NV48?

[overclocked]

I noticed that in some of the latest BETA Forceware driver´s it´s support NV48 so maybe it means that nV will release it before christmas afterall, any speculation´s?

 

[Evildeus]

Or maybe it's just FUD. It would be interesting to see a 500+MHz (110 micron?) GPU with some tweaking in PS/VS

 

[overclocked]

He, na it´s there and it seem´s to be two target´s for it, maybe a "Ultra" and a GT.

 

[trinibwoy]

Or maybe it's just FUD. It would be interesting to see a 500+MHz (110 micron?) GPU with some tweaking in PS/VS

Haha it's FUD now to include support for future hardware in your BETA drivers?

 

[Razor1]

Definitly not fud, it wouldn't be there unless the card has already been tapped out and is beng tested.

 

[Guden Oden]

I'm curious... Could you describe what precisely 'tapping out' a card involves. Are they perhaps stored in barrels and left to mature for some months before first time they're plugged into a PC?

 

[MasterBaiter]

I'm curious... Could you describe what precisely 'tapping out' a card involves. Are they perhaps stored in barrels and left to mature for some months before first time they're plugged into a PC?

Ahem, I think it's slange. Like when you see a girl and say "I'd tap that".

Okay, it's not nice to make phun of someone's spelling mistake. It's just petty!

Back on-topic. With the type of stuff that Nvidia has pulled of late, it very well could be FUD, but I think it would be foolish of ATI to assume it as such. What if it isn't? Worse thing that will happen is ATI will have an ace up it's sleeve that Nvidia wouldn't , so for this reason I would say it is not FUD.

This is kinda reminding me of the part in Princess Bride where Wesley & Vizzini have their battle of the mind.

 

[Pete]

Olde schoole slange--noe, waite, newe schoole. ;P

 

[Evildeus]

Or maybe it's just FUD. It would be interesting to see a 500+MHz (110 micron?) GPU with some tweaking in PS/VS

Haha it's FUD now to include support for future hardware in your BETA drivers?

Well, it could but that's not what i think (but i did saw it somewhere else)

 

[jvd]

We will see what happens next week still can't find around the states , would have been a good gift to my sisters roomate for her 21st .

 

[overclocked]

One of the "big" questions is if nVidia will manufacture NV48 at IBM or TSMC. I don´t know if IBM has any equal to 110nm but i heard recent that they have a 100nm node if i recall correct.
If i speculate and say that nV will produce NV48 at TSMC 110nm node they could in theory add more pipelines(internal) while keeping 16ROP´s
and still have a slightly smaller core than NV40.
The NV40 AT IBM´s 130nm node is as i understand around 280mm square
and a transition to 110nm would allow them to increase the number of transistor´s to around 300million while still not have a larger core than NV40. All in theory and speculation of course.

edit: spelling

 

[MatiasZ]

The NV40 AT IBM´s 130nm node is as i understand around 280mm square
and a transition to 110nm would allow them to increase the number of transistor´s to around 300million while still not have a larger core than NV40. All in theory and speculation of course.

Wouldn't heat be a problem then? 220 million transistors running at 400Mhz are already producing some nice heat on 280mm square, wouldn't 300 million on the same heat disipation surface running at say the same speed (more pipelines*same speed=faster gpu) be outputting too much heat for being usable? I get the impression they need newer and better technologies to make their chips run cooler and be more efficient. Look at what happened to prescott for being 90nm on the same technology and speed.

 

[cho]

i think you have better find out the the details of nv47

 

[overclocked]

Wouldn't heat be a problem then? 220 million transistors running at 400Mhz are already producing some nice heat on 280mm square, wouldn't 300 million on the same heat disipation surface running at say the same speed (more pipelines*same speed=faster gpu) be outputting too much heat for being usable? I get the impression they need newer and better technologies to make their chips run cooler and be more efficient. Look at what happened to prescott for being 90nm on the same technology and speed.

Ães i assume it would but that Prescott is running hot seem´s more like
a design flaw from Intel, compare it to the 90nm Athlon64 that seem´s to
have a superb thermal profile and it´s AMD´s first revision of their 90nm
core.
Also keep in mind that GPU´s is not designed for crazy Mhz speed(relative) and would not have such big problem´s. Of course the 6800Ultra
draw´s quite a bit of power so there are problem´s already now but i think that it could be more of a engineering problem from the design team consider the XTPE don´t have this flaw´s. If we do a quick comparison and this is just a guess and say that the 130nm Low-k XTPE @520Mhz/160m transistor´s equal´s to a NV40U @400Mhz/222mt because of the R420 much higher clockspeed. I´m just guess that the
low-k and less transistor´s "advantage" of ATI´s design vanish because of the much higher speed vs the nVidia part.
Then and now i´m guess again i think that ATI´s experience in lot´s of their mobile and god know´s what gives them the experience to deliver a part with great thermal´s.

i think you have better find out the the details of nv47

What the hell is the NV47

 

[DegustatoR]

Right now it looks like NV48 is gonna be a simple overclock of NV40. There is a possibility that it's gonna be remade for 0.11 TSMC process but it's really slim right now. So i'm betting that NV48 is just a speedbined NV40 (or NV45) with 450-475MHz core and 1200-1400MHz GDDR3. So no need for a new tapeout.

The really interesting part is NV47 which is likely to be NV's spring refresh.

 

[Dave Baumann]

Note: X700/6600 discussion moved off to "X700 Launch 'n Stuff" thread.