voeding is het zeker ni , kheb ier wa ligge zoeke op google en het is echt ongelooflijk hoeveel 7900gt's problemen vertonen , vooral in fear. Hier is interessante post van Evgfa forum:
"Hello Everyone,
I've been reading with interest the posts on the EVGA forums regarding the large number of faulty 7900 GT cards. I work for a system manufacturer in the UK and we've had a large number of failed cards. We've so far tried Asus, Galaxy and Forsa (an OEM brand) stock 7900 GT cards. It also seems from the large number of posts on the net that XFX and EVGA are affected and probably a bunch of other manufacturers.
And guess what we've managed to break nearly all ( 30 plus cards and counting ) of them doing nothing more than running them on a continuous 3DMark03 (build 3.6.0) loop for around 24-48 hours. Unlike most of the people posting to this forum we have the facilities to test a large number of different set ups. We run everything with standard settings. Clean install of XP with latest service pack and updates. Latest official WHQL nForce 4 AMD chipset drivers (6.7) and latest official WHQL Forceware release 80 graphics drivers (84.21).
In short everything stock, so nothing that should cause any problems. We've so far tested with the following hardware.
PSUs: Enermax 600W Noisetaker (36A across the two 12V rails), Enermax Liberty 620W, Hiper 525W, Hiper Type R 580W. All but the 525W PSUs are SLI certified.
CPU: AMD Athlon 3500, 4000, 4400, 4600, 4800, FX57, FX60
Motherboard: Asus A8N-SLI standard, SE, Premium and DFI Lanparty SLI-DR
The problem is consistant across all set ups, and very similar to alot of the problems on these forums and others have been describing. We get the display flashing on and off when exiting a 3D application, crashes and lockups in FEAR etc. One thing we have noticed is that OpenGL games are not as badly affected as Direct3D games. If you're in doubt whether you have the problem, the easiest way we've found is to test in 3DMark03, at first you'll get mild artifacts in Battle For Proxycon test then after a time if you esc out of the benchmark, it gives you the flashing display where the desktop isn't redrawn correctly, only a reboot will fix it.
We are fairly certain it isn't a power issue, or a particular motherboard. It is almost certainly a fault with the 7900 GT cards themselves. What we have noticed is that all the boards we have tested use the reference PCB design and have the same serial number designation (which probably means they all came out of the same factory). I would be interested to hear if anyone out there with a Gainward card ( which appears to not use the reference layout) has had similar problems.
There have been numerous posts regarding the very hot voltage regulation ICs, which may or may not be the problem as I've seen several posts now with people who have working cards and have added extra ramsinks to cool them only to have the problem occur again.
We are currently working with Galaxy to get an answer to this question, and if we get any news I'll post again. Just thought I'd say something so people can see that it is a real problem and not one exclusive to EVGA or any other manufacturer. One telling statistic, try typing in this to google "7900 GT artifact" and then try "7800 GT artifact". Given the fact that the 7800s are 10 months old, it surprised me how many hits you get with the 7900."
echt wel schandalig dat de makers het probleem goe genoeg kennen maar er zich niet echt iets van aantrekke

wat bij mij nu wel werkt is mijn memory underclocken