12 years old is peanuts. Many people have perfectly working video cards twice or even three times that age.

If a card was working fine and died after handling ("changed the thermal paste") that was most likely not coincidence - it suggests poor ESD precautions lead to the memory chips being fried by static.

I've no idea what is the expected lifetime of any component. My oldest VGA card is from 1989 I believe and still runs fine. But things can and will go bad from time to time. For example I bought used boxed Radeon HD6870 about 4 years ago. Tested it for few days and then put it back in a box. A few months ago I wanted to play around with it and installed it in the same computer as 4 years ago. Artifacts all over the display 🙁 Card was in an anti-static bag in original box. I guess bad investment 🤣

that is what im talking about

is it possible, that because of factory or components (new procedures) made by the memory makers that are not even tested for long periods of time, can it be possible that memory chips can go bad just because of time? or chips can fail because of age, memory clock, exposed to "normal" heat.

im sure many people here had issues with videocards that were fine and suddenly...artifacts on post.

Just speculation, but I would say that "modern" graphics cards (from around 2005 or so) with BGA chips and lead-free solder are much more likely to fail earlier due to the higher probability of developing cold joints over time.

Life expectancy of modern video cards is usually less due to more complicated production.

Fan failure is what killed my GeForce 3. The fan failed twice over 5 years. The first time (after 3 years of use) I was able to shut down the computer in time to prevent damage to the GPU. However the replacement fan also died 2 years later, and that time I didn't react quickly enough and lost the card. Do modern GPUs still die if they overheat or (like CPUs) do they automatically throttle down to prevent damage?

kixs wrote:

I've no idea what is the expected lifetime of any component. My oldest VGA card is from 1989 I believe and still runs fine. But things can and will go bad from time to time. For example I bought used boxed Radeon HD6870 about 4 years ago. Tested it for few days and then put it back in a box. A few months ago I wanted to play around with it and installed it in the same computer as 4 years ago. Artifacts all over the display 🙁 Card was in an anti-static bag in original box. I guess bad investment 🤣

Still I'd suspect handling in a situation like this. RAM chips don't spontaneously die (unless exposed to ionizing radiation, in which case it wouldn't just be the hardware that was dying). The trouble with ESD is that damage doesn't instantly surface - sometimes it can weaken connections and have them fail days, weeks, months or even years down the line. That means it could be the previous owner who killed the card, but nothing being visible at time of sale.

Newer cards are made to finer tolerances than older ones, and - generally speaking - they run hotter, which is also known to shorten lifespan of cards, particularly if the cooling solution decides to crap out - like Errius describes. His story is what I'd expect of generic age-related failure - which theoretically could have been avoided by monitoring the fan better. The cases of 'working' cards being put away then failing on/soon after re-use just scream ESD.

Nice example here by the way: our main systems aren't officially retro, but first/second generation i3/i7 are almost 9 years old now - and still perfectly serviceable for daily use. My partner wanted to upgrade her SSD from 80GB to 250GB and add 8GB more RAM. Sounds simple - stick the DIMMs in the free slots, get the new drive, hook it up to PSU and a free SATA port and then do rest in software. Somehow though she managed to fry the motherboard&CPU (memory controller) while she was at it. Nothing was connected incorrectly and the system was disconnected from the mains while being worked upon, and there was no visible damage to any parts - indeed, apart from CPU all parts worked in my other So1156 system. So that leaves ESD: she probably discharged straight into the memory slot while installing the new DIMM...

Yeh I haven't seen many stories about the ram failing, not saying it isnt a thing just its not as common as the lead free solder problem TheMobRules mentioned above.
added with cards around that time designed to run incredibly hot, then cooling own only made things worse.
This is why people "bake" or reflow their cards, reworking the solder.

So I guess keeping the cards well under normal temperatures would help, this would also happen on older cards like a voodoo, just not to the same extremes.

im buying a Anti Static Wrist Strap
and keeping my cards on anti static bags.

i do have some cards that give artifacts because of ram
voodoo3, radeon9800pro, viper770

i wonder if those can be saved making a memory transplant.

Some limited experience also shows that high-end cards tend to have shorter life spans. Probably a combination of having more components, running at hotter temperatures, and being stressed more (because most people who buy high-end cards intend to stress them).

THat's what I find so great about the 486 era - PLCC Socketed SRAMS on my S3 805.

Though I can't think it'd be impossible to resolder RAM chips on most vintage video cards from the 80's, 90's, and maybe some even from the early 2000's with some care. I've owned several of my video cards for over 15 years and those cards were 15 years old when I got them, and they are still alive now. Of course, were talking 386/486 era here.