OC'd 648/1000 8800GTS compared to 2900XT

[dnottis]

Since this is Xtremesystems and I wanted to put this information out for some feedback... I thought you guys might like this.

My 8800GTS has been flashed to 648/1000, comparing to stock 2900XT and some OC'd 2900XT @ 860/980. Overclocking the GTS made a huge difference, the 2900XT gained little and as you can see the 2900XT got pretty well smacked down.

GRRR - I'll update this when I get home, I've got the 860/980 2900XT completed this is the old revision that was missing most of the overclocked data. The results were un-impressing.

At 1280x1024 resolution:
http://3dxtreme.net/other/ontap/ATI%202900XT/Capture.jpg

At 1680x1050 resolution:
http://3dxtreme.net/other/ontap/ATI%202900XT/Capture%201680x1050.jpg


Lemme know what you guys think.

[zanzabar]

why are u benching them at 4x aa only if u would have used 2x the ati card would have won, and if u look at the img quality a game running 2x on low and med resolution will look better on the ati card than an nvidia card (especially the g80 cards) that has 4x aa


and try bioshock in dx10 with msaa witch is what the ati card is meant to be used with


what hardware profile did use for the texture filter in the ccc, if u use the quality setting (the default) it pre filters all textures resulting in a 10-18% loss in fps, but way better img quality

[dnottis]

why are u benching them at 4x aa only if u would have used 2x the ati card would have won, and if u look at the img quality a game running 2x on low and med resolution will look better on the ati card than an nvidia card (especially the g80 cards) that has 4x aa


and try bioshock in dx10 with msaa witch is what the ati card is meant to be used with


what hardware profile did use for the texture filter in the ccc, if u use the quality setting (the default) it pre filters all textures resulting in a 10-18% loss in fps, but way better img quality



First of all there is no AA in DX10 with Bioshock, only DX9 with 8800 cards.
Why would anyone run 2x AA? I always run 4x AA, never 2x. I used High Quality for both cards btw.

[CandymanCan]

Hell i dont even use 4x AA i use 16xAA lol

[binormalkilla]

I don't know man, this conflicts with all of the latest reviews using new drivers.....in Bioshock the 2900xt only loses to the GTX by 5-15 FPS according to this review:
http://www.digitalreport.net/content/view/52/29/
Far Cry is WAY off....I thought I remembered seeing better results so I googled it:
The Radeon HD 2900XT was 10fps faster than the GeForce 8800 GTS (640MB), while it was just 7fps slower than the GeForce 8800 GTX. Therefore once again on a performance vs. price ratio the new Radeon does very well.
http://www.techspot.com/articles-info/52/bench/FarCry.png
http://www.techspot.com/review/52-asus-radeon-hd-2900xt/page4.html

[Daveb2012]

I would like to see more COD2 benches, this also conflicts with the benches I found that just made me order a x2900xt to replace my 8800gts 320 OC'd:

Gaming: Maximum Quality, 1280x1024
Game
AMD ATI Radeon
HD 2900 XT
NVIDIA GeForce
8800 GTS 640MB
Call of Duty 2
73.5 FPS
56.7 FPS
Company of Heroes
92.1 FPS
90.1 FPS
F.E.A.R. 84.0 FPS
83.3 FPS
Half Life 2: Episode 1 112.0 FPS 57.4 FPS *
Oblivion
47.9 FPS
39.5 FPS
3DMark06
11447
9836

this is from http://www.dailytech.com/ATI+Radeon+HD+2900+XT+Performance+Benchmarks/article7043.htm

[dnottis]

I don't know man, this conflicts with all of the latest reviews using new drivers.....in Bioshock the 2900xt only loses to the GTX by 5-15 FPS according to this review:
http://www.digitalreport.net/content/view/52/29/
Far Cry is WAY off....I thought I remembered seeing better results so I googled it:


Got the updated charts up. Ya, you are looking at NON AA results...non AA the 2900 XT is another beast, once you enable it the card starts to drop significantly. My results without AA are spot on with that...they didnt even test with AA on... geez

[dnottis]

I would like to see more COD2 benches, this also conflicts with the benches I found that just made me order a x2900xt to replace my 8800gts 320 OC'd:

Gaming: Maximum Quality, 1280x1024
Game
AMD ATI Radeon
HD 2900 XT
NVIDIA GeForce
8800 GTS 640MB
Call of Duty 2
73.5 FPS
56.7 FPS
Company of Heroes
92.1 FPS
90.1 FPS
F.E.A.R. 84.0 FPS
83.3 FPS
Half Life 2: Episode 1 112.0 FPS 57.4 FPS *
Oblivion
47.9 FPS
39.5 FPS
3DMark06
11447
9836

this is from http://www.dailytech.com/ATI+Radeon+HD+2900+XT+Performance+Benchmarks/article7043.htm


Keep in mind my GTS has been flashed to 648/2000/ shader 1512. Its about as fast as a GTX now, slightly faster at lower res's.

[Leeghoofd]

If you get these powerfull cards I don't see the use in playing with 2x AA...

Good job Dnottis, there are many contradictory reports on these matters, I kindda like it simple as you did it....

For Bioshock it all will be done in the drivers... too early to judge await the full game release and proper drivers from both companies involved...

[ChaosMinionX]

Why on the first set is the 2900XT using 5-5-5-15 settings at DDR2 900.... that right there is a major hit to performance. When the 8800 is using 5-4-4-12, and what subtimings and trfc are you using for each? that can have an effect on overall fps as well. And if you had Static DRAM read enabled, or another sort of ram booster on for one, and not the other. But its a good start, gaming scores on XP seem lower than Vista performance for me with the HD2900XT 1GB, I dont have an Nvidia card to compare to though :)

[binormalkilla]

Got the updated charts up. Ya, you are looking at NON AA results...non AA the 2900 XT is another beast, once you enable it the card starts to drop significantly. My results without AA are spot on with that...they didnt even test with AA on... geez

Hmmm....yea there used to be quite a huge raping performance hit with AA at high res, but for me the 7.8 Catalysts fixed that. Hopefully they are still working on it though. AA won't work in Bioshock either.

[dnottis]

Why on the first set is the 2900XT using 5-5-5-15 settings at DDR2 900.... that right there is a major hit to performance. When the 8800 is using 5-4-4-12, and what subtimings and trfc are you using for each? that can have an effect on overall fps as well. And if you had Static DRAM read enabled, or another sort of ram booster on for one, and not the other. But its a good start, gaming scores on XP seem lower than Vista performance for me with the HD2900XT 1GB, I dont have an Nvidia card to compare to though :)


The timings were the same, that was just a typo from a previous review I was working on in the specs.

Same ram, same timings, same OC used throughout. Nothing in the system changed but the GPU.

[dnottis]

Actually, I had run all those benchmarks with 7.8 Cats, the bioshock scores are with hotfix (7.9 beta) offered by ATI.

[purecain]

basically the hd2900xt is cpu limited, i think you would find the hd2900 out performs the gts at higher clocks... then again your right it isnt performing as well as i thought from looking at those benches( just noticed the timing differances, not a valid comparison)..... it does scale better than the gts/gtx's in crossfire though.....

[NKD]

i play at 1920x1200, and if u want my honest opnion i never look at the edges when i am playing a game, and at that resolution i really dont think u need 4xaa, yes radeon sucks with high aa, but why would i use 4xaa at such a high resolution, i have read every review saying that at high resolutions u dont need 4xaa, heck i use 4xaa but i i dont think there is much difference between 4xaa and 2xaa at the resolution i play at,

i think ati is really working hard at AA performance hit, they have stated some games in 7.8 and i believe there is more to come, hey i have had nvidia and ati both and i am not a fanboy, 8800 is a great card and r600 is a pretty awesome card too, and i got two for 600 and i aint complaining.

[dnottis]

basically the hd2900xt is cpu limited, i think you would find the hd2900 out performs the gts at higher clocks... then again your right it isnt performing as well as i thought from looking at those benches( just noticed the timing differances, not a valid comparison)..... it does scale better than the gts/gtx's in crossfire though.....

There are no timing differences. That was a typo. Read posts above, all testing has been identical settings, timings, OC. Thanks.

[dnottis]

i play at 1920x1200, and if u want my honest opnion i never look at the edges when i am playing a game, and at that resolution i really dont think u need 4xaa, yes radeon sucks with high aa, but why would i use 4xaa at such a high resolution, i have read every review saying that at high resolutions u dont need 4xaa, heck i use 4xaa but i i dont think there is much difference between 4xaa and 2xaa at the resolution i play at,

i think ati is really working hard at AA performance hit, they have stated some games in 7.8 and i believe there is more to come, hey i have had nvidia and ati both and i am not a fanboy, 8800 is a great card and r600 is a pretty awesome card too, and i got two for 600 and i aint complaining.

I dont have a monitor that does 1920x1200. 1280x1024 and 1680x1050 were tested. I prefer to play 1280x1024 with 4xAA, I cannot stand to not have AA on even at 1680.