OK, so I installed Oblivion last night and half way thru character creation, my machine blue screens, error 10000050. I know this is a general error, but the event IDs and error types point to hardware failure or device driver conflict.

I have the GeCube x1950xt. This card has been known to overheat on me in the past, but it has never generated a blue screen. I did some googling and discovered that for most people, this is a driver going into infinite loop, because of a half-failed installation, or corrupted registry entry. I have the latest Catalyst driver from ATI/AMD's website.

Before my question, system specs:

Asus A8V
Opteron 185
2g Corsair XMS DDR400
GeCube x1950xt
700w OCZ Gamexstream PSU
2x WD Raptor 36g RAID 0

I have heard thru the rumor mill that Oblivion doesn't like VIA chipsets. Can anyone confirm??

OK, my question. I thought this was heat until last night. I ran a stress test on my VPU and it didn't shut down even as high as 90c. This seems to only happen in 3d apps, and monitoring the tempratures, they never got near 90c at the blue screen.

Does that make sense to anyone?

Also, I replaced the stock cooler on this card with the Zalman GPU cooler, which is great in it's own respect. However, and this is what has my curiosity at the moment; the stock cooler had a TEC on it. Which was wired into the card. After removing it, there is an exposed "header" on the card now where nothing is connected. Do you guys think that there is some hardware trigger that is calling the driver to turn on the TEC in 3d apps and when it doesn't sense it, it just loops? Should i jumper that header? Is that even a possible cause??

Any suggestions would be nice. I am going home tonight to do a full diagnostic on my mobo/ram to remove them as variables of the crash.

This GPU has reset my machine before, but only after 6 hours of Battlefield 2142 and there was no blue screen, just a reboot. Oblivion didn't take more than 5 mins to blue screen and turn off my GPU.

As a side note, when it blue screens, my monitor stops receiving a signal from my GPU, which is normal if the GPU is shutting down. However! After the bluescreen, the CPU resets and boots normally into windows, but I still have no video. I just know it boots because I hear the lovely XP startup jingle.

Sorry for the long post, didn't want to leave any info out.

Thanks in advance!

Yes, patched it before I ran the game for the first time.

It's sometimes hard to say with oblivion, it's so hardware intensive, It uses as much raw power as it can, that I can use it to find problems when I over clock. IE if the RAM is underpowered or not able to run at the bus rate I have the FSB set on, or if the CPU is undervolted.

I would uninstall and reinstall both the drivers for the video card and then the game, and then if you still have the problem start running test on the hardware. IE Orthos, Memtest, and 3D mark.

I would have to say no on that. I would fear it's simply a power header. Jumpering them would short the power leads. This would cause your card to have a tendency towards spitting sparks.

How many prongs on the header is there?

2 Prongs.

I would have to assume it's stricly a power header then.

Question about your Opty, did you install the AMD dual core optimizer? I believe I read that the Opty's don't like it. You might try uninstalling it if you installed it. Did you try troubleshooting by dropping some of the video settings?

Didn't install the dual-core optimizer.

Yes, I turned down the graphics in the game to ultra-low settings and played fine, but looked like an 8bit nintendo game

OK, new development. I took an old monitor home and hooked it to 2nd DVI port to do some real time temp. monitoring. The GPU dies at 83c.

I know that's warm, but according to the little dial on CCC, its JUST entering the orange. Is that too hot? If so, what are my choices? This Zalman cooler seems to be working properly. Does this card *require* a TEC?

Another new development...

It seems that I see a lot of people having problems with the 1950xt running too hot. They say they adjust the core voltage using ATI Tool, but I do not see a setting to adjust the core voltage. They said lowering it to 1.22 still allows for smooth operation with no artifacts. Anyone know how to do this?

I think I stumbled upon the solution.

ATI Tool. I downloaded it, and let it handle the overclocking/voltages instead of CCC and it ran 10c cooler during Oblivion character creation. It ran steady at 72c at what seemed to be the "peak". With the CCC driver, it kept climbing rather rapidly to over 80c.

Could simply intsalling ATI Tool have this effect? Does it use different vcore voltage.

Anyone? Does installing ATI Tool somehow lower load temps?

I don't kow, ATI Tool doesn't work with my card, but if it lowered the temps and performance is good, then go with it. Is Oblivion stable now? The game is a memory/bandwidth hog. If there is a glitch in your system, Oblivion will expose it.

Oh, and my 939 system scores 9600 in 3D Mark 06.

The main thing ATItool will give you as regards to controlling overheating is better control over fan speed. The ATI defaults put quietness above heat concerns, ATItool (or Rivatuner, ATItray) can allow for much higher fan speeds thus lower temps.

Yes I have had crashes when I used to run Oblivion on an overclocked 1900xt, which went away once ATItray was installed. (The machine now has an 8800.)

If one is not overclocking the ATI defaults are generally ok for voltages.

That's the strange thing though....

I didn't overclock the card when it was crashing. Only default CCC settings were enabled. The only difference now is that ATI Tool is controlling the clock speeds. My best guess is that ATI Tool undervolts the card, generating much less heat. Several posts I read when researching said that the default voltage on the 1950xt is high for stock clocks.

I dunno what it did, but yes, everything is stable now and running at the least 10c cooler. I'll leave this one to dumb luck

Heh, maybe your card just needed some burn-in time? 8) If you can add another or larger fan to your case or something it may help your situation if you think you're having overheating issues. I had to add a new NB cooler to my motherboard, as it was getting way way too hot with the default one. It's certainly something more people should consider. Touch the coolers on your NB and SB and your CPU when you put the system under load. Just tap them because they could be hot enough to burn you a little. If it feels really hot like a stove, I'd recommend changing out coolers. Doesn't necessarily have to be fan cooled either. There's a low-profile copper cooler made by Enzotech This can make your system a whole lot more stable and not get in the way of your CPU cooler. (http://www.coolerguys.com/840556068440.html) I wish I'd bought that one instead of the one I did buy for my mATX system. (http://www.newegg.com/Product/Product.asp?Item=N82E16835701003&ATT)
It gets in the way of any wider CPU coolers because it's tall and my NB is too close to my CPU socket. Anyway, for about $20 you can get a sweet copper low-profile NB cooler that can help cool your motherboard temps and looks nice too. If your CPU cooler blows air around it it should probably be enough. That's my advice anyway.

Strange, I used my CCC to unlock my card and let it run the overclock tool, and it ran the GPU as high as 92oC, and my card gave me no problems. (I also found out that the noise that comes from the card is the fan running at medium to low speed, and it gets quiet at high speeds)