Question AMD HD 4850 Errors/Failing, - need input

Aug 26, 2020
Hey guys, been having pretty serious GPU issues with a Sapphire ATI Radeon HD 4850 that I just cannot seem to fix. It may well be the card itself beginning to break, I just don't know. GPU is quite old, obviously, and the computer itself is not exactly current, but it's a quality build and runs fine for my purposes.

It began about 2 weeks ago with the computer freezing, no input possible for about 8 seconds, then it would recover and be usable again, with a notification in the bottom right of the screen saying 'your card has recovered' etc. It would go back to normal at that point, but inevitably, it would happen again. At first, it was hourly, then after a few days, every 45 min, then after a week, every 25 min. Then, yesterday, it started chain-freezing, to the point I could literally do nothing.

I began checking the temps on Everest, and they were way up there. Normal temps as far as I know for GPU, max, are around 185ºF. I only had a browser and 1-2 games open, something it was always able to easily handle in the past, and temps were reading @ 250ºF+. (!!!) So, I pulled all the plugs on the back of the computer and made use of the fact that I was removing my air conditioner from my window on the same night, utilizing the open window to blow out all the dust/debris from the computer case with ~30 psi via air compressor. The sound card - which I have long since stopped using/needing due to poor drivers(I use onboard sound) - was only a few millimeters beneath the fan of the GPU, not allowing much room for air intake, so I removed the sound card entirely, freezing up significant space for GPU cooling/air flow. I also tied the cords/wires back even more cleanly, creating as much air flow as possible. It's also now late Fall, almost winter here in Boston, so the nighttime and morning temps are very cool if not just downright cold. Long story short, between all of these things, I was able to get the temps right back down to normal, as you can see in the screenshot. Even under load with a few gaming windows open, 2 browsers with 30 tabs between them, several other apps, it's still not getting up to 185ºF again.

So, on paper, this all seems good, but, it's still malfunctioning.

The card - or whatever is causing this - is still peppering the screen sporadically with multicolored dots that tend to vanish when cursor hovers over them or the screen changes to a new page. It is still restarting on me with the same 'driver stopped responding' error, also seemingly randomly, and I just don't know, A) what is causing this, B), how long this will go on for until it fails, and C) how I fix it. I will buy a new card if necessary, I just need to know that's what's causing this. I've seen one instance when I worked in IT of a PSU causing something similar, but typically it's been the GFX card itself. I'm just not sure. I purchased 16gb of used good condition high quality RAM off ebay about 6 months ago, replacing my NewEgg-purchased 4gb RAM, and have had no problems that I can identify that would trace back to the RAM. RAM/CPU hardly ever malfunction in my experience anyway, so I'm not thinking those as culprits. The boot drive that windows, and even the games, are running off, is a 240gb SSD, also recently purchased, maybe 6 mos ago. Even if it(or any of my 4 hard drives) were to malfunction, I don't know how/if it would cause these types of errors I'm seeing. Again, I'm just speculating, thinking out loud; any input/help is much appreciated. Thank you.

Here is an Everest screenshot of as much of my PC(which I built) info as I could fit on the screen. If you need more detailed info about the PC just let me know.