 

[AtenRa]

When you start from 0, any growth looks pretty nice. Based on your numbers, AMD replaced ~13M sold (non-bobcat) cpus, mostly phenoms, with ~9.4M sold cpus, mostly BD. BD is selling more than yesterday, but it's selling a lot, lot worse than what AMD was selling before they had BD.

Looking at the BD growth alone is misleading, because when you plan to replace all your sales with BD, it paints a picture of "BD grew by 7M, it must be awesome", when what really happened was that AMD lost 4M chip sales compared to what they had before.

The biggest problem AMD has right now that they are not able to sell chips.

Bulldozer/Piledriver 8C and 6C replaced Phenom X6
BD/PD 4C and Llano/Trinity replaced Phenom/Athlon X4, X3 and X2

So again,

BD/PD 8C/6C sales are rising and they are almost the same volume in Q3 2012 as Phenom X6 was in Q1 2011.

 

[AtenRa]

It's easy for you to add up Bulldozer/Trinity sales numbers on that table and verify that BD/PD sales are going up, but you cannot add the total numbers because they are shrinking, much less see that Bulldozer isn't replacing K10.x sales 1:1. This info is all there, in the same table you quoted, and on top of that conveniently forgetting that AMD paid GLF to not manufacture chips for them on the following quarter.

Af course BD/PD doesnt replace 1:1 the Phenom, it wasnt meant to in the first place. That was the job for the APUs, Llano and Trinity. BD/PD was only ment to replace the high end Phenom X6 and it actually have as the volume of both BD/PD C6 and C8 are almost the same a Phenom X6 was in Q1 2011.

This is your already habitué show of intellectual dishonesty in plain sight for everyone to see and quite frankly, I'm full of it, spare me if you can next time. But if the urge for another show is too much for you to resist, just try to be a little more competent when twisting the numbers.

First learn to understand what CPU model replaces what and then talk about dishonesty.
BD/PD never meant to be a high volume product, it was only meant to replace the Phenom II X6 at the High End segment and it seams it did managed to have the same volume in Q3 2012.
The APUs, both small core (BobCat) and big core (Llano/Trinity) were meant to replace everything else including Phenom II/Athlon II X4, X3 and X2.

 

[mrmt]

First learn to understand what CPU model replaces what and then talk about dishonesty.
BD/PD never meant to be a high volume product, it was only meant to replace the Phenom II X6 at the High End segment and it seams it did managed to have the same volume in Q3 2012.

Yes, you *are* dishonest. Definitely dishonest. You are conveniently "forgetting" the Phenom II X4 that could not be replaced by APU, and had volumes *far* higher than X6 ever reached, and also saying that AMD decided destroy their value proposition by replacing a solid 6-core processor with an anemic 2-module processors that comprises the bulk of Bulldozer/Piledriver sales.

Enough of you, I'm putting you in my ignore list. I have nothing to say to someone as dishonest as you.

 

[grimpr]

how this thread became a discussion from AMD financials ?

We have an attack from financial speculators and stock market players, just put them on your ignore list and all problems are solved, you now know who they are and they're definetely not pc entusiasts.

 

[AtenRa]

Yes, you *are* dishonest. Definitely dishonest. You are conveniently "forgetting" the Phenom II X4 that could not be replaced by APU, and had volumes *far* higher than X6 ever reached, and also saying that AMD decided destroy their value proposition by replacing a solid 6-core processor with an anemic 2-module processors that comprises the bulk of Bulldozer/Piledriver sales.

Enough of you, I'm putting you in my ignore list. I have nothing to say to someone as dishonest as you.

Only someone ignorant like you would believe that AMD would replace Phenom II X4 (Deneb) 258mm2 die with Bulldozer 315mm2 die. You seriously dont have a clue of what you are talking about.

Bulldozer 6C and 8C replaced Phenom II X6 (Thuban) 346mm2 at the Hign-End segment.

ps: next time youll personally attack/call people i will report you.

http://forums.anandtech.com/showthread.php?t=2192303

1) No trolling, flaming or personally attacking members. Deftly attacking ideas and backing up arguments with facts is acceptable and encouraged. Attacking other members personally and purposefully causing trouble with no motive other than to upset the crowd is not allowed.

 

[Ajay]

I hate to unsubscribe from a thread that has some excellent technical financial info around FD-SOI and semicon manufacturing in general. But the S/N is just getting lower and lower.

 

[SocketF]

I hate to unsubscribe from a thread that has some excellent technical financial info around FD-SOI and semicon manufacturing in general. But the S/N is just getting lower and lower.

Yes it is getting worse

I try to post something on topic:

STE got their FD-SOI ARM chip now up to 3.0 GHz:

http://www.stericsson.com/press_releases/L8580_demo.jsp

No big news, they still don't say anything about the power-draw at that speed, but scaling seems good. 3 GHz is not that bad for a 8-11 stage pipeline ;-)

 

[mrmt]

No big news, they still don't say anything about the power-draw at that speed, but scaling seems good. 3 GHz is not that bad for a 8-11 stage pipeline ;-)

Are there any phones on the pipeline that will use the chip?

 

[Xpage]

I'd rather have a 1.5ghz chip that gives me more battery life imho.

 

[SocketF]

Are there any phones on the pipeline that will use the chip?

Not that I know of. If there would be anybody then I would bet on Sony and some Xperia model. They are using currently one STE-chip in the low-budget Xperia U.

The problem with STE is, that STM is not paying for them any longer. That means that they will be bankrupt soon ... some articles mentioned summer.

So I guess their 3 GHz chip is a means of advertisement to get an investor.

I'd rather have a 1.5ghz chip that gives me more battery life imho.

Depends ... that STE chip only has 2 cores. 4 cores @1.5 like most high-end phones have it now wont be much better.