Post your breaking NV30 news links here!

[Sabastian]

I interviewed one of the tech guys and assuming my tape caught it all I'll have it up as soon as I can.

Excellent.

 

[nAo]

Well, all I have to say is..my sources suck

 

[LeStoffer]

...

3.) There are no GigaPixel deferred rendering / geometry caching elements. ZCULL can reject more pixels per clock than GF4.

4.) FSAA is similar to GF4 and there's no programmable grids etc.

I interviewed one of the tech guys and assuming my tape caught it all I'll have it up as soon as I can.

Fairly important info at this point. Keep it coming Mister Baumann! 8)

 

[Bambers]

Maybe they just knocked together their own map for it.

Still, even now the game must be very buggy not the best choice for a benchmark, still this is anand who used the ut2k3 performance test for ages so its expected really.

 

[maskrider]

4.) FSAA is similar to GF4 and there's no programmable grids etc.

This point alone got me disappointed.

 

[Colourless]

Me too. Non programmable FSAA is disappointing for sure.

Plus, IMO, the Cooling solution looks like a 'hack' just to get the clock rate as high as possible. Even though its 0.13 micron, this is going to be one damn hot chip.

 

[Bambers]

This point alone got me disappointed.

I agree.

Looks like nv30 will not be on my list of possible gfx cards for a spring upgrade. I was dissappointed at the 8500s lack of decent quality AA after what was initially said about SV and I really can't believe that nvidia would stick with their old AA modes in the nv30, especially after they have been saying about higher quality pixels not more of them, obviously they were only thinking pixel/vertex shaders at that point. Maybe they're hoping that people will only look at benchmark scores.

 

[Laa-Yosh]

4.) FSAA is similar to GF4 and there's no programmable grids etc.

This is an attack point for ATi. They could claim thet their 4x FSAA is equivalent to Nvidia's 8x Ordered Grid method - that is, if they really want to resort to such weapons

 

[Joe DeFuria]

Still, even now the game must be very buggy not the best choice for a benchmark, still this is anand who used the ut2k3 performance test for ages so its expected really.

No, it's not.

At least with UT2K3, Anand actually had the program to run himself, as well as the hardware to test it on. Neither is the case here.

 

[Dave Baumann]

I'm confused at this point about the 8X as well - the guy I was talking to confirmed to me that GF FX still only has 4 z units per pipe, so exactly how they are achieving 8X I'm not 100% sure.

 

[Dave Baumann]

Oh, yeah, Ansio is stil 8X as well.

 

[demalion]

Welll, one of the previews listed gamma correction concerning AA. This is a significant improvement. ATI seems to have pretty good results with this, what else was wanted? EDIT: Ah, Rotated Grid I guess? Is it clear that this isn't offered?

What it looks like is it will offer an enhancement over ATI FSAA (8x) and it will be interesting to see how visible the difference is, and also an addition that ATI can conceivably match according to what they've said (6XS)...but it remains to be seen if they ever will. (*pokes* OpenGL Guy)

I'm interested to see how the fillrate figures work out (it sounds like they include color compression in the calculation, but perhaps I'm misunderstanding).

Also, what does "two textures per texture unit mean" (words from one of the previews I think)?

 

[alexsok]

It's not 8x1 Dave.

 

[Nebuchadnezzar]

I thing that still is a mystery to me :

For what the heck are those 125M trannies???

Since it's olny a 8x1 design, where are they used??? In the shaders???

 

[mboeller]

well, one question!

How fast would be the R300 with this cooling :

according to tomshardware.com

The demo board, which NVIDIA demonstrated in an nForce2 system, produced a lot of heat. The air coming out of the fan grille is hot to the touch. While the system was quite loud overall, we could still make out the Flow FX fan - not a very positive trait. NVIDIA has promised to refine the design to make it quieter.

Even with this cooling-solution the air, coming out of the board is hot!


PS.: according to c't the highest AA-mode is 8x OGMS for OpenGL and 6xS(?) for DX. So no improvement compared to the GF4. Even the 4xMode of the R300 should look as good as the 8xmode of the NV30.

 

[Psikotiko]

It's not 8x1 Dave.

Explain?

 

[alexsok]

Explain?

There is no longer a definition of "TMU".

 

[Nebuchadnezzar]

So mister Alexsok, can you tell us more?

EDIT: Outdated; damn you guys post 4 post in 10 seconds, too fast for me

 

[Dreadnought]

Its been mentioned already in this thread - but I was most struck by the seeming lack of power of the Nv30 vertex shader.

Based of the normalisations from Toms:

NVIDIA GeForceFX (500MHz): 0,7 Mverts / clock
ATI Radeon 9700 PRO (325 MHz): 1 Mverts / clock

Considering all we'd seen pushed by Nv about efficiency over power, and a de-emphasising of bandwidth - The 1 thing I really expected to see was a v strong showing from the vertex shader.

 

[maskrider]

Even with this cooling-solution the air, coming out of the board is hot!

I think the fan is one of the reasons for adding a power connector, this fan will certainly draw a hell lot of current. Loud is somewhat expected when you see the size of it.