NVIDIA GF100 & Friends speculation

[entity279]

Yes but if GF100 has only 480 active SPs it sort of means that Arun was right, and nV was wrong .

 

[silent_guy]

If you read up a little bit on physics API's, you'll see a number of bits specifically put in there to avoid numerical errors.

I believe you need to be most careful for conditions where your objects are stationary. Nobody cares about an object being off by a pixel when it's flying around the room, but a not-yet-destroyed wall made of individual bricks with a single pixel movement every 10 seconds can be very annoying.

I remember reading about this on the Bullet Physics site as well as the one of Newton Dynamics. Can't find back the exact place, but this is a good indication that the problem is real.

 

[chavvdarrr]

So, are we waiting announcement of 3 cards?
470, 480 and 480Ultra?

 

[Silus]

So, are we waiting announcement of 3 cards?
470, 480 and 480Ultra?

No. GTX 470 = 448 SPs, GTX 480 = 512 SPs

Clock frequencies and TDP are still up in the air, although in terms of TDP, the more reasonable 250 is probably what should be expected, given the PSU recommendation.

If an Ultra exists, it will be a tweaked chip in the future or simply the cherry picked cores that can be highly overclocked, although with the supply constraints, I don't think that's feasible for a while.

 

[A1xLLcqAgt0qc2RyMz0y]

Gtx490

So, are we waiting announcement of 3 cards?
470, 480 and 480Ultra?

I believe that nVidia is holding back the 512SP parts to combat the upcoming AMD release of the 5990.

So my take is:

Now
GTX470=448SP 225watt vs 5850
GTX480=480SP 250watt vs 5870

Later when 5990 gets released
GTX490=512SP 295watt vs 5990

 

[Skinner]

No. GTX 470 = 448 SPs, GTX 480 = 512 SPs

Clock frequencies and TDP are still up in the air, although in terms of TDP, the more reasonable 250 is probably what should be expected, given the PSU recommendation.

If an Ultra exists, it will be a tweaked chip in the future or simply the cherry picked cores that can be highly overclocked, although with the supply constraints, I don't think that's feasible for a while.

480, 512, 480, 512 what will it be
Well, the only thing which is certain is that it's called GTX480

I'm very curious how it will perform in Metro2033 with everything to the max.

 

[CarstenS]

I'm very curious how it will perform in Metro2033 which everything to the max.

Hopefully double-digit Fps.

 

[rpg.314]

I believe that nVidia is holding back the 512SP parts to combat the upcoming AMD release of the 5990.

The only thing that is being held back is the B1 spin.

 

[PSU-failure]

Regarding power draw, it seems to be quite bad considering NV's PR statement when asked about it... "It's quiet at idle".

Quite strange since after inspection it seems GT215 doesn't have the infamous RV770 idle power issue, but answering "noise" when asked about "power" is suspect.

 

[Silus]

I believe that nVidia is holding back the 512SP parts to combat the upcoming AMD release of the 5990.

So my take is:

Now
GTX470=448SP 225watt vs 5850
GTX480=480SP 250watt vs 5870

Later when 5990 gets released
GTX490=512SP 295watt vs 5990

So, in your view, enabling a single SM, yields an increase of 45 watts ?

From 448 to 480, it's only 25 watts. But from 480 to 512, it's 45 watts

 

[Dave Baumann]

From 448 to 480, it's only 25 watts. But from 480 to 512, it's 45 watts

Other than changes in clocks that we may not know about as well, there are real technical reasons why the should be , some of which have already been suggested.

 

[A1xLLcqAgt0qc2RyMz0y]

So, in your view, enabling a single SM, yields an increase of 45 watts ?

From 448 to 480, it's only 25 watts. But from 480 to 512, it's 45 watts

No that is not my view. More likely higher clocks also.

 

[Silent_Buddha]

There's also the possibility of a GTX 480 with 480 active ALUs and then a GTX 480/512 similar to the GTX 260/216...

See how many possible combinations of 480 or 512 we can envision.

I still wouldn't be surprised at a 480 with 480 and the a 480 Ultra/PE with 512 that only the press ever see.

Then again, at this point of silly season, I would be surprised at a 480 with 512 with ultra low quantities just to make sure it does as well as possible in preview benchmarks. And no 480 SP part to make sure PR doesn't bite them in the arse. All the while hoping B1 allows significant quantity of 512 SP parts.

At this point I think it's ALL NV PR running through the numbers and possible PR implications before they settle on final numbers to ship to retail. Do they do a 512 low quantity launch? Do they do a 480 SP launch with greater quantity? Do they do a 512 and 480 launch to win benchmarks but then use the 480 for volume without benching? Do they want better PR or better return? And if quantities are as low as rumored, then returns may not even matter so back to a 512 low quantity launch for PR, and supply be damned.

Regards,
SB

 

[Kaotik]

This doesn't sound too promising
First this:

(yes, it's GTX480 in the box there)

and then this shortly after

voi pojat... YouTube - Nelly

"Voi pojat..." = "Oh boy(s)..."

 

[ShaidarHaran]

Oh noes! A flagship part that produces heat! Unacceptable!

 

[mboeller]

So, in your view, enabling a single SM, yields an increase of 45 watts ?

From 448 to 480, it's only 25 watts. But from 480 to 512, it's 45 watts

layman answer: powerwall

see CPUs for a comparison

 

[Picao84]

This doesn't sound too promising
First this:

(yes, it's GTX480 in the box there)

and then this shortly after


"Voi pojat..." = "Oh boy(s)..."

Have you thought he might mean "getting hot in here" like: Hey! I have a Fermi! And not really talking about the card temperature? Yes, we all *might* know Fermi is hot, but that is not reason to say "oh this doesn't sound too promising"... Damn fear mongers!

 

[Silus]

Other than changes in clocks that we may not know about as well, there are real technical reasons why the should be , some of which have already been suggested.

Which is why that post I quoted is absurd at this point now isn't it ?

 

[Vincent]

Just in case refresh your memory years ago.

http://www.youtube.com/watch?v=WOVjZqC1AE4

http://www.youtube.com/watch?v=lmYrYcDp07Q

 

[KimB]

I believe you need to be most careful for conditions where your objects are stationary. Nobody cares about an object being off by a pixel when it's flying around the room, but a not-yet-destroyed wall made of individual bricks with a single pixel movement every 10 seconds can be very annoying.

I remember reading about this on the Bullet Physics site as well as the one of Newton Dynamics. Can't find back the exact place, but this is a good indication that the problem is real.

Yes, basically it's a problem that can only reasonably arise in the case of an object that is allowed to oscillate in place for a while. A chain attached to a wall or ceiling, for instance. My suspicion with the Crysis "magic board" is that they may simulate some oscillation modes across the board's surface that allow the board to bend, and if those vibrations get perturbed enough, you could easily get strange-looking behavior as it bounces off of things.

In any case, if nVidia chooses to allow support for double-precision in PhysX, any problems related to numerical errors that exist in the current implementations should simply disappear. This isn't to say that double precision is completely devoid of numerical errors, but rather that it's enough of an improvement that the few situations in which it occurs for single-precision are likely to disappear entirely with the move to double.

Now, since the situations where physics simulations break down are relatively few, devs may choose to just ignore the potential of using double precision due to the 2x performance hit. I don't know.