Discussion GPU Lifespans - Experiences over theory

[JackrumMadthing]

I'd specifically like to have people share their complete hardware failures from age alone

I couldn't find any official MTBF(Mean Time Between Failures) on major GPU's in the past decade out there, so lets presume a daily usage of at least 4hrs at maximum operating temperatures and that you have a decent overall case cooling system as well as maintenance regime (cleaning fans/sinks) with a steady Power Supply (no spikes or sudden shutdowns)

I'd also be interesting in hearing what people experienced in the months/days leading to the failure, if nothing at all and just one day nope... or flickerings/artifacts, wierd fan behaviour, etc..

My personal failure experiences:

• 3 month/Gigibyte nVidia GTX1650: Loadshedding power failure, board and chip fried wonderfully (doesnt really count I guess)
• 6month/Gigabyte nVidia GTX1650: Serious artifacts that degenerated swiftly over days to complete garbage on screen, clearly a lemon
• 7yrs/Gigabyte nVidia GTX960: Just refused to power on one morning
• 18yrs/ASUS nVidia GeForce FX5950 Ultra: Flickering during usage including full black screens for a few seconds, then normal function over about 6 months, consistent behaviour in this regard. Eventually refused to power up.
• 19yrs/Dell? Radeon 9600XT: About 1yr before failure had serious artifacts during POST, removed cleaned reseated and fine. Repeated again 1 month before failure, then just died outright.

My personal still running cards:

• 26yrs/Voodoo 2: Admittedly not daily, but still used occasionally in a nostalgia box.
• 26yrs/Rendition Verite V2200: Still the best way to play original Tomb Raider, also not daily but...ya
• 21yrs/ATI Radeon 9550: It's in a Linux box so sue me. Daily usage but hardly being stressed.
• 12yrs/ASUS nVidia GTX660: Mine craft and similar, daily usage.
• 8yrs/Reference nVidia 1080: Daily extreme usage (never over 70 degrees though), occasionally I get a black screen for 2s which is freaking me out.
• 5yrs/ASUS nVidia GTX1650: Daily usage, Star Citizen is its most taxing but functionally played title

So my secret desire here is to really figure out if my 1080 is on its way out through other peoples experiences, but I think it would be cool to get an idea of other cards general unofficial MTBF's as well!

 

[Joseph_138]

I've never had a video chip fail, going all the way back to the days of 16 bit ISA and VL Bus video cards. The things that have failed, are fans, and they are a lot less costly to replace than a card with a failed GPU, you just have to catch it in time before the GPU overheats.

 

[quickcheck]

I made it 7 years with my Asus ROG 1070, still works. Fairly regular gaming, sometimes very long sessions.
I had one XFX 6850 fail after 5 years in another system. Didn't seem to draw any power from one day to the next.

 

[PEnns]

Never had a GPU that failed. They just got too old and had to be replaced.

Till recently, I had perfectly running Radeon 9600XT, 8800 GTX and GTX 960 sitting in a box for over 10 years after being replaced!

 

[CelicaGT]

Only ever had one failure. GeForce 256 (Yes, that one). I killed it. I keep it as a reminder to not ***k with things I don't understand.

 

[JeffreyP55]

I'd specifically like to have people share their complete hardware failures from age alone

I couldn't find any official MTBF(Mean Time Between Failures) on major GPU's in the past decade out there, so lets presume a daily usage of at least 4hrs at maximum operating temperatures and that you have a decent overall case cooling system as well as maintenance regime (cleaning fans/sinks) with a steady Power Supply (no spikes or sudden shutdowns)

I'd also be interesting in hearing what people experienced in the months/days leading to the failure, if nothing at all and just one day nope... or flickerings/artifacts, wierd fan behaviour, etc..

My personal failure experiences:

• 3 month/Gigibyte nVidia GTX1650: Loadshedding power failure, board and chip fried wonderfully (doesnt really count I guess)
• 6month/Gigabyte nVidia GTX1650: Serious artifacts that degenerated swiftly over days to complete garbage on screen, clearly a lemon
• 7yrs/Gigabyte nVidia GTX960: Just refused to power on one morning
• 18yrs/ASUS nVidia GeForce FX5950 Ultra: Flickering during usage including full black screens for a few seconds, then normal function over about 6 months, consistent behaviour in this regard. Eventually refused to power up.
• 19yrs/Dell? Radeon 9600XT: About 1yr before failure had serious artifacts during POST, removed cleaned reseated and fine. Repeated again 1 month before failure, then just died outright.

My personal still running cards:

• 26yrs/Voodoo 2: Admittedly not daily, but still used occasionally in a nostalgia box.
• 26yrs/Rendition Verite V2200: Still the best way to play original Tomb Raider, also not daily but...ya
• 21yrs/ATI Radeon 9550: It's in a Linux box so sue me. Daily usage but hardly being stressed.
• 12yrs/ASUS nVidia GTX660: Mine craft and similar, daily usage.
• 8yrs/Reference nVidia 1080: Daily extreme usage (never over 70 degrees though), occasionally I get a black screen for 2s which is freaking me out.
• 5yrs/ASUS nVidia GTX1650: Daily usage, Star Citizen is its most taxing but functionally played title

So my secret desire here is to really figure out if my 1080 is on its way out through other peoples experiences, but I think it would be cool to get an idea of other cards general unofficial MTBF's as well!

Almost 8 year old Asus 1080ti. Everything is AOK.
I also store older obsolete GPU's for testing and a just-in-case situation.. Never had a GPU failure. 25 years + a couple of dead CPU's but many motherboard failures.

 

[Eugene$$]

never had issues with gpus

 

[Gururu]

Since 1995, the only card that died before a replacement was a EVGA GTX 590, which I had with another in SLI. Lasted about 5 years and started to produce major artifacts and screen corruption. I just took out the other still working 590 this month and am currently on iGPU.

 

[Avro Arrow]

There is no way to tell if a GPU is going to fail based on time alone. I've owned 19 video cards in my life and the only one to not work flawlessly for as long as I owned it was an XFX Radeon RX 5700 XT Triple-Dissipation. Sending it in for RMA ultimately fixed the problem and now it works perfectly as well (they even replaced it with a higher-end THICC-III model). Even my old GeForce 8500 GT and two Radeon HD 4870s still work perfectly (as glorified video adapters anyway).

There are some people still plugging away with an old RX 580 while some people have had cards as new as the RTX 3070 fail on them. In that way, it's a total crapshoot.

 

[Eximo]

Only ever had one GPU with even an issue. GTX285 from BFG has intermittent display outputs. Usually just means a restart or two to get a video signal out. I haven't fired up my Voodoo2 (no really compatible system at the moment) or Voodoo 5 in a while. Presumed working. Also have an old AGP Geforce 2 with no board to power it up.

I've always wondered about these new smaller process nodes and how robust they will be pulling hundreds of watts through a relatively small area, but so far no problems. But I have also been watercooling since my GTX580. Never attempted a GPU overclock before that, wasn't much call for it back in the days of 1024x768 and CRTs. 30 FPS was considered pretty good.

I have had friends with cards that died seemingly quicker then they should have. Smoking seemed to be part of the problem, which likely started with poor airflow and eventually internal tar build up on the PCBs.

From watching repair videos, seems like memory chips are a general problem. That could be down to poor cooling or overclocking for long periods.

 

[kanewolf]

I'd specifically like to have people share their complete hardware failures from age alone

I couldn't find any official MTBF(Mean Time Between Failures) on major GPU's in the past decade out there, so lets presume a daily usage of at least 4hrs at maximum operating temperatures and that you have a decent overall case cooling system as well as maintenance regime (cleaning fans/sinks) with a steady Power Supply (no spikes or sudden shutdowns)

I'd also be interesting in hearing what people experienced in the months/days leading to the failure, if nothing at all and just one day nope... or flickerings/artifacts, wierd fan behaviour, etc..

My personal failure experiences:

• 3 month/Gigibyte nVidia GTX1650: Loadshedding power failure, board and chip fried wonderfully (doesnt really count I guess)
• 6month/Gigabyte nVidia GTX1650: Serious artifacts that degenerated swiftly over days to complete garbage on screen, clearly a lemon
• 7yrs/Gigabyte nVidia GTX960: Just refused to power on one morning
• 18yrs/ASUS nVidia GeForce FX5950 Ultra: Flickering during usage including full black screens for a few seconds, then normal function over about 6 months, consistent behaviour in this regard. Eventually refused to power up.
• 19yrs/Dell? Radeon 9600XT: About 1yr before failure had serious artifacts during POST, removed cleaned reseated and fine. Repeated again 1 month before failure, then just died outright.

My personal still running cards:

• 26yrs/Voodoo 2: Admittedly not daily, but still used occasionally in a nostalgia box.
• 26yrs/Rendition Verite V2200: Still the best way to play original Tomb Raider, also not daily but...ya
• 21yrs/ATI Radeon 9550: It's in a Linux box so sue me. Daily usage but hardly being stressed.
• 12yrs/ASUS nVidia GTX660: Mine craft and similar, daily usage.
• 8yrs/Reference nVidia 1080: Daily extreme usage (never over 70 degrees though), occasionally I get a black screen for 2s which is freaking me out.
• 5yrs/ASUS nVidia GTX1650: Daily usage, Star Citizen is its most taxing but functionally played title

So my secret desire here is to really figure out if my 1080 is on its way out through other peoples experiences, but I think it would be cool to get an idea of other cards general unofficial MTBF's as well!

What benefit do you believe there is in single instance reporting? Just as a single person can die shortly after birth or live to 100, that provides no insight into the average death age. It is the aggregates that provide data which can be used for prediction.

 

[Eximo]

In essence, looking for symptoms to identify if his current old GPU is dying.

 

[Rokinamerica]

Never had a desktop GPU die (since '93), but in early 2000's had a laptop GPU go. As someone said earlier, I have had fans go, but not before I was able to replace.

 

[kira-faye]

Had a Sapphire Radeon HD4070 (I think?) die around year 5. Reference card. Started artifacting randomly, eventually got to where the machine would not POST with the card in. Its twin is still going strong last I pulled it out of the spares box, though.

No other card has ever failed on me. I did have a DOA card from Amazon Warehouse at one point, but that was clearly someone else committing fraud with their return, don't know if it counts.