Bitcoin Forum

Anybody know how long these gpus will last? Mining 24/7? I have an amd 7950. Wondering if i should add a 3 yr extended warranty?

Doubt it. Mining will not degrade your gpu anymore than gaming.

Yeah but people usually dont game 24/7

if they are being OCed and overvolted they will degrade to a degree, depends on the operational temperatures of the cards more than anything. You can just as well kill a undervolted cards with high temps, but they can usually handle less stress the higher voltage you run.

In 2 to 3 years you will most likely get new GPUs anyways, I don't think GPUs will die because of mining in that timeframe unless abused. My cards do have 3 year warranty as a stock (Gigabyte Windforce 3 7970).

I've heard of GPUs mining in the 70Cs that have failed after a couple of months, and others that have run in the 90Cs for years. Typically keeping them cooler will always be better, but also remember that temperature variance will do a lot more harm than you think. Running at GPU at 80C nonstop for months at a time is better than running a GPU up to 70C -> down to 30C -> up to 70 -> down to 40 -> etc.

Ok thanks for the advice

Lots of truth here. Power cycling is horrible for the solder joints. That being said some of my cards have been far more prone to dying up than others. My watercooled HD 5970s with underclocked cores, underclocked ram and vddc @ 0.95v still have cores die. I haven't found the problem but frankly they've hit ROI 4-5 times by now and I just retire the dead cards. For newer cards chances are you'll kill the fan before you kill the core/vrm unless you go with HD 7990s(which likely will be hot temperamental bastards).

Only think you need to worry about is temps. I have folded on cards which are over 5 years old at this stage and they still work fine.

I've had a couple 5970s die, which I'm pretty sure saw some pretty high temps at least some of the time.

So far I've not had any single-chip cards die.  They're not as easy to overheat.

I never overvolt my stuff, though I do OC sometimes, or undervolt.

I've a rig that use 2 ATI 5970 (a bit overclocked) 24/7 fans at 85% since 3.5 years now.. Everything is still fine !

Bad VRM cooling is a common culprit on a lot of cheap cards using non stock coolers. Can have really good gpu temps while they don't even have passive heatsinks on the VRMs, especially common among 58xx cards I've noticed.

Older cards might need the TIM refreshed, too.  I just got another HD5870 off the Bay of E.  Temps were 92C pretty much instantly, fan at 100%.   :-\

Took the heatsink off, and the previous owner had put some sort of cheap silver compound on there, which was really thin and gloopy.  It had all squeezed off the GPU core and was puddled inside the shim.  Cleaned it up best I could, put a blob of generic grey compound on, reassembled, now getting 69C at 80% fan.

My HD6970 was the same - the factory TIM had all dried up, again dropped 20C and about 30% fan speed just by renewing the TIM.