Question Help with troubleshooting RX 560 2GB

[Glyrin]

A family member bought a used GPU and it worked fine for 2 months but now it's having issues. So they asked me for help but I ran out of ideas.

Out off nowhere, the screen will freeze in whichever image it was showing, even though the rest of the computer will still be working, once this happens only hard-resetting the PC will work. After the hard-reset the GPU stops giving a signal even though the PC boots and gives no error messages at all it just doesn't give a signal out. To make it work again I had to take the GPU out, boot with integrated graphics and turn off the PC, put the GPU back in. It worked, until it froze again.

Now for the weird parts:
- when it is working I am able to test the GPU with Furmark, Heaven, Superposition and synthetic benchmarks for hours without any issue;
- the computer works just fine without the GPU, just can't game on it cause old CPU integrated graphics but it does everything else;
- the GPU takes all power from the PCIE slot so no issues with the PSU;
- it's not the monitor or the cables since I tested with mine and issue keeps happening;
- it happens with both the HDMI and DP ports, didn't try the DVI port because i don't have a DVI monitor, cable or adapter;
- the crash even happened when on the desktop and only with Discord app open.

The GPU is an ASUS RX 560 OC 2GB

My current line of thought is that the display out connectors might be damaged because of the rust on them (former owner lived near the ocean) even though the board itself seems fine.

Any ideas what could be causing this? How this could be fixed? He is trying to old on getting a new GPU until things calm down (for 1 year), hence why he bought a used GPU.

Thanks for any help or tips

 

[solidsnake1298]

I wonder if this card was sold by the previous owner because it was problematic.

I would download MSI Afterburner and try underclocking the GPU and memory a little. I wonder if the factory OC is no longer 100% stable. Maybe it was never stable. Even if you stress it with Furmark and whatnot, it may have transient stability issues.

 

[solidsnake1298]

Also, MSI Afterburner would let you monitor temperature vs clock speed vs memory speed during stress tests. Do you know if, during your stress tests, the GPU is thermal throttling?

 

[Glyrin]

Also, MSI Afterburner would let you monitor temperature vs clock speed vs memory speed during stress tests. Do you know if, during your stress tests, the GPU is thermal throttling?

I know that after 3 hours of Furmark the GPU wasn't even hitting 70ºC with 30ºC ambient, I actually forgot to check the speeds during the test, I will have a look when I get a day off and report back. Thanks for the input and I will try to mess with the clocks as well, I usually have AMD Radeon Software running everything on auto and standard without any additional overclocks.

 

[solidsnake1298]

OC cards like these come with BIOS that include higher clocks than the AMD/Nvidia spec. So even if you are not software OCing with MSI Afterburner or anything like that, the card's BIOS will boost a bit higher than stock. I've had GPU overclocks (manually set in my case) that passed hours long stress tests, but when I was gaming later the games would randomly crash even after an hour or two of problem-free gaming.

Dialing back my manual OC solved the stability issues. I am wondering if the factory OC on your RX 560 is not 100% stable and that you need to dial it back a little in software.

 

[blckgrffn]

Usually hard locks for me have been PSU related. What’s the PSU situation looking like in this guy?

Particularly if the PSU worked for a bit then with a GPU is over stressed.

I hate PSU problems as they can multiply the issues…

 

[Arkaign]

I had a similar issue recently that scared the crap out of me on my 2080ti rig. It turned out to be something bizarre with windows itself. If you have a spare HDD or SSD, you can download the latest W10 ISO onto a USB stick of 8GB or more, and install it without a key for testing, leaving your original install alone.

An advantage here is that if you've got a windows install that's been going for a while, it builds up a fair amount of excess bulk even if you run windows disk cleaning and other utilities. Nothing is ever as effective as a fresh install.

Also, a friend of mine had a similar issue with his Ryzen 3700X/2070 Super rig, and of all the things it turned out to be caused by a cheap power cable. It was just really thin and caused erratic behavior. I gave him a fatboy type of cable that I got from an old server, and his problems went away entirely.

So there are a couple of things to try 👍

 

[blckgrffn]

I had a similar issue recently that scared the crap out of me on my 2080ti rig. It turned out to be something bizarre with windows itself. If you have a spare HDD or SSD, you can download the latest W10 ISO onto a USB stick of 8GB or more, and install it without a key for testing, leaving your original install alone.

An advantage here is that if you've got a windows install that's been going for a while, it builds up a fair amount of excess bulk even if you run windows disk cleaning and other utilities. Nothing is ever as effective as a fresh install.

Also, a friend of mine had a similar issue with his Ryzen 3700X/2070 Super rig, and of all the things it turned out to be caused by a cheap power cable. It was just really thin and caused erratic behavior. I gave him a fatboy type of cable that I got from an old server, and his problems went away entirely.

So there are a couple of things to try 👍

Also had a bad power cable. Out of hundreds (thousands?) I've deployed personally and professionally, it happens. It can be incredibly non-obvious.

In college I was over drawing a UPS and getting very erratic behavior too - it wasn't that my PC was over drawing it but a combo of more than one PC and other factors. It would be fine until the UPS heated up and then magically terrible.

 

[Glyrin]

Well I finally got some time to test the card, ended up not testing since it wasn't even giving a picture when i tried to boot the pc.
Took it to a store to see if it can be repaired, let's see it it's worth it or not

 

[funboy6942]

Everyone forgets these guys have thermal paste on them just like cpu, and board bits. But if it is freezing, gaming slow, have to underclock or volt, the its time to slap on some new digs on her. She is getting hot and locking up. May also want to grab a can of air, and a cheap toothbrush and clean them fans out, especially the fan fins, for they will not cool good if there is crap growing on them.

It is best to redo the paste on GPU/CPU/Board HSF around every two years or so, and give the old gal a once over, cleaning out all the rest of the bits if you filtration is the suck. Id do laptop stuff about every year, and clean its fan fins n bits as well.

I bet when you pull the HS off, the "paste" is a ceramic by now. Also can be like many new and used I have taken apart to find the OEM hardly used anything and it pretty much is and has been see through between the two. And then you have others with nothing better to do but glob the crap on, spilling onto the GPB chic below, causing it to have too MUCH paste between the bits to cool off right as well