GFFX Tests / Test Apps

[Brent]

42.69 don't have missing effects

 

[noko]

Some IQ comparisons which 3dMark03 I think would be useful. There are other features of 3dMark03 which is untouched by any other means.

 

[cho]

Some IQ comparisons which 3dMark03 I think would be useful. There are other features of 3dMark03 which is untouched by any other means.

this site tried:

http://www.gzeasy.com/nparticle.asp...%B4%BB%F7%D5%BD++GeforceFX+5800U&offset=2

 

[noko]

And what do you think about that? Good, bad?? I couldn't read it so I can't really evalute what was shown. Seeing what is being rendered with the numbers is the only way to accurately reflect the merit of each card. Without that then the comparison is pointless and I think is better not to have any comparison whats so ever.

For example, using the Texture Filering test of 3dMark03 you can show where the texture aliasing and moire problem stems from on the Radeon 9700 Pro. Also this test shows pretty well the antialiasing capability of a card. Maybe I will post some images here to show the usefullness of some of these unique tests.

 

[Neeyik]

Does anyone know if the Parhelia-512 demos will run on other cards? The Shark demo is supposed to highlight the vertex shading capabilities of the product (VS2.0 caps if I'm not mistaken). If possible, how about seeing how the DX9 cards run that demo?

 

[Hyp-X]

Does anyone know if the Parhelia-512 demos will run on other cards? The Shark demo is supposed to highlight the vertex shading capabilities of the product (VS2.0 caps if I'm not mistaken). If possible, how about seeing how the DX9 cards run that demo?

That demo uses Displacement Mapping w/ Adaptive Tesselation which isn't supported by any other card.

 

[mboeller]

Amazing;

the R9000 is nearly as fast as the R8500 in Dual-Texture mode.

R8500 4 x 2 = FFP - Dual texture - 563.419495M pixels/sec
R9000 4 x 1 = FFP - Dual texture - 472.467651M pixels/sec

So now, we only need to know the actuall speed (MHz) of each card to draw an conclusion.

Regarding to the R9200Pro;

I would really like to know the MHz-numbers of both chips tested above.

If we assume that the R9000 was the R9000Pro @ 275MHz and the R8500 was the R8500-Retail @ 275MHz too, then an R9200 would need to run at 327MHz (with 275MHz RAM) to have the same multitexturing fillrate as the R8500-Retail. So maybe, with an R9200Pro at >= 333MHz with >= 275MHz DDR-RAM the R9200Pro would really be faster then an R8500/R9100 and so the higher number would be justified.

Sidenote : as it seems the R8500 is really bandwidth limited here.

 

[Babel-17]

Matrox's reef demo on non Matrox cards



http://www.warp2search.net/article.php?sid=11113&mode=thread&order=0&thold=0

 

[2B-Maverick]

Test Application for Pixel-Shaders:
IL2-Sturmovik - Forgotten Battles (out in the US since march 3rd, in Europe march 20th)
Huge Pixel-Shader penalty when enabling "perfect" landscape settings in game options: i get a 50% fps drop in higher resolutions. Looks ver nice though

Benchmark with track playback and fraps is suggested (and very reliable - reproducable).
In game fps measurement with ">fps SHOW START" on console.

 

[psurge]

Does anyone know how aniso is implemented in the DX9 refrast?

Does it use the EWA algorithm by any chance? If so, would it be possible for someone to post a refrast images for the 3dmark03 texture filtering tests?

Regards,
Serge