Issues with the 6870?

I might be looking at a GPU issue here, what do you think?

Every single time I try to launch a game, I get hit with a black screen. Sometimes the audio keeps playing in the background, sometimes it doesn't, but the display stays completely dark. I'm running Windows 10 and I have the latest Crimson drivers installed directly from the AMD website.

The power supply is an EVGA 500W model, specifically the A500 Silent, which features two 6-pin connectors. I highly doubt the PSU is the culprit.

I went ahead and installed Nvidia Psyche 🙂, but it didn't change anything. I also disabled power saving modes in the power management settings.

Motherboard is an Asrock B85M.
The rest of the PC is just an Intel Celeron.

Maybe try some older drivers

http://www.guru3d.com/files-details/...-download.html

I'd recommend grabbing this tool first—it'll help you completely wipe out your current graphics drivers. Once that's done, just head over to the AMD website and download the latest official drivers for your card. Just a heads-up: skip the beta version and stick to the stable release. After you've got those installed, give the game another shot. Is this happening just with this one title, or are you seeing issues in other games too? If you want to be extra sure, you could also try running this: http://www.ozone3d.net/benchmarks/fur/

It’s a good way to put the GPU through its paces. If the system crashes right at the start of the test, then you're likely looking at a hardware issue with the card itself.

Larry Ortiz5 said:http://www.guru3d.com/files-details/...-download.html

Download that utility to wipe your current graphics drivers completely. Once you've done that, head over to the AMD site and grab the latest stable driver for your card. Avoid the beta versions; stick to the standard release. After that, try launching the game again. I should ask—is this happening exclusively with this one title, or are you seeing similar issues in other games as well? You might also want to run this program: http://www.ozone3d.net/benchmarks/fur/

Use it to stress test the GPU. If the system crashes right at the start of the test, then the hardware itself is likely the culprit.

The screen goes black in any game I play, including 3D Mark. Thanks, I'll give those steps a shot.

Try hunting down an older driver, ideally from back in 2016. I just don't buy that a power supply issue would act out like this...

EDIT: Rewriting this because it took me half an hour to type 😬

I'll give it a shot later once I'm off the clock, then I'll let you know if I managed to pull it off.

Listen, I followed the advice and uninstalled the drivers. I even went ahead and installed some older, Windows-certified versions, but the result is exactly the same.

I downloaded Furmark from the link provided and started the test. The moment the test kicked in, the screen just froze. I've tried this multiple times now, always the same outcome. The GPU temperature sits at 42 degrees while running Windows, so there isn't any obvious overheating issue.

It's worth noting that I bought this card used. Before this, there wasn't a dedicated GPU in this PC; it was just running on the Intel HD Graphics from the processor.

I also attempted to run an older title, Call of Duty 4. It loads into the game just fine, but as soon as the 3D sequence starts, I get a black screen. It happens every single time. No luck.

I'm at a loss. Either the graphics card is shot, or perhaps the PSU isn't providing enough juice under Load—though I doubt that. The label says it provides 20A and 400W on the 12V rail, and technically, it has two 6-pin connectors.

Could this be a Windows 10 issue? I found some similar problems online, but they were all related to Steam games, so they aren't particularly helpful here.

Dammit.

My only remaining move is to test this GPU in another PC to be certain. It would be foolish to go out and buy a new power supply or a different card on pure guesswork.

Yeah, I guess your safest bet is just testing it on a different PC.

I managed to scrape together a few results. I installed Afterburner, manually set the fan speed to 60%, and dialed back both the memory and GPU frequencies by about 50MHz. The card actually managed to launch Furmark, but then it just crashed after roughly 10 seconds.

Which utility should I use to undervolt the card, and what's a safe threshold for dropping the voltage if I want to give that a shot?

What could be causing this?

What kind of GPU temps are we looking at here?

analogtiger19 said:What kind of GPU temperatures are we seeing?

Ugh. I cranked the fan up to 80% manually—sounds like a Boeing taking off at this point. Running Furmark is fine for a moment, but after about five seconds, I have to force the program to close because Furmark shows temps hitting 85C and climbing fast.

I actually took the heatsink off already. The paste seems fresh enough. I put everything back together and tightened it down, but the temperatures are still skyrocketing. I suppose I’ll try some new thermal paste later and make sure I tighten those spring screws to the absolute max.

It seems like the card is just hitting sudden overheating spikes. I honestly can't say why. The fans are working; I set them to 80% manually myself.

I’ve honestly lost my mind here. I applied fresh Cooler Master thermal paste, tightened the heatsink down as much as possible, dropped the clock speeds 150MHz below stock, set the fan to 100%, and fired up Furmark. About 15 seconds in, it hit 105°C. At that point, I just gave up on the test.

It is as if the GPU heatsink is somehow failing to actually dissipate any heat—I have no idea what is going on. We're talking 105°C even with the fan pinned at 100% and frequencies set well below stock levels.

Temperatures within Windows stay steady at around 45°C at stock frequencies with the fan sitting at roughly 25% auto. For some reason, once you hit a 3D load, the temperature just skyrockets. In Windows, the card is idling at a mere 300MHz.

Hey, if it's hitting 30-35+ degrees Celsius, what’s the airflow like in the case? Are the GPU fans actually spinning up when they should?

It’s possible a fan died and it keeps stalling out—starting and stopping constantly. Try watching it closely to see if the speed stays steady, especially once you kick off a benchmark.

analogtiger19 said:Look, if it’s 90 degrees outside, what’s the airflow like inside the case? Are the GPU fans actually spinning at the right speed?

It is entirely possible the fan is failing—stuttering, stopping, then starting again. Try watching it closely to see if it maintains a steady RPM, especially once you fire up a benchmark.

The speed is constant. It sounds like a jet engine at 100%. I have to wonder if, for some reason, the heatsink isn't making proper contact with the GPU core.

The case is a Cooler Master 343 with three fans, and the side panel is off. The AC in the room is set to 77F.

If gas leaked from those copper heat pipes, then your card's cooling system is basically dead...

Alexander Moore67 said:if gas leaked from those copper tubes, then your card's cooling system is effectively dead...

The copper tubing is dented in several spots and looks pretty rough from the outside. I have a cooler from a 4080, though I am not entirely certain if it will fit this specific model. I suppose I will attempt the swap later. It should work, theoretically.

Maybe those heatpipes really are shot and just aren't doing their job anymore...

It’s my first time hearing that heat pumps could actually be a bad investment

It’s not exactly a perfect example, but you'll get the point. This shows why a GPU is basically a paperweight once those heat pipes get compromised...