Graphics Card Temperatures

[prwilson]

I have an original Titan graphics card - ran GPU-Z yesterday - turns out it's running at 60C at idle with 38% fan - this seems a bit high to me.

Is it normal for the idle temperature to rise as a card gets older? Should I be worried about it? Anything I can do to eke out the lifespan a bit?

Cheers in advance.

[prwilson]

BTW I'm running a midi case, with three case fans, all other temperatures seem to be in what seem to be more normal ranges.

[i7Baby]

As TIM gets older it dries out and has to be periodically replaced. I suggest you do the same on you Titan.

It just means it could thermally throttle easily under high load.

[prwilson]

A small amount of Googling suggests TIM = Thermal Interface Material - yes, I think you're probably right, I'll give that a go.

I checked it out playing a game yesterday and the temperature went upto 81C - which is more-or-less what I would have expected. If the thermal paste was losing it's effectiveness you'd think it would be at least as apparent, if not more, when under load.

[prwilson]

I've had a problem for a while now whereby (mostly on desktop) the screen suddenly goes blank, then the desktop comes back up andI get a message saying "nVidia graphics driver has stopped responding and recovered". Everything seems fine, but video (Youtube, or embedded) runs really slowly, it's fine again if I reboot.

I've tried updating the graphics drivers, updating with clean install, completely removing the drivers using Display Driver Uninstaller and reinstalling, but the same thing still happens.

Could this be related to the high idle temperatures?

[i7Baby]

"Could this be related to the high idle temperatures?" Yes

[prwilson]

Any explanation as to why the temperatures should be excessive at idle, but not under load though?

[scaryjim]

Any explanation as to why the temperatures should be excessive at idle, but not under load though?

They shouldn't be - certainly not enough to cause crashes. I'd be surprised if a crash on desktop is heat-related if you can play games without crashes. At load most modern GPUs will adapt their clock speed to a temperature target, so it's most likely that your card is staying at its target temperature by reducing the clock speeds.

Sounds more likely a software conflict, and I'm going to point at Flash - it's a problem I've seen with a few different video cards. Virtually all software nowadays is GPU accelerated, including your desktop, but flash videos is one that tends to cause problems. Worth keeping an eye out and trying to identify if the driver crashes occur when you have videos playing in a web browser.

If so, as a temporary workaround you can tell the browser to stop using GPU acceleration. It might also be worth rolling back the driver to an earlier version - sometimes new drivers can introduce bugs, and since you're on a not-the-current-latest-super-generation card from nvidia you might find they're not paying as much attention to fixing the problems you face as they are for the GTX 10 series.

As to the high temperatures, I'm dubious about the likelihood of it being the TIM - in an older card it's far more likely to be an accumulation of dust on the fan and heatsink, and those are much easier to deal with: you can simply blast them with a can of compressed air (hold the fan to stop it spinning excessively while you're doing this). I'd try that before I'd go messing around taking the heatsink off and trying to replace the TIM. To give you an indication of what dust can do to a cooling system, I cleaned my laptop recently and it went from having the fan close to 100% all the time and still hitting its thermal limit to the fan being almost silent and being 20^oC below thermal limit...

[kalniel]

Yeah I'd agree with the above - the two most likely cause of issues are dust and software. Check you aren't overclocking anything, as some OC methods result in a loss of power saving modes. It could also be that something software wise is preventing the card from stably entering low power modes. I almost always turn off GPU acceleration for web browsers - even on a 7 year old CPU I don't need it and it's better to keep the GPU in a low power state.

[prwilson]

Thanks for the replies. A few things to go on. I have recently tried 372.70 and the most recent 378.66 drivers - both have experienced crashes (the latest ones just now typing this ). I'll try and implement things individually to see if I can pinpoint the cause.

Additionally, a friend of mine speculated that maybe the drivers are throttling back under load in order that things don't go over some pre-determined settings - but there is no sign of excessive fan use or anything like that.