So last night my GTX 670 went up to the big PCB pile in the sky after 4 years of pretty much 24/7 use.. At first thought I was ok with that.. I mean its run pretty much anything I've thrown at it for 4 years.. but then I remembered I still have a 7800 GT which I got in 2005 that sits at home in our media PC that has been fine for over a decade... Is it that I've got a little unlucky and my 670 has kicked it early, or as GPU's get more complex their lifespans are shortening? I've never actually seen any information on it and would be interested to hear what people though.
A few thoughts on that: 1) Media PC = low load, low temperatures, therefore low thermally induced fatigue. 2) I agree with the sentiment that as GPUs get more advanced, their life span shortens, based on that's what I see in the real world (my parents had an old Sony Trinitron TV that lasted 20 years for example) 3) Kind of related to the above, but I'd imagine that older cards are less susceptible to silicon breakdown by quantum tunnelling since transistor size is larger. 4) I don't think GPUs are supported to last beyond three years, like most PC hardware. 5) Did you overclock the 670?
I've been building pcs since the days of 486dx. I've overclocked every cpu I've had since those days. Never had a cpu fail. Intel cpu i should say. I have had a BFG nvidia fail. I had one motherboard fail in year 3 (evga). I killed an ocz memory stick with static. I killed an abit motherboard when i had it out of case. I caused one psu to explode by powering a TEC i had fitted to cpu to chill it. I never buy seagate hard disks anymore. Gpus tend to run hotter than cpus... A german friend a mine is very proficient at killing gpus. He buys them used and then end up killing every one. He's pretty dumb. Zotacs come with 5 year warranty.
Heat is usually the enemy. I've had a few GPUs die over the years but then I've always been very unlucky. Nowadays I just slap an AIO on a GPU and haven't had one die since doing that. Plus running them in the 60s is a lot less stressful on the solder than running one constantly at 90c. The flux degrades over time..
I.. Actually can't remember the last time I had a piece of PC hardware (HDD's not included) DIE on me. Replaced due to performance issues sure, but outright death? I can't recall. I generally run stock clocks on everything, and go a little bit crazy with fans though.
Surely the increase in the number of transistors means an increased liklihood of failure unless the manufacturing process has also managed to up the quality of the product.
My Asus GTX 670 died about in March. And I wasn't pushing mine too hard. It wasn't a 24/7 machine, if i'm not using it, the PC was usually off. So I don't think it's only you. But it's highly likely that there is a silicon lottery somewhere regarding lifespan.
Four dead Asus motherboards - 2 x M3F, 2x P6T-WS... Two dead Zotac H55 mitx boards.... Two dead R9 280x's - under water cooling ! One GTX 460 One i7 900 cpu All run 24/7 OC'd and thus killed two PSU's in the process.... burnt out 12V pins on P6 series boards, and one M3F as well ! Apart from that all's fine ! dunx P.S. Currently oc'd and 24/7 @ 100% load :- GTX 460 GTX 480 HD 5870 GTX 960 GTX 970 HD 7970 i7 3770K, i7 4790K, E3-1231 v3, i7 870, plus a spare i7 870 looking for a new home ! EVGA 750 GQ PSU, Silverstone 450W SFX PSU, Coolermaster 1250W PSU, Fractal Array 300w psu.... I need a new EVGA 850 P2 for my repairs.... and a new Maximus Gene III for my dead PC....
