[adrianlee]

GeForce GTX 480 and GeForce GTX 470 will be the names of the first two GPUs shipped based on our new GF100 chip!

Twitter / NVIDIA GeForce:

Share|

[MrBungle]

That is a fun fact!

[psolord]

So they jumped straight to the rebrand? That's awesome!

[sutyi]

Originally Posted by psolord

So they jumped straight to the rebrand? That's awesome!

More likely they jumped because of their own rebranding demise of renaming budget GT2xx DX10.1 parts to GT3xx. Seems like they actually did a favor for potential customers by simplifying their future line up, if that makes sense.

[Teemax]

Originally Posted by psolord

So they jumped straight to the rebrand? That's awesome!

Actually, I think it should be called a "rebranding anticipation move".
All those left over GTX275, GTX260 and GTX285 now can "safely" move to GTX3xx line. They already did this with almost the entire mobile line.

Expect a mass order of new carton boxes from NVIDIA partners...

[psolord]

Originally Posted by Teemax

Actually, I think it should be called a "rebranding anticipation move".
All those left over GTX275, GTX260 and GTX285 now can "safely" move to GTX3xx line. They already did this with almost the entire mobile line.

Expect a mass order of new carton boxes from NVIDIA partners...

Well, GT200B is a pretty good chip and the cards based on it, are pretty good cards as well.

Now if they shrink it to 40nm and call it GT200C and base their medium line up (ie GTS/X 3XX) on that, but with MUCH better prices, I am ok with that. Of course they will lack DX11, Compute Shader 5.0 and the lot but that's ok I guess. DX11 seems to be very heavy so it will need the more transistors you can throw at it, to get decent performance (that's 60fps+ at 1080P)!

Nvidia has found themselves between a rock and a hard thing. ATI 5850 sells for 220 euros where I live, while GTX 285 is easily 60-80 euros more expensive. Noone will go for that. So they need a 40nm higher clocked GTX 285, so it can reach the same performance as the ATI 5850, but even cheaper, lets say at 180 euros, since it will lack basic features.

[Teemax]

Originally Posted by psolord

Well, GT200B is a pretty good chip and the cards based on it, are pretty good cards as well.

Now if they shrink it to 40nm and call it GT200C and base their medium line up (ie GTS/X 3XX) on that, but with MUCH better prices, I am ok with that. Of course they will lack DX11, Compute Shader 5.0 and the lot but that's ok I guess. DX11 seems to be very heavy so it will need the more transistors you can throw at it, to get decent performance (that's 60fps+ at 1080P)!

Nvidia has found themselves between a rock and a hard thing. ATI 5850 sells for 220 euros where I live, while GTX 285 is easily 60-80 euros more expensive. Noone will go for that. So they need a 40nm higher clocked GTX 285, so it can reach the same performance as the ATI 5850, but even cheaper, lets say at 180 euros, since it will lack basic features.

GT200b is "good" or not is entirely depending on your point of view.

Yes, it was the fastest GPU for the majority of 2009. But is it really that much faster than the much smaller HD4890 to justify the huge die size and complex PCB? NVIDIA can shrink the GT200b to 40nm all they like, but the 512-bit memory bus and DDR3 will:
- Keep the die size big
- Keep the PCB complex
- Keep power consumption high

What possible market would they go after with those re-branded GT200b? Low resolution gamers can benefit more from a muchcheap ATI HD5770; and only the fanboys among the high rez gamers would go for these cards.


Instead of focusing their energy on rebranding and reusing old stuff, can we just move on to the new arch, NVIDIA? I know that buying new printed carton boxes are cheaper than designing a new chip, but this is getting really really old now.

[Niceone]

Well.. at least it's not the worst case scenario if they rebrand G200-generation chips as Geforce 300-series and Fermi-generation chips as Geforce 400-series. Think about what kind of mess it would be if there would be both G200 and Fermi-generation chips in Geforce 300-series.

..Of course Nvidia can rebrand G200-generation cards with Geforce 400-names..

[power666]

Originally Posted by Teemax

Actually, I think it should be called a "rebranding anticipation move".
All those left over GTX275, GTX260 and GTX285 now can "safely" move to GTX3xx line. They already did this with almost the entire mobile line.

I really don't think that there is going to be any further rebranding of the GT200/GT200b chip. The last batch of order for board partners was made in the fall of 2009 (I want to say September but I could be wrong on this date). Even then, supply of these chips at retail is dwindling. Board partners were simply expecting Fermi to be released sooner instead of keeping these cards in stock to tide things over.

Originally Posted by Teemax

GT200b is "good" or not is entirely depending on your point of view.

Yes, it was the fastest GPU for the majority of 2009. But is it really that much faster than the much smaller HD4890 to justify the huge die size and complex PCB?

For the most part, I'd answer "no". This is a moot point now as the Radeon 5870 is here and Fermi still hasn't seen the light of day.

Originally Posted by Teemax

NVIDIA can shrink the GT200b to 40nm all they like, but the 512-bit memory bus and DDR3 will:
- Keep the die size big
- Keep the PCB complex
- Keep power consumption high

For all the reasons you list are reasons I believe that nVidia won't release a 40 nm die shrink of the GT200.

However, I do think a new midrange part from nVidia is on the horizon. If I had to guess, it'ill come with 256 shaders, 256 bit wide memory bus, support both GDDR3 and GDDR5 memory and built on a 40 nm process. This would solve most of those problems listed and fill the role a 40 nm based GT200 would fit.

The real question is when nVidia will launch a new midrange chip. They've generally trailed the high end chip by 6 to 9 months but the high end Fermi card is getting close to when a midrange variant would be expected. Everything has been focused on the high end and there has been no hint to a new midrange model.

Originally Posted by Teemax

Instead of focusing their energy on rebranding and reusing old stuff, can we just move on to the new arch, NVIDIA? I know that buying new printed carton boxes are cheaper than designing a new chip, but this is getting really really old now.

I agree. Case in point: the 9800GX2 is now being rebranded: about two years after it was first launched. Sure, the 65 nm built G92 has been replaced by the 55 nm build G92b but that only means reduced power consumption and a boost in clock speeds. The feature set has simply remained the same.

[NerosDevil]

That's my beloved company, way to go Nvidia !

Originally Posted by sutyi

More likely they jumped because of their own rebranding demise of renaming budget GT2xx DX10.1 parts to GT3xx. Seems like they actually did a favor for potential customers by simplifying their future line up, if that makes sense.

Urgh well, some of GT(s/x)3xx still stuck with DX10 shits

[Teemax]

Originally Posted by power666

However, I do think a new midrange part from nVidia is on the horizon. If I had to guess, it'ill come with 256 shaders, 256 bit wide memory bus, support both GDDR3 and GDDR5 memory and built on a 40 nm process. This would solve most of those problems listed and fill the role a 40 nm based GT200 would fit.

I don't remember seeing any Evergreen card with GDDR3; so GDDR5's price is probably at the point where it makes more economic sense over GDDR3. Hopefully GDDR3 will be for the very low end only.

I hope that Fermi has more success yielding a true mid-range card than the GT200b. On paper the Fermi architecture looks quite scalable, so cutting away 2 memory controller (down to 256 bit) and maybe throwing away 1 GPC (down to 384 "cuda cores") would likely make a decent mid-range card.


But even then, that mid-range chip would probably be as big as RV870...