OK but pushing 55nm GT200 as fast as possible will mean lower costs for NVIDIA and possibly a little faster clock. All these should make 'a new' GTX280 more competitive to HD48xx/4870x2 than current 65nm revisions.
Nvidia GT200b rumours and speculation thread
- Thread starter nicolasb
- Start date
OK but pushing 55nm GT200 as fast as possible will mean lower costs for NVIDIA and possibly a little faster clock. All these should make 'a new' GTX280 more competitive to HD48xx/4870x2 than current 65nm revisions.
Ehh.. I don't think the mentioned GT200 55nm, but ofcourse they already had G92b in production..
But with no announcements at nvision it could be a bit more than a die shrink and take of coure (probably as they know they can't take the performance crown back with just a shrink and it's maybe 10% clock increase).
But with no announcements at nvision it could be a bit more than a die shrink and take of coure (probably as they know they can't take the performance crown back with just a shrink and it's maybe 10% clock increase).
Well, i don`t think NVIDIA wants to take performance crown back with GT200B. The only thing what it has to do is release GPU which will be more competitive (price/performance) for HD4870 and HD4850x2. So that means GT200B doesn`t have to bring significant performance bump but more likely better price. Performance equal to GT200@65nm with about 100$ lower price should be great. When all these thing will come true of course.
It could mean many things.
That'd be about the smartest thing, Nvidia could do atm apart from actually rushing out a single-chip card that's beating the X2s hands down - and we all know what the chances for this are...
Silent_Buddha
Legend
The problem with that is what Carl B is pointing out.
Nvidia finds itself in the corner. They have a part that is selling much lower than they expected. Probably low enough that they are making little to no money on it. Nvidia isn't in a position to squeeze it's AIBs right now for fear they'll lose more exclusive AIBs.
Added to that is the fact that it's not selling well. Thus creating a situation where supply is, oddly enough considering the rumored low yields, much greater than demand.
Thus Nvidia is in a situation where it needs a part that remains competitive but is cheaper to produce, but at the same time it has an inventory of existing chips it has to clear out.
What do they do? If they bring out a competitive chip that is cheaper to manufacture and sells for the same price point they an get some margins back although it won't necessarily help with market share.
A result of that is they either have to continue selling existing stock at it's current prices or reduce the price to move it quickly.
In the first case, they erode the capability of the new chip to make money and it doesn't really move the old chips.
In the second case, they can now make much better margins or see a profit with the new chips, however they'll take massive losses on existing inventory. This might not be a bad thing however, as inventory that is just sitting there not selling is costing them money every day it sits there.
Nvidia isn't in a very enviable position right now. Facing the prospect of having to write off whatever inventory (and they make it sound like a lot) of G8x, G9x, and G2xx in order to introduce a new chip that is still competitive from a price/performance standpoint that is cheaper to produce.
I'm positive Nvidia will come out of this just fine. But in the short term, 1-2 years I'd expect, it's going to hurt them a bit financially. Whether it has the same impact as the situation ATI had with R600 remains to be seen. None of the products can be seen in as low a light as R600, yet ATI had far less inventory of R600 to deal with (when it released the Rv670) than Nvidia has stock of G8x, G9x, G2xx (when it releases whatever the next chip is).
Regards,
SB
Not sure if this is news but HardOCP claims the new 216SP GTX 260 is a 55nm part...
HardOCP: BFGTech GeForce GTX 260 OCX MAXCORE Review
HardOCP: BFGTech GeForce GTX 260 OCX MAXCORE Review
I'm gonna go ahead and call shens on that. There hasn't been any news about a 55nm GT2xx-derived part coming back from the last respin yet.
We've measured really low power consumption on our sample - the BIOS readout revealed, that it might be at least in part due to the voltages, which where at the lowest end of possible choices with 1.05v for 2D and 1.06v for 3D and "Extra". This in spite of the model being a heavily OCed AMP²-Version from Zotac.
http://www.pcgameshardware.com/aid,660194/Reviews/PCGH_Review_Nvidia_Geforce_GTX_260_with_216_ALUs/
http://www.pcgameshardware.com/aid,660194/Reviews/PCGH_Review_Nvidia_Geforce_GTX_260_with_216_ALUs/
Why hasn't anyone taken the HSF off to see if it is GT200 or something else?
Cookie Monster
Newcomer
Brent at [H]OCP has already confirmed with BFG that its based on the old 65nm. Its weird how BFG referes to G200 as D10U.
Cookie Monster
Newcomer
Its just that the term is rarely used around these parts.
Anyway getting back to topic, any news on G200B?
If the elsa roadmap is correct, surely the G206 = G200B?
Anyway getting back to topic, any news on G200B?
If the elsa roadmap is correct, surely the G206 = G200B?
GT200 refresh on Oct 22 according to these guys, not sure how credible it is.
http://www.hardware-infos.com/news.php?news=2406
http://www.hardware-infos.com/news.php?news=2406
Looks like they claim same core clocks but a bump to 1.5GHz shader domain clock. Hmm...
Hopefully that's accompanied by a price reduction else I will continue to be uninterested.
Seems that it was an wrong info...
