Good Benchmarks but Lousy Game Performance?

[Chipa66]

I replaced my Geforce Ti4600 with a Geforce FX 5900 yesterday. I benchmarked the 4600 using 3dMark03 before I removed it and got a reading of around 1750. I installed the 5900 by totally reformatting a new hard drive and doing a clean install of Windows XP Home. I have installed and updated the Nvidia drivers, the Soundblaster drivers, and my motherboard/chipset drivers so that they are all current. I have also downloaded the most recent version of Direct X 9 from Microsoft.

Well, I benchmarked the FX5900 and got a reading of about 5100 marks or so(compared to 1750 from the Ti4600)! I was pretty ecstatic and was eager to see how smooth my games were going to run. Well to my surprise, I'm getting a real slide show at times in both HALO and Medal of Honor! I had no problems running either of these with my previous setup and with much lower benchmark scores. I have the latest patches to both games and have tried lowering the in-game settings as well as reducing lowering the values of my card at the desktop but to no avail. does anyone know how I could get such high scores in 3DMark and still run my games so poorly?

My Specs:

Alienware Area 51
P4 2.4 Ghz
Asus P4T-E Motherboard (Intel 850 chipset)
Geforce FX 5900
Soundblaster Audigy Gamer
512 Mb PC800 RDRAM (Samsung)

 

[Turbo-Athlon5]

strange, maybe some drivers or something got f*cked up when u reinstalled. i duno, do u have directx 9.0 INSTALLED? maybe ur using an old directx.

maybe the card u got is f*cked up?

sorry i wasnt much help to ya

 

[Spinox]

Benchmark is wroted to do compares as you did,
but do not expect games to run faster.
Many games is not wroted correctly to use even the
half power from the newest GPU, but more the CPU.

 

[preet2u]

Just a couple of suggestions ....

First ..Run dxdiag .. to figure out if u r having any issues with your drivers ..

If there r problems .. u could just reinstall the drivers from Nvidia ..

If that's not the case .. i think u should try reverting back to 52.16 (ie .. provided ur using 53.03 ) I found those more reliable then the 53.03.