NVIDIA GF100 & Friends speculation

Sontin said:
That's a good question. I know that a GTX285 gets 45 FPS with DX10 at the same position. So the GTX480 would be only 25% faster with DX11 and without Tessellation...
Click to expand...
Would the GTX285 frame look identical?

I've got no idea if Heaven is picture-equivalent across D3D10 and 11 with tessellation off.

Jawed
 
Sontin said:
Tha
Yes yes. But we know nothing about GF100's DC performance. All this DX11 downplaying is a result of GF100 tessellation performance.
Click to expand...

where is this tessellation downplaying?? all i see is questioning if its a typical Tessellation workload and for very good reasons

http://www.beyond3d.com/content/reviews/54/9

One thing worth noting is that between 60 and 80(!)% of these primitives get culled, which doesn't strike us as terribly efficient:
Click to expand...


care to point out the downplaying in this thread of tessellation?
 
Alexko said:
Hmm I don't know why that would be, but I don't know much about processes so let's say that's the case.

NVIDIA can still clock down significantly in idle and presumably lower voltage at least a little. Running the GPU at 40% of its maximum clock and 90% of its maximum voltage would make it draw 97.2W, assuming a 300W TDP. Surely, if the cooling system can handle 300W without perforating your eardrum, it should be able to dissipate ~100W fairly silently.
Click to expand...

So what ratio of static power are you assuming? Remember, static power doesn't scale with frequency. Oh, and for a 500-600 mm^2 chip, static power is going to be pretty high.
 
rpg.314 said:
KonKort doesn't know or does the no. of ALUs on your $600 gpu depend on which side of the bed you woke up on?

Personally, I doubt if nv will launch a castrated top dog after feeding so much info about it to public. They'll sell 5 480's if they have to, but AFAICS a 512 ALU gpu should ship.
Click to expand...


That's always been my take on the matter, neither variant is going to be in particularly high supply anyway, so they might as well benefit from higher benchmarks and not run the risk of being beaten by the 5870 in some benchmarks.They can always introduce a "GTX 475" later on if necessary but so long as Nvidia can produce enough "full fat" GTX 480s to ship to all the major hardware sites, then that's what they'll do. Its not as if Nvidia don't have previous form in this area.
 
Sontin said:
Yes yes. But we know nothing about GF100's DC performance. All this DX11 downplaying is a result of GF100 tessellation performance.
Click to expand...

DX11 downplaying of GF100 due to tesselation in Unigine?

Hell, I even said way back when the demo was released that the tesselation load in Unigine was unrealistically high, and that it's purpose was more to push video cards as hard as possible in order to see how they will react.

Likewise, that the tesselation of the cobblestones was intentionally exaggerated in order to emphasize the effect of lighting (and shadow) calculations when interacting with tesselated surfaces. It also looks far more impressive and easier to see than if they had made it more like real cobblestones.

It's a demo of the engine. Everything about it is either to showcase features (with exaggeration) and stress video cards.

There's no other reason to have that amount of poly's for Doorframes (run the demo on a Dx11 card, turn on tesselation and wireframe mode and look at the doors) other than to push a video card to see how many poly's it can handle while doing tesselation.

It's VERY impressive that GF100 can do a massive number of triangles. Noone is doubting that.

What people are questioning is whether tesselation loads in actual shipping games in the next year or two are going to give GF100 enough of a leg up if the rest of it isn't much faster than Cypress.

BTW - I've been a huge advocate of tesselation ever since Dx11 was announced (not so much before that since it was proprietary), but as said before Tesselation does not equal Dx11. I'm almost as excited about what Direct Computer is going to bring to games.

I'm going to guess that Direct Compute will be far more noticeable and prevalent in near term (1-2 years) games than Tesselation. Although I personally hope for more and better uses of Tesselation with regards to world geometry as well as objects and characters.

Regards,
SB
 
Well simple subdivision isn't what makes tessellation useful. It's the displacement mapping that happens on top of that which really makes a difference. No DX11 game shipping so far has done anything of note with displacement. Apparently Metro 2033 does some of that on its monsters so we'll see how that turns out.
 
10030713431c04665107db1b1a.jpg



Which one is real :?::?:
 
Jawed said:
With tessellation off the engine is still running in D3D11 mode. Not sure why GTX480 isn't any faster than HD5870 running the apparently "slower" version of this benchmark.
Click to expand...
Er, well, we really shouldn't expect the GTX480 to be faster than ATI's hardware in every benchmark.
 
chavvdarrr said:
The one with Default clock 17634 :)
Click to expand...
Well , since we know that GF100 cards have 4 domains (Core , Text , Shader , Memory ) , it is almost sure that GPU-Z will not be able to recognize their frequencies correctly , however , we can extrapolate some of the frequencies based on the assumption that the program recognized some of them but was not able to categorize them correctly :

1-GPU-Z could have swapped Core frequency with shader frequency , that means Shader frequency is really : 1600MHz

2-GPU-Z could have swapped shader frequency with memory frequency , and that means memory frequency is 1x50 MHz .

3-GPU-Z could have swapped memory frequency with core frequency and then got confused with (Core/texture) domain , and that means the core frequency is possibly 650MHz and texture frequency is 800MHz .

4-finally , all of this could be 100% WRONG ! , and the program is juts missed up !!
 
aaronspink said:
So what ratio of static power are you assuming? Remember, static power doesn't scale with frequency. Oh, and for a 500-600 mm^2 chip, static power is going to be pretty high.
Click to expand...

I admit I neglected static power.

How high would that ratio be, in your opinion? How does leakage scale with voltage?
 
You must log in or register to reply here.

Similar threads

M
Interesting old GPU architecture
Replies
17
Views
3K
Ailuros
Ailuros
T
GPU Stability testing for Nvidia cards - PowerSweep
Replies
0
Views
443
T2098
T
J
Graphic card/CPU configuration for best memory graphics in Flight Simulator 98?
2
Replies
37
Views
5K
homerdog
homerdog
R
HDR settings uniformity is much needed
2 3 4
Replies
68
Views
9K
DavidGraham
D
D
Speculation and Rumors: Nvidia Blackwell ...
45 46 47
Replies
928
Views
103K
Kaotik
Kaotik
Back
Top