3DMark03 PS2.0 IQ comparison: R300(6307) vs NV31(4300/4345)
- Thread starter Zephyr
- Start date
Interesting, though I don't know the significance of some of the differences I was able to see (by d/l'ing the images and flipping thru them using IRFanView).
1. 43.00 has some very visible errors in the wooden base and background.
2. 43.00 and 43.75 seem to render the animals the same, while ATi shows a noticably "grittier" texture.
3. 43.00 and ATi show the same dynamic range, far greater than that of 43.75 (just look at the base).
4. Does the ATi card have AA on? The animal outlines look less jagged than with the FX.
5. Why do ATi cards render those white dots on these IQ tests? My Xpert 128 did the same with the zebra or whatever in 3DM01SE.
Zephyr, I'm guessing you meant INT when you said FX?
1. 43.00 has some very visible errors in the wooden base and background.
2. 43.00 and 43.75 seem to render the animals the same, while ATi shows a noticably "grittier" texture.
3. 43.00 and ATi show the same dynamic range, far greater than that of 43.75 (just look at the base).
4. Does the ATi card have AA on? The animal outlines look less jagged than with the FX.
5. Why do ATi cards render those white dots on these IQ tests? My Xpert 128 did the same with the zebra or whatever in 3DM01SE.
Zephyr, I'm guessing you meant INT when you said FX?
Pete said:Interesting, though I don't know the significance of some of the differences I was able to see (by d/l'ing the images and flipping thru them using IRFanView).
1. 43.00 has some very visible errors in the wooden base and background.
2. 43.00 and 43.75 seem to render the animals the same, while ATi shows a noticably "grittier" texture.
3. 43.00 and ATi show the same dynamic range, far greater than that of 43.75 (just look at the base).
4. Does the ATi card have AA on? The animal outlines look less jagged than with the FX.
5. Why do ATi cards render those white dots on these IQ tests? My Xpert 128 did the same with the zebra or whatever in 3DM01SE.
Zephyr, I'm guessing you meant INT when you said FX?
Terrible
yes, i dont notice that my r300 card has both AA(2x) and AF(2xQ) on R300 only can provide 4 bit subpixel precision.
FX = fixed point
NocturnDragon
Regular
43.00 has some very visible errors in the wooden base and background. But the color of the wooden base is more similar to the ati screenshot.
43.75 The background and the animal are almost identical to the ATI ones, but the wooden base has different lighting
ATI looks to render it in the right way to me, but for the white pixels here and there.
Could someone provide a screen shot without AA and AF?
Edit: change 43.75 with 43.45 typo
43.75 The background and the animal are almost identical to the ATI ones, but the wooden base has different lighting
ATI looks to render it in the right way to me, but for the white pixels here and there.
Could someone provide a screen shot without AA and AF?
Edit: change 43.75 with 43.45 typo
NocturnDragon
Regular
andypski said:
Oh yeah i supposed so.. that's why i asked if someone could send a screenshot without AA.
Thanks for the answer
NocturnDragon
Regular
I uploaded the animated gif with the reference image
OpenGL guy
Veteran
Good luck "playing" 3D Mark 2003jjayb said:Sorry, the reference pic and the ATI pic are wrong. Everyone know's the Nvidia pics show it "the way it's meant to be played".
OK, the NV31 with 43.00 drivers obviously has some errors that results in cracks between the polygons in the base. I have been flipping fast between the picture of NV31 with 43.45 drivers and the R300 picture.
The animals are exactly the same for the eye except:
1) Some pixels at the outline are different for each card possibly because of a slightly different subpixel detection.
2) A small difference in the texturing of the smaller spot at the butt of the elephant.
3) A small difference in the texturing of a spot on the left of the left ear of the rhino.
Ofcourse the wooden plate is not correctly lighted and does vary quite big in texturing.
However, the animals are almost indistinguishable for the eye when flipping between the images fast. So if nVidia is cheating, then they are doing it very well.
The animals are exactly the same for the eye except:
1) Some pixels at the outline are different for each card possibly because of a slightly different subpixel detection.
2) A small difference in the texturing of the smaller spot at the butt of the elephant.
3) A small difference in the texturing of a spot on the left of the left ear of the rhino.
Ofcourse the wooden plate is not correctly lighted and does vary quite big in texturing.
However, the animals are almost indistinguishable for the eye when flipping between the images fast. So if nVidia is cheating, then they are doing it very well.
Mintmaster
Veteran
Does this sound reasonable?
ATI: FP24
43.00: FP16, black dots due to 10-bit mantissa and its poor precision
43.45: I12, low dynamic range due to lack of exponent, but higher precision than FP16
Reference: FP32 because that's what CPU's use
ATI: FP24
43.00: FP16, black dots due to 10-bit mantissa and its poor precision
43.45: I12, low dynamic range due to lack of exponent, but higher precision than FP16
Reference: FP32 because that's what CPU's use