2900xt owners\benchmarks thread

Sumanji · 15 Jul 2007 at 13:03

Not when the average is only 61.

Klo · 15 Jul 2007 at 14:26

Sumanji said:
Not when the average is only 61.

61 fps is easily smooth enough for fluid play. However, 26 is quite low, although even then you would be hard pushed notice it if it only happened occasionally.

rickfin · 15 Jul 2007 at 15:06

After you added the black box to your steam account do you see it under my games? I can only see DOD.

killer_uk · 15 Jul 2007 at 15:08

rickfin said:
After you added the black box to your steam account do you see it under my games? I can only see DOD.

Yup, you will only see DoD until the black box games are released

rickfin · 15 Jul 2007 at 15:10

killer_uk said:
Yup, you will only see DoD until the black box games are released

Thanks for the fast answer.

Tom|Nbk · 15 Jul 2007 at 15:51

FEAR
Res: 1680x1050
AA setting 4xAA (in game)
AF setting 16xAF (in game)
In game setting maxed (except soft shadows)

Avg: 66 - Min: 25 - Max: 148

Tom|Nbk · 15 Jul 2007 at 16:03

Vogon said:
IL2 Sturmovik 1946:
Res 1920 x 1200
AA setting 4xaa 8xwide tent (control panel)
AF setting App controlled (control panel)
In game setting maxed, perfect settings with water mode 2.
Avg: 53.629 - Min: 19 - Max: 118

I am dissapointed with IL2 Sturmovik performance, I would hope that can be improved in newer drivers, it is an older title, but is pretty demanding still with all the eye candy, but those figures are not much above the X1900Pro.
Other games obviously are much faster than the X1900.

Vogon, you do realise that the Narrow tent and wide tent filters take more of a hit to performance over the traditional box filters?, If you go try that benchmark again but this time with 8xAA BOX you should find you get better performance over 8xAA Wide Tent. This would explain why the performance is some what lacking.

Vogon · 15 Jul 2007 at 16:37

Not Really Tom|NBK.
In that game at least 4aa with 8x wide tent, performs better than 8x box, for straight 8xAA I get: Avg: 43.776 - Min: 18 - Max: 92

fornowagain · 15 Jul 2007 at 16:56

The box filter is a traditional 8 sampled MSAA. The CFAA narrow and wide filters are not direct samples, but pixels from the neighboring box samples. If you use a 4xMSAA + custom filter to get to 8xCFAA it will use less resources than the direct 8xMSAA box. Its pretty clever, the card uses the shaders to combine the primary+secondary MSAA samples to give the overall CFAA product. I suppose its a bit of a disadvantage to the competitors traditional ROP resolve in shader heavy games, but then its also fully programmable unlike hardware AA.

http://www.techreport.com/reviews/2007q2/radeon-hd-2900xt/index.x?pg=6
http://www.beyond3d.com/content/reviews/16/10

Custom Filter AA, or CFAA for short, is AMD implementing non-box filters that look outside the pixel being processed in order to calculate final colour and antialias the image. The sample resolve for that is performed on the shader core, data passed in so the programmable hardware can do the math, with the filter function defined by the driver. That means AMD can implement a pluggable user filter system if they want, but even if they don't they can update, add and remove filters from the driver at will whenever they see fit.

The big advantage is the ability to perform better filtering than the standard hardware resolve. However the disadvantages include possible implementation of bad filters, and speed issues because the driver now has to issue, and the hardware run, resolve calculation threads which chew up available shader core cycles.

Tom|Nbk · 15 Jul 2007 at 20:55

Guess it depends on the type of game then as I've found Box to perform better in some.

Vogon · 16 Jul 2007 at 14:46

I really really miss the old 6x AA, bring it back ATi/AMD

Tom|Nbk · 16 Jul 2007 at 14:52

Vogon said:
I really really miss the old 6x AA, bring it back ATi/AMD

Yea it would be nice to have a 6xAA Box resolve

Tom|Nbk · 17 Jul 2007 at 00:45

Not my benchies, these were pulled from hardware france (hardware.fr) havn't seen these before some nice benchmarks of the card :cool:

. I'd link to the article but I dont think anyone speaks French

Tom|Nbk · 17 Jul 2007 at 00:45

RavenUK · 17 Jul 2007 at 01:32

Here's some more benchmark results from a recent review, all tested with 4xAA and 8XAF @ 1920-1200, not looking to bad.

http://www.overclock3d.net/reviews.php?/gpu_displays/sapphire_hd_2900_xt/1

Gerard · 17 Jul 2007 at 01:32

Be interesting to see how crysis performs as it will be the first true dx10 game. If ati have a lead in that then it could help them shift some cards.

Tom|Nbk · 17 Jul 2007 at 01:36

Cheers for those Gerard things are certainly shaping up. Let's hope the 7.7s either this week or next week wrap things up a bit. :cool:

Azza · 17 Jul 2007 at 02:02

Those drivers are sure improving performance.

Vogon · 17 Jul 2007 at 13:13

I have to say I don't know were they are coming from re the fan noise .... mine is quiet at idle unoticable over my tuniq @ 50% speed, yes when it ramps up its louder than the 8800gtx, but not bad, certainly not audible over the gaming noises.
If you manually set it to 100% its loud, but even then its the air being pushed, no actual whiney fan noise, and it would never get there in normal use tbh.

I think there is more to come with driver improvements, but have to say I am dissapointed that the drivers are not better sorted by now.

BTW does anyone else get a major slowdown when the dragon comes out of the water on 3D06 ? I am thinking maybe its only 1gb cards as I have not seen it mentioned anywere, it does it on both the drivers I have tried (4.5 and 4.6).

Tom|Nbk · 17 Jul 2007 at 13:25

As you can see with ATi's new OpenGL driver it seriously is kicking ass against the likes of the Ultra/GTX, this would also apply to Prey and Doom 3 and pretty much all other OpenGL titles :cool:

.