ELSA hints GT206 and GT212

What? I thought that the NVIO's existence was purely to increase yeilds, again based on die-size.
Actually, it's to remove process steps and therefore very slightly reduce the wafer price (while also reducing risk a bit). It barely even saves any space on the main die...
 
Well there's candidate dies per wafer, for one. Statistics are on my side here. You can play the "depends on what the meaning of is, is" game, but we all know I'm right.
Obviously (we do) not (know that or take it for a fact).

You responded to

Yields rates are fine (as in as expected) but ASPs are not.

with

Not what I've heard.

So please - if you could stop going on about statistics for a second - tell us, where - from what you claim to have heard, GT200s yields are lower than Nv expected them to be.


--
All I'm saying is that GT200 is a huge chip, and can't be giving them great yields on a 65nm process.
Actually, that was not what you were saying above. ;)


edit:
Sorry if that sounds a bit harsh.
 
Last edited by a moderator:
Do you have information that indicates otherwise?

Maybe.

Why cancel a project that's already taped-out for one that hasn't?
If it has been cancelled, how do you know at what stage exactly it has been cancelled?

There is a part due out next month, and it's not GT212.
Who said anything about GT212 within 2008 anyway?

That leaves GT200b. Unless GT200b = GT206, in which case your attitude is wholly uncalled for.
Who exactly has an attitude here? I asked a simple question and if you should have any weird sensitivities on simple questions then the problem is all yours aside with attitude problem for which I suggest you buy a couple of mirrors if you shouldn't have any.

A change in internal naming convention isn't exactly information the public would be privy to, now is it?
If (no make that IF) GT206 replaces GT200b then its hardly "just" a name change.

***edit: almost forgot...

Did they know "up front" what the die size would be?

Most certainly yes.
 
512-Bit + GDDR5, while a little core clock increase through 55nm? :???:

If GT200B is existence, it is more likely just an optical shrink around 470mm², to get a bit higher margins over GT200A and likely there are no big changes in clocks or memory.

But GT206 looks more like a G92-like performance part, with some redesign(32SPs-TPCs, no-DP) and smaller MC, with high possibility to GDDR5.
 
I wonder whether the GT206 will be faster than a GTX280. At least the posts in nvnews forums indicate so.

Who said, that they are talking about GT206?;)
Maybe it is GT200B, with a slightly clock upgrade (~650/1500MHz; GT200-400 which is listed since some months in the driver), which will be offered ~350-400€, while I would see GT206 below this, with the possible 16 ROPs, 48 TMUs, 192SPs and a ~ 200-250€ MSRP and a 4870 like performance ratio, since ALU:Tex would be increased in this scenario.
 
Who said, that they are talking about GT206?;)
Maybe it is GT200B, with a slightly clock upgrade (~650/1500MHz; GT200-400 which is listed since some months in the driver)
I suppose the GT200b has been cancelled in favor of other chips (GT206, GT212). The GTX260 refresh would have been a good opportunity for a die shrink.
 
All right. So what?
;)
I'm 99% certain that GT206 has 256-bit bus. So it's not a GT200b which is a simple shrink of GT200.
Plus it makes no sense at all to use GDDR5 with 512-bit bus for GT200-level GPU. I'd say that GT200 as it is has too much bandwidth, no need to increase it even further.
So GT206 is either 8 24/8 TPCs or 6 32/8 TPCs with 256-bit GDDR5 and has nothing to do with GT200b. I think it's the first one.
 
I'm 99% certain that GT206 has 256-bit bus. So it's not a GT200b which is a simple shrink of GT200.
Plus it makes no sense at all to use GDDR5 with 512-bit bus for GT200-level GPU. I'd say that GT200 as it is has too much bandwidth, no need to increase it even further.
So GT206 is either 8 24/8 TPCs or 6 32/8 TPCs with 256-bit GDDR5 and has nothing to do with GT200b. I think it's the first one.

OK.
;)
 

Why even post that?

If it has been cancelled, how do you know at what stage exactly it has been cancelled?

Well unless I were an insider I wouldn't *know* this, and if I were an insider I wouldn't be sharing this. I'm speculating. Which is the purpose of this thread, last I checked.

Who said anything about GT212 within 2008 anyway?

There are three rumored successive parts to GT200, GT212 being one of them. I was simply ruling out the possibility of it coming out this year so we could narrow the field down to GT200b or GT206 being next month's mystery part.

Who exactly has an attitude here? I asked a simple question and if you should have any weird sensitivities on simple questions then the problem is all yours aside with attitude problem for which I suggest you buy a couple of mirrors if you shouldn't have any.

Your tone is unmistakable.

I hardly think the following was an appropriate response:


Certainly not a useful post. If you have something to share, share it.

If (no make that IF) GT206 replaces GT200b then its hardly "just" a name change.

How would any outsider know this? We don't even know what GT200b/GT206 are, nor when they were first designed. For all we know, GT200b was the original plan, but perhaps they've decided to use only numeric codenames with the GT200 series, hence my change in nomenclature comments.

***edit: almost forgot...



Most certainly yes.

Well certainly they had a budget to work within, I was trying to be more specific though.
 
Well unless I were an insider I wouldn't *know* this, and if I were an insider I wouldn't be sharing this. I'm speculating. Which is the purpose of this thread, last I checked.

I'm not aware of any chip that has reached what is usually known under "final tape out" and has been cancelled, but I might be wrong also.

Your tone is unmistakable.

Can you stick to the point and stop being so senstitive about something that wasn't even intended to be offensive? Unless you equal someone giving a one liner as offensive as killing your dog or your mother.

I hardly think the following was an appropriate response:

I don't see it breaking any TOS of the board, nor being inappropriate. As I said you obviously have issues I cannot and will not solve.

Certainly not a useful post. If you have something to share, share it.

Or maybe I'm speculating with a bit more common sense.

How would any outsider know this? We don't even know what GT200b/GT206 are, nor when they were first designed. For all we know, GT200b was the original plan, but perhaps they've decided to use only numeric codenames with the GT200 series, hence my change in nomenclature comments.

Something like GT200b should have been on shelves for quite some time now and something like along the line of the 216SP/GTS but on 55nm doesn't really cut the cake either as we've seen. Either NV changed its plans on relative short notice or whatever is to come at 55nm is going to be as boring as that thing above on 65nm. And pardon me if you would had bothered reading the thread from the beginning I have already suggested what I would consider an interesting possibility for something like a "GT206" and in extension a "GT212" next year, which might be just based on a theory that might or might not make sense and might or might not reflect reality.

Well certainly they had a budget to work within, I was trying to be more specific though.

Any IHV that cannot predict a chip's die size before it's done, is IMHLO doomed. There might be minor mistakes in their "estimates" but these are negligable. For GT200 NVIDIA had given a quite long time before its release a die size to some of its partners and it was too exact on the last square millimeter to be an "estimate" after all.
 
Something like GT200b should have been on shelves for quite some time now

No. GT200 launched in June. Your statement has no historical precedent to back it up.

and something like along the line of the 216SP/GTS but on 55nm doesn't really cut the cake either as we've seen. Either NV changed its plans on relative short notice or whatever is to come at 55nm is going to be as boring as that thing above on 65nm.

Sounds exactly like what I've said in this thread and others.
 
I think he is referring to the GT200b in late August/early September rumor on Fud/NordicHardware/Inq (and many others).

My point is that never before has such a thing occurred. Ship a high-end chip, shrink it, and re-release 2-3 months later?
 
No. GT200 launched in June. Your statement has no historical precedent to back it up.

GT200's launch had been postponed. Its been said over and over again that there had been endless internal negotiations whether to release it or not. A simple shrink doesn't take that long after all; unless of course their sitting again on another finished design and are breading on it.

Sounds exactly like what I've said in this thread and others.

Meaning in other words that a GT200b@55nm wouldn't be the wisest move to make after all.

Even Fudo seems to "think" along that line:

http://www.fudzilla.com/index.php?option=com_content&task=view&id=9597&Itemid=1
 
Back
Top