Kyro II or Geforce II Pro?

[BFG10K]

You fail to understand the irony ? SS uses the OGL transformation pipeline, and as such it will use GF3?s T&L unit which is part of the videocard, not a CPU process as you stated (the majority of the work is being done on the videocard

There's no irony there at all PotNoodle. I think you don't understand what I'm saying.

If you're video card limited, any speed advantages offered by hardware T&L are negated, since the CPU has to wait for the card to finish what it's doing. In effect you have extra CPU cycles to burn. That's the exact principle of how software HSR works, and it's exactly why your claim that the Kyro2 isn't punished for not having T&L is false, since the setting you were using was video card limited.

As for your SS benchmarks, we can't assume that they're 100% accurate. Anyone can run benchmarks and post them up somewhere.

 

[PotNoodle]

?There's no irony there at all PotNoodle. I think you don't understand what I'm saying.

If you're video card limited, any speed advantages offered by hardware T&L are negated, since the CPU has to wait for the card to finish what it's doing.?

heheh ? I understand what you are trying to say, but the point being is you are saying it wrong. There is no such thing as ?Videocard limited? since the videocard, with T&L undertakes many of the tasks that is traditionally part of the CPU?s work; the words you want are ?Fillrate? or ?Geometry? limited.

What you meant to say was: ?If you are fillrate limited any speed advantages offered by hardware T&L are negated since the scene has to rendered before the next scene?s geometry can be calculated?.

However, the point I was making by mentioning T&L is that the apparent cap of 70FPS in this game on GF3 doesn?t seem that impressive ? given its an OpenGL title that should make full use of the hardware T&L on GF3 I would expect it to be getting much higher frame rates at 800x600, which as you poit out isn?t really much of a fillrate limited case for GF3.

?As for your SS benchmarks, we can't assume that they're 100% accurate. Anyone can run benchmarks and post them up somewhere.?

Well, considering the guy they originated from was from B3D I would assume they weren?t falsified. Perhaps if B3D comes back (Rev said it will) we can ask him.

 

[BFG10K]

There is no such thing as ?Videocard limited? since the videocard,

You mean all this time you've been arguing in circles because you were nitpicking over one word?
Sheesh, you're a good time waster, aren't you?

If the CPU is waiting for the video card, you're video card limited. It's as simple as that.
You can call it fillrate limited or whatever else you wish, but calling it "video card limited" is still 100% correct.

 

[PotNoodle]

?If the CPU is waiting for the video card, you're video card limited. It's as simple as that.?

Even if the CPU is doing no 3D related function, and they are all moved to the videocard?

(That?s an illustration BTW, not the reality ? but its not far from reality since all the static geometry will be performed on the videocard with a T&L unit through the OpenGL pipeline; in all likely hood the CPU is only generating the physics and the dynamic ? model ? geometry in this case, so its not likely that that GF3 is going to be CPU limited)

?You can call it fillrate limited or whatever else you wish, but calling it "video card limited" is still 100% correct.?

In the case of GF3 saying ?Videocard limited? could suggest that you are Fillrate and/or Geometry limited ? how do you make the distinction?

 

[BFG10K]

In the case of GF3 saying ?Videocard limited? could suggest that you are Fillrate and/or Geometry limited ? how do you make the distinction?

In this case there's no need to make such a distinction. The bottleneck is the video card, in some form or another. If the bottleneck was the CPU, T&L would be boosting framerates.