Poor performance from my Geforce FX 5900?

[MournSanity]

I bought a BFG Geforce FX 5900 from Fry's the other day. The thing rocks but I think it is not working at the level it should be. My games run fast enough but when I run benchmarks like 3dmark2001, I get a score of 14915 on my system. This seems really low, I was expecting it to be in the low 17000s with my rig. Any ideas on how I might be able to get it to work better than it does now?

BTW, this is a real FX 5900, with the 850 Mhz memory.

Oh, and another thing, I know a lot of you like to complain about how benchmarks like 3dmark2001 are synthetic and blah blah blah, but it really does accurately represent how your computer compares to others. So please don't start with that

My specs are in my sig.

 

[VIAN]

I also have a 5900, I think that your score is fine. You can look at my score and machine by clicking my machine.

 

[modedepe]

It's not that low; I wouldn't be worried.

 

[Richdog]

I get 18'000 with my rig in sig so yours seems about right.

 

[blazerazor]

Have you tried it with just 256x2 in ddr mode instead the current set up.

 

[MournSanity]

Originally posted by: blazerazor
Have you tried it with just 256x2 in ddr mode instead the current set up.

No, I haven't actually. I will go try it now. I added the 512 MB stick later on and it's made by a different company so it might have some effect on my results.

 

[jim1976]

Originally posted by: Richdog
I get 18'000 with my rig in sig so yours seems about right.

 

[wkabel23]

Try updating your drivers.

 

[MournSanity]

Originally posted by: wkabel23
Try updating your drivers.

53.03

 

[videoclone]

A few people have asked this same question hehe
When the Geforce FX5900 first came out i ran 3Dmark2001 and scored 17000 just like you wanted with a 2500+ AthlonXP at default 1.8Ghz

With the drivers 53.03 I now score 15000 in 3Dmark.

The new drivers concentrate on Pixel shader performance and DX9 speed problems that the old drivers weren?t doing very well in turn they have sacrificed overall speed for better performance in DX9 and Pixel shader 2.0 which I think is a good idea

Your old games wont run at 200FPS only 150FPS BUTTTTT

The new games with DX9 Pixel shaders will run at 50FPS and not 30FPS

So don?t complain about your 3Dmark2001 score because I?m sure when u grab that DX9 game u would rather have it running as fast as possible

Ohh and the new drivers also get a worse score in 3Dmark2003 because I used to score
6000 in 3Dmark 2003 with drivers 45.xx and now with the 53.xx I score 4800

PS: Halo runs allot better with the new drivers cause it uses heaps of pixel shaders and so does duceX2

 

[Regs]

I was scoring that much with a Barton @ 12.5 x 175. (2.21 GHz) and a 9800np, so I don't think the benchmark is to out-of-whack. I've find it likely to be because of the motherboard and bios. Even so, when it comes to gamming, 1k points on 2001 SE doesn't really mean anything. It means you likely get the same performance as others with little higher scores, or you may have a unnoticeable framerate discrepancy.

And Richdog, I like to see a link for a comparison.

 

[MournSanity]

Originally posted by: videoclone
A few people have asked this same question hehe
When the Geforce FX5900 first came out i ran 3Dmark2001 and scored 17000 just like you wanted with a 2500+ AthlonXP at default 1.8Ghz

With the drivers 53.03 I now score 15000 in 3Dmark.

The new drivers concentrate on Pixel shader performance and DX9 speed problems that the old drivers weren?t doing very well in turn they have sacrificed overall speed for better performance in DX9 and Pixel shader 2.0 which I think is a good idea

Your old games wont run at 200FPS only 150FPS BUTTTTT

The new games with DX9 Pixel shaders will run at 50FPS and not 30FPS

So don?t complain about your 3Dmark2001 score because I?m sure when u grab that DX9 game u would rather have it running as fast as possible

Ohh and the new drivers also get a worse score in 3Dmark2003 because I used to score
6000 in 3Dmark 2003 with drivers 45.xx and now with the 53.xx I score 4800

PS: Halo runs allot better with the new drivers cause it uses heaps of pixel shaders and so does duceX2

Thanks for confirming it for me, I suspected that might have been the case. I don't mind a speed decrease for an overall IQ increase, I was just worried that my 5900 was underperforming compared to other 5900s.

 

[bates550]

i also purchased a BFG fx5900 from frys. how do you tell whether the memory clock is 850mhz? its not listed under the specs.

with the updated driver, im benching at 11200, so obviously something isnt right. even before i updated, i only benched at around 13000. im running a gig of ram (dual 512 ddr), 80gig HD, p4 2.4HT, and an asus p4p800 mobo, so i dont think its my computer that's bogging it.

i read the other threads about 256bit vs. 128bit and the frys label states that its 256bit, but the back of the box reads: "Additionally, CineFX 2.0 continues to deliver the highest level of image quality and film-industry levels of precision through true 128bit precision computing..." did frys pull a fast one on me?

please help...
alex

 

[modedepe]

Use powerstrip, coolbits, or rivatuner to check the clockspeeds. And no, that does not mean fry's pulled a fast one on you.

 

[MDE]

That score is right. My BFG 5900 scores about 15,000 at stock with my Barton at 2.2 with 512MB of RAM. I get about 17,000 if I overclock the card to 450/900 (highest stable it'll go for some odd reason) and get my CPU up to 220x11 (2.42 GHz) I score in the low 17,000s.

EDIT: Didn't read the whole thread. Ignore me for now .

 

[bates550]

thanks for the info. downloaded coolbit and my card defaults at 850mhz. found out that the card was in 2d mode, so the core clock was at 300mhz. i set it to 3D performance mode, which bumped the core clock at 400mhz. benched it again at 400/850 and got 11497 =(.

i guess the card isnt complementing the rest of my comp that well. i noticed that most of you with the 14000 benchmark are running athlons.

any comments? or tips on how to fix this? i made sure everything is updated.

alex

 

[thedarkwolf]

You weren't changing the card between 3d and 2d mode. The newer Nvidia cards allow you to run the core at different speeds for 2d and 3d thats all that is. Its pretty cool cause you can turn the clock speed way down in 2d and not be able to tell a difference but the heat output will be lower. Then when you start up a game it bumps the speed back up. Memory speed however stays the same in 2d and 3d.

Bates what are your specs for the rest of you computer?

 

[bates550]

p4 2.4ghzHT
asus p4p800
1gb ram (512x2 ddr)
WD 80gb 7200rpm
winXP pro

 

[Pete]

Do you have AA or AF forced on via the drivers? IIRC, "texture sharpening" is 2xAF, in case you have it selected.

 

[Bucksnort]

Originally posted by: bates550
thanks for the info. downloaded coolbit and my card defaults at 850mhz. found out that the card was in 2d mode, so the core clock was at 300mhz. i set it to 3D performance mode, which bumped the core clock at 400mhz. benched it again at 400/850 and got 11497 =(.

i guess the card isnt complementing the rest of my comp that well. i noticed that most of you with the 14000 benchmark are running athlons.

any comments? or tips on how to fix this? i made sure everything is updated.

alex

I get 1095 in 3dmark01 with amd1900+ and 133 fsb no overclocking evga5900se. Bates, before you run 3dmark you need to set v sync to always off in the 3d panel and turn off aa and af, put them on application controls. Note you will not see the option for setting v sync in d3d or overclocking unless the extra features are unlocked by installing coolbits or something similar.

 

[bates550]

followed your instructions and just benched 13619. sounds close enough to the average ~14000. thank you very much for your help. =)