Supposed futuremark score on 6600

Zcull schemes are more efficient at higher resolutions since polys span more pixels, so fill relative fill-rate req's can drop.
 
But z compression also works better at higher resolutions, so I really don't think it makes much difference.
 
Whether it makes a difference in higher resolutios and to what degree we'll have to wait and see.

Logically the majority of the 3dmark03 score should come from GT1/GT4 for a 6600.
 
That is a nice overclock, almost 20%. Wonder how much the 6600 300MHz version overclocks ? Maybe it will be as fast as a 9800 after all in no AA/AF situations ? 6800's look even worse now.

Looks like Doom3 / HL / Far Cry will be playable at 1600x1200, not bad for a midrange card at all.
 
dizietsma said:
Maybe it will be as fast as a 9800 after all in no AA/AF situations ?

I would think that it's going to be faster then the 9800 in no AA/AF situations since it's clocked higher and is faster/MHz. AA is another thing though but it'll be interesting to see how it does in supposedly shader limited games like Half Life 2.
 
I think he was talking about the 6600n not the GT ;) I think it's possible if the OC is really good, otherwise, noway.
 
Looks like Doom3 / HL / Far Cry will be playable at 1600x1200, not bad for a midrange card at all.

If you meant HL2, I don't think all three games will be really playable in that resolution with maxed details, even if AA/AF are entirely switched off.
 
Just theoretically... Is it possible that 6600 has two texturing units per pipe?
Fillrate numbers seem funny and GT1 is relatively low...

edit: or what causes this?

Gt6600
Fill Rate (Single / Multi)
1747.6 / 3892.5 (more than double increase)

Radeon x600
1239.3 1973.0 (~60% increase)
 
That would be, well, very odd, and memory bandwidth constraints could easily cause those numbers (remember it's only got a 128-bit bus).
 
Mendel said:
Xmas said:
No, it's just that NV43 can only output 4 pixels/clock (8 Z/stencil)

so the 8 pixel pipe talk was what exactly?
It's true, it has 8 pixel pipes, or rather 2 quad pipes like those in NV40. But it only has 1/4 the ROPs of NV40, which affects its peak output.
 
Xmas said:
Mendel said:
Xmas said:
No, it's just that NV43 can only output 4 pixels/clock (8 Z/stencil)

so the 8 pixel pipe talk was what exactly?
It's true, it has 8 pixel pipes, or rather 2 quad pipes like those in NV40. But it only has 1/4 the ROPs of NV40, which affects its peak output.

I don't understand. 1/4 the ROPs of NV40 would mean 8 ROPs; with 8 pixel pipes it should be able to output 8 pixels/clock (in theory). should be limited to 4 pixels/clock only in 2xAA mode.

?

rant: ROPs in nv40 are integrated in pixel pipeline. I really don't see any reason for Nvidia to redesign the pixel pipeline for NV4x series. one of the main reason they went this gen with 16x1 was that this architecture goes down the line very easy. they could make 2 quads and one quad chips with minimum redesign. maybe for NV43 they cut down early Z rejection units?
 
Back
Top