weird scores for ati video cards with ut2004?

[nemesismk2]

Powercolor X800 Pro 256MB GDDR3 review

"The graph does not contain errors even though it looks otherwise. The framerate did indeed improve when going from 1024x768 to 1280x1024. I checked it several times and it did it every time. This happens because the CPU has become a severe bottleneck, even at the reasonably high resolution of 1024x768. I think it?s also very impressive that the framerate remains at 106 - the same as it was on 1024x768, when going as high as 1600x1200. By adding AA/AF the framerate did take a hit, but remained perfectly playable (even with the maximum settings)."

fx5950 ultra scores:-

1024x768 = 109
1280x1024 = 105
1600x1200 = 95

x800 pro scores:-

1024x768 = 106
1280x1024 = 111 *
1600x1200 = 106

9800 xt scores:-

1024x768 = 106
1280x1024 = 109 *
1600x1200 = 86

If the x800 pro was cpu limited as the reviewer mentions then the 1280 score would be the same the 1024 and 1600 scores. Does anyone know what could be causing this odd result?

 

[ForceCalibur]

I dont see whats odd. THe fact that they still use such an old efficient engine that hardly even stresses old 5 year old graphics cards for benchmarking purposes on these high end cards is odd. Its like using Counterstrike to benchmark an X800XT and an X800Pro.

The engine is great and efficient, but it doesn't do anything to help evaluate modern day video card performances. Now if every single game that comes out will look and play as good as UT2k4 with such minimal system and GPU requirements, then this benchmark will be fine. But as it is, games like Farcry, Halo, Unreal 3, etc will be ruthless on the video card.

 

[CaiNaM]

Originally posted by: ForceCalibur
I dont see whats odd. THe fact that they still use such an old efficient engine that hardly even stresses old 5 year old graphics cards for benchmarking purposes on these high end cards is odd. Its like using Counterstrike to benchmark an X800XT and an X800Pro.

The engine is great and efficient, but it doesn't do anything to help evaluate modern day video card performances. Now if every single game that comes out will look and play as good as UT2k4 with such minimal system and GPU requirements, then this benchmark will be fine. But as it is, games like Farcry, Halo, Unreal 3, etc will be ruthless on the video card.

it's consistent with what i get:

1024x768 123.26 fps botmatch (dm-rankin: 173.08 fps, as-convoy: 73.66 fps, br-colossus: 123.03 fps )
1280x1024 127.25 fps botmatch (dm-rankin: 190.01 fps, as-convoy: 70.00 fps, br-colossus: 121.75 fps )
1600x1024 123.09 fps botmatch (dm-rankin: 172.75 fps, as-convoy: 73.51 fps, br-colossus: 123.01 fps )

there's always going to be a bit of a "margin of error" from each run.

 

[nemesismk2]

Originally posted by: ForceCalibur
I dont see whats odd. THe fact that they still use such an old efficient engine that hardly even stresses old 5 year old graphics cards for benchmarking purposes on these high end cards is odd. Its like using Counterstrike to benchmark an X800XT and an X800Pro.

The engine is great and efficient, but it doesn't do anything to help evaluate modern day video card performances. Now if every single game that comes out will look and play as good as UT2k4 with such minimal system and GPU requirements, then this benchmark will be fine. But as it is, games like Farcry, Halo, Unreal 3, etc will be ruthless on the video card.

Well that was a useless answer to my question wasn't it!

 

[nemesismk2]

Originally posted by: CaiNaM

Originally posted by: ForceCalibur
I dont see whats odd. THe fact that they still use such an old efficient engine that hardly even stresses old 5 year old graphics cards for benchmarking purposes on these high end cards is odd. Its like using Counterstrike to benchmark an X800XT and an X800Pro.

The engine is great and efficient, but it doesn't do anything to help evaluate modern day video card performances. Now if every single game that comes out will look and play as good as UT2k4 with such minimal system and GPU requirements, then this benchmark will be fine. But as it is, games like Farcry, Halo, Unreal 3, etc will be ruthless on the video card.

it's consistent with what i get:

1024x768 123.26 fps botmatch (dm-rankin: 173.08 fps, as-convoy: 73.66 fps, br-colossus: 123.03 fps )
1280x1024 127.25 fps botmatch (dm-rankin: 190.01 fps, as-convoy: 70.00 fps, br-colossus: 121.75 fps )
1600x1024 123.09 fps botmatch (dm-rankin: 172.75 fps, as-convoy: 73.51 fps, br-colossus: 123.01 fps )

there's always going to be a bit of a "margin of error" from each run.

I think that increase is a pretty big margin of error don't you think? Your scores for 1024 and 1600 are almost identical, why aren't your scores for 1280 the same?

 

[CaiNaM]

Originally posted by: nemesismk2

Originally posted by: CaiNaM

Originally posted by: ForceCalibur
I dont see whats odd. THe fact that they still use such an old efficient engine that hardly even stresses old 5 year old graphics cards for benchmarking purposes on these high end cards is odd. Its like using Counterstrike to benchmark an X800XT and an X800Pro.

The engine is great and efficient, but it doesn't do anything to help evaluate modern day video card performances. Now if every single game that comes out will look and play as good as UT2k4 with such minimal system and GPU requirements, then this benchmark will be fine. But as it is, games like Farcry, Halo, Unreal 3, etc will be ruthless on the video card.

it's consistent with what i get:

1024x768 123.26 fps botmatch (dm-rankin: 173.08 fps, as-convoy: 73.66 fps, br-colossus: 123.03 fps )
1280x1024 127.25 fps botmatch (dm-rankin: 190.01 fps, as-convoy: 70.00 fps, br-colossus: 121.75 fps )
1600x1024 123.09 fps botmatch (dm-rankin: 172.75 fps, as-convoy: 73.51 fps, br-colossus: 123.01 fps )

there's always going to be a bit of a "margin of error" from each run.

I think that increase is a pretty big margin of error don't you think? Your scores for 1024 and 1600 are almost identical, why aren't your scores for 1280 the same?

? 4 fps @ 120+? that's well within a reasonable margin of error...

 

[LTC8K6]

1600 only is cpu limited? It would be higher with a faster cpu?

 

[nemesismk2]

Originally posted by: CaiNaM

Originally posted by: nemesismk2

Originally posted by: CaiNaM

Originally posted by: ForceCalibur
I dont see whats odd. THe fact that they still use such an old efficient engine that hardly even stresses old 5 year old graphics cards for benchmarking purposes on these high end cards is odd. Its like using Counterstrike to benchmark an X800XT and an X800Pro.

The engine is great and efficient, but it doesn't do anything to help evaluate modern day video card performances. Now if every single game that comes out will look and play as good as UT2k4 with such minimal system and GPU requirements, then this benchmark will be fine. But as it is, games like Farcry, Halo, Unreal 3, etc will be ruthless on the video card.

it's consistent with what i get:

1024x768 123.26 fps botmatch (dm-rankin: 173.08 fps, as-convoy: 73.66 fps, br-colossus: 123.03 fps )
1280x1024 127.25 fps botmatch (dm-rankin: 190.01 fps, as-convoy: 70.00 fps, br-colossus: 121.75 fps )
1600x1024 123.09 fps botmatch (dm-rankin: 172.75 fps, as-convoy: 73.51 fps, br-colossus: 123.01 fps )

there's always going to be a bit of a "margin of error" from each run.

I think that increase is a pretty big margin of error don't you think? Your scores for 1024 and 1600 are almost identical, why aren't your scores for 1280 the same?

? 4 fps @ 120+? that's well within a reasonable margin of error...

No your dm-rankin scores of 173fps at 1024 and 190fps at 1280!

 

[CaiNaM]

well, i said my findings were consistent with the reveiw you referenced - 1280 was slightly faster. could have something to do with the aspect ratio being different... also could have something to do with that level being particularly "easy" to render?

 

[nemesismk2]

Originally posted by: CaiNaM
well, i said my findings were consistent with the reveiw you referenced - 1280 was slightly faster. could have something to do with the aspect ratio being different... also could have something to do with that level being particularly "easy" to render?

Even if the level is easy to render it doesn't explain why the 1280 results are different to the 1024 and 1600 results when they are almost the same.

 

[CaiNaM]

Originally posted by: nemesismk2

Originally posted by: CaiNaM
well, i said my findings were consistent with the reveiw you referenced - 1280 was slightly faster. could have something to do with the aspect ratio being different... also could have something to do with that level being particularly "easy" to render?

Even if the level is easy to render it doesn't explain why the 1280 results are different to the 1024 and 1600 results when they are almost the same.

...1280 was slightly faster. could have something to do with the aspect ratio being different.

frankly tho you seem rather obsessed with this issue, so i'll leave you to contemplate it.

 

[CaiNaM]

Originally posted by: LTC8K6
1600 only is cpu limited? It would be higher with a faster cpu?

yea..

btw, that was with no aa/af.

even at 6xaa/16xaf ut2004 is cpu limited in my pc:

max detail no aa/af
1024x768 123.26 fps botmatch (dm-rankin: 173.08 fps, as-convoy: 73.66 fps, br-colossus: 123.03 fps )
1280x1024 127.25 fps botmatch (dm-rankin: 190.01 fps, as-convoy: 70.00 fps, br-colossus: 121.75 fps )
1600x1024 123.09 fps botmatch (dm-rankin: 172.75 fps, as-convoy: 73.51 fps, br-colossus: 123.01 fps )

max detail 4xaa/8xaf
1024x768 121.69 fps botmatch (dm-rankin: 169.33 fps, as-convoy: 73.56 fps, br-colossus: 122.18 fps )
1280x1024 126.12 fps botmatch (dm-rankin: 187.71 fps, as-convoy: 69.94 fps, br-colossus: 120.70 fps )
1600x1024 122.77 fps botmatch (dm-rankin: 172.66 fps, as-convoy: 73.56 fps, br-colossus: 122.09 fps )

max detail 6xaa/16xaf
1024x768 121.82 fps botmatch (dm-rankin: 170.19 fps, as-convoy: 73.37 fps, br-colossus: 121.90 fps )
1280x1024 116.58 fps botmatch (dm-rankin: 163.51 fps, as-convoy: 69.71 fps, br-colossus: 116.52 fps )
1600x1024 122.84 fps botmatch (dm-rankin: 172.54 fps, as-convoy: 73.48 fps, br-colossus: 122.50 fps )

tho interestingly dm-rankin takes a bit of a hit with 6xaa/16xaf... that'll give nemisis something else to mull over

 

[LTC8K6]

Why couldn't it be CPU limited at 1600? Maybe it's just a coincidence that it slowed down to the same speeds as 1024. Somebody overclock their CPU and see what happens.

 

[nemesismk2]

Originally posted by: CaiNaM

Originally posted by: nemesismk2

Originally posted by: CaiNaM
well, i said my findings were consistent with the reveiw you referenced - 1280 was slightly faster. could have something to do with the aspect ratio being different... also could have something to do with that level being particularly "easy" to render?

Even if the level is easy to render it doesn't explain why the 1280 results are different to the 1024 and 1600 results when they are almost the same.

...1280 was slightly faster. could have something to do with the aspect ratio being different.

frankly tho you seem rather obsessed with this issue, so i'll leave you to contemplate it.

Being obsessed is a negative, I am a very curious person so I would like to find out why!