Sounds like it could have a very narrowly defined threshold for when to spin up the fan due to heat… or perhaps it calculates fan speed based on throughput. If the fan’s spinning up faster without more work, the temp should drop. Maybe use ATI Tools to keep the card at a specific fan speed so it won’t speed up unnecessarily.
I don’t think ATI tools would do me much good, since it’s an 8800 GT. (Sorry, I should have clarified that it wasn’t a 4850, didn’t consider that it would be taken that way.) It has to be the GPU fan, but it’s happening in such an odd set of situations as to be almost random. It seems to have no connection to what’s happening on the screen or how much it’s having to render. To wit, outdoors with maximized viewing range, no problem. In a dark cave talking to one person with a close up on their face, HUGE FAN NOISE. It’s just bizarre.
Yeah I didn’t know what 420 meant. Well, if there’s an nVidia tool to set fan speed, definitely try that, it would probably override the default smart fan settings of your card. If not then you’re boned, unless someone else on the internets figured it out.
I’m probably boned. I think I’ve taken most of the reasonable steps to try and fix it. The odd thing is, even if I don’t pause the game to let the fan quiet down, I never see any expected heat induced errors - no texture corruption, no visual artifacting - and the few times I’ve opened the case to try and figure out what the hell was going on, there’s been no noticeable heat whatsoever. There is, however, an odd blue plastic brace that runs from the bottom of the case past the GPU, and I’m wondering if it might be pressing against the metal door, transmitting fan vibrations into it and causing the intense noises.
Hell, at this point I’m clutching at straws. Worst case scenario the GPU dies and I get something a little more modern, although I’m hoping to wait until I can find a reasonably priced card to last until next year’s (presumably) new generation.
If you can open the case while it’s happening, rule out the brace by pressing your finger against it.
Otherwise… bonage.
The problem is, opening the case is done by removing the metal door that I suspect is pressing against the brace. So, it’s impossible to test. Next time I shut down, I’m going to try better securing that arm to see if it fixes things.
Could be a defective temperature sensor. I recently had to replace the Intel stock fan on my CPU because its sensor had died (yes, the main CPU temp sensor is actually inside the fan… surprised me too). Everything kept working, except that my motherboard no longer reported a CPU temperature – and the fan always ran at full speed. With a new fan temperature reporting and fan speed regulation once again worked fine.
There are some aftermarket cooling solutions for graphics cards, and installing one on your graphics card should fix this issue, assuming the temperature sensor is not build into the card itself. But that’s a lot of work and probably as expensive as getting a new card.
MotherFUCK. Well, the temperature sensor isn’t defective, I just wasn’t giving it long enough. I gave it a workout in one of the main quest-areas in FO3 with the works - lots of colored lighting, shadows, particle effects, missiles, physics, fighting, etc. It got up to 98 degrees for about three seconds, stayed at 97 degrees for a total of about five minutes, and averaged between 90 and 94 degrees. Little worried now. It’s idling at a temperature of 60 degrees, and seems to idle at 65 after I play FO3 for a while. I’ve taken the case apart, blown out dust everywhere I can - can’t really get inside, as the card has a screwed/bolted on heat shroud that the fan is integral to.
I’ve been doing some reading, though, and it would seem to indicate that the card is known to run hot, and can handle these temperatures.
“Ever since the GeForce 8800 GT/GTS series have been released everyone was flabbergasted by two things, precisely… the vast amount of bang for bucks (this card really rocks for the money you pay for it), and more unfortunate… the heat the card creates. It’s pretty safe to say that the 8800 GT & GTS with reference coolers run rather hot, very hot as we’ve seen temps over 90 degrees C when the GPU is fully utilized. The cards can handle it though, so don’t be worried about it.”
I’m years out of date in expertise on this shit - can anyone let me know what level of panic-defcon I should be at here?
I’d say you’re at threat level yellow. You may want to read up on better cooling solutions for the card. I don’t give a fuck what internet pseudo-expert suggests that any piece of consumer electronics is “designed” to handle high heat, it will lower the life of your card.
Well, right now I’m going to try increasing the fan speeds with RivaTuner. We’ll see what, if any, effect this has.
Well, I think I’ve managed to get Rivatuner set up to crank the fan up to 60% if the temperature hits 80 degrees, 80% if the temperature hits 80 degrees, and 100% if the core temp hits 91 degrees or higher. I guess we’ll see if this actually works, and what kind of a difference it makes. Frustratingly, the only real guide to setting up fan response to hardware events is out of date for the current edition of RivaTuner, but it doesn’t seem terribly complex.
That internet pseudo-expert would be you, in this case…
I have an 8800 GTS with the standard Nvidia cooler myself, and so do many other people – the 8800 is an old and very popular series. Never heard about any excessive failure rates and never had any trouble myself, despite the card being over two years old and reaching 90 or 100 degrees under load. Those temperatures are in fact normal for the design, and the stock cooler should have no trouble with them.
Of course any piece of equipment is going to fail eventually, and since quatoria’s system has no other issues I still think it’s a defective fan controller. Making it spin up sooner may be a valid solution if you can get a more even noise level, though.
Well, I’m now using rivatuner to set the fans to 40% fixed, when I fire up Fallout 3, and that seems to be enough to keep the temperature from ever creeping above 78 degrees, even under heavy load. Previously it was sitting at 29 degrees until the temperature hit 90 or so, and only then cooling at about 60%. For whatever reason, that was just not working.
Touche. Yet that design leads to a smaller margin of error when, say, the temperature sensor is faulty or the cooling system starts to get faulty. I’d rather listen to me than that other guy.