Ever since I got my
MSI N9800GT-MD1G, I've been having TDR's (nvlddmkm stopped responding and has recovered) randomly in any game or 3D application. (even Google Earth)
Usually it happens within the first five minutes, other times it can take an hour or longer before it times out.
I've traced this problem down to the power supply's 12V rail value dropping very low while doing anything 3D. I do not believe this card should be using a lot of power, since it's a "green" card and gets all its power from the PCIE slot anyway, which is limited. I have not been able to acquire a multimeter and I understand software readings can be inaccurate. But the readings I get must mean something is wrong.
Using CPUID Hardware Monitor, these are the power readings:
Only the +12V seems to drop a lot when running something 3D, while the others are pretty stable/static.
The +12V has dropped as low as 11.01V while running FurMark.
I'm already as sure as you that it's the power supply is bad, but I just want to make sure before buying anything new, because I find it strange how the power supply only suddenly became "bad" when I installed this 9800GT.
I never had any problems (at all) with a 9600GT installed. I never looked at the power reading before though, but my PC was stable. I don't have that card anymore to test.
And just by the way, the 9600GT was not a "green" card and had an extra power connector, so surely it used more power?
I do not play very graphics intensive games. I play Guild Wars, The Sims 3, Sam & Max, Black & White 2, Burnout Paradise.
I run all games at 1440x900 since that's my monitor's native resolution.
With the graphics card removed and with onboard graphics, the +12V idles at 11.90V.
So ultimately my question is: which is the culprit, the graphics card or the power supply?
Some relevant specs in no particular order-
The power supply is a generic 450W.
Running Windows 7 Ultimate 32 bit.
Latest beta NVidia drivers. (257.15)
Biostar MCP6P-M2 motherboard.
Samsung HD322HJ hard drive.
Lite-On IHES208-08 Blu-Ray drive.
1 x Zeppelin 2GB RAM.
AMD Athlon 64 X2 5200+ Brisbane
Nothing is overclocked.
My PC is very humble, like my budget, so please be nice.
Some logical and redundant things I've tried-
Increasing the TDR timeout to 20 seconds. At a TDR event this just makes the screen freeze for 20 seconds before recovering.
Disabling TDR. At a would-be TDR event the screen freezes as above, but doesn't recover. This calls for a cold reboot.
Installing older drivers and drivers from MSI.
Updating DirectX.
Updating motherboard drivers.
Updating motherboard BIOS and graphics card BIOS. (using official utilities)
Ran Memtest86 3.5 for 16 hours and got no errors.
Ran MemtestG80 multiple times with the maximum amount of GPU memory allocatable and got no errors.
Ran Prime95 for 19 hours and got no errors.
Checked temperatures. Nothing ever goes over 60° C.
Taking out the PCI WLAN card and the Blu-Ray drive. (the only extra peripherals that are in the machine)
Cleaned out dust.
Reseated graphics card, CPU, RAM.
Various BIOS tweaks.
Underclocking the graphics card. Only after the clocks are 1/4 of the stock clocks can I play without TDR's, but the framerates are awful and it will randomly lock up and then I have to do a cold reboot. So it's not a solution either.
Clean install of Windows 7
Disabling Avira and closing other programs.
Possible solutions-
New PSU. I can get this one for a decent price, so please take a look and advise: Cooler Master eXtreme - RS-650-PCAR-E3 - 650 Watt
New graphics card. Hopefully not... But I'm thinking about an HD5770.
Notes-
Is it at all possible that the power supply is not getting enough AC power? I couldn't find anything about that on Google.
Running FurMark doesn't seem to cause a TDR, but I've not run it for longer than an hour.
Thanks for taking the time to read my post! I hope you could keep up with me.