PC Graphics

[amdpleasedontblackscreen]

CPU: AMD Ryzen 7 2700 Eight-Core Processor

GPU: Radeon RX 580 8GB

Ram: 32GB

PSU: 650w

Been having issues like this for the past year and I'm unable to fix the issue. I read all the reddit posts, amd support forum posts, tom's hardware forums, youtube and even the second page of google trying to fix it with no avail. I did memtest, no errors. System scan, nothing wrong. DDU in safe mode 3 times nothing changed. Set fan curve really high, still crashes when system is 50 C. Undervolted gpu, crash. Swapped power supply from another pc, crash. Disabled MPO, crashed 10 minutes ago. Revert drivers to the stable one, crashed right after.

Mainly it's a black screen and the computer restarts or it freezes and I am forced to restart. This is very painful since it crashes at the most inconvenient moments. I'm 99% sure it's the gpu since everything bad that happened before could be narrowed down to the gpu. 

[amdpleasedontblackscreen]

Just fully cleaned the pc and replaced thermal paste, just crashed.

[DimkaTsv_alt]

Try to make registry edit:
[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\Dwm]
"OverlayTestMode"=dword:00000005
That will limit or disable MPO effect on system. Currently there are problems with MPO/HW Acceleration/VSync conflictations on software level. maybe this your issue. 
Issue becomes worse with frequent alt-tabbing (and happens on returning to game) or using multimonitor setup (Including cheaper extra wide displays that abuse 2x matrix glued together as technology)

--------
Was forced to create alt, because of answer number limitation (timeout basically).

[amdpleasedontblackscreen]

I said in the post that I already disabled MPO in the registry

[DimkaTsv_alt]

Sorry, read diagonally and missed that line.
Can it be problem with weak displayport cable? What monitor are you using?
Flickering Issues With AMD GPUs? We've Got You Covered! Driver Settings And Choosing Proper Cables -...
Even though it shouldn't appear as serious as game freezing, though. But maybe for older GPU it can cause really wonky stuff when display output going missing suddenly.

Or are there any specific consistensies on crashes? Types of load? Applications or their combination? (i have 2 games that cause BSOD if they are both opened simultaneously, so here is that for software incompatability)

[amdpleasedontblackscreen]

Victoria 3 and Apex Legends seem to crash the most but I haven't play a lot of other games recently so could be more.

[cpurpe91]

While I do not know much about Victoria 3, Apex Legends has been causing crashes for months on AMD GPUs. Weird issue with crashing apex legends - AMD Community

It has caused problems that persist even after the game is closed, and I experienced crashes in other games after playing Apex for a few hours. The only way I could fix it was to not play the game anymore, and DDU then reinstall drivers. It is a problem on EAs end where they keep patching the game and breaking it for AMD users. 

I hope this helps.

[MADZyren]

Might be interesting to create a restore point before installing or running the game, then install and play it until issues occur, then go back to restore point.

[Gost]

Disable hdcp

Reduce clock speed to 1256 for RX580 8gb,issue with boost clock.

This worked for me.

[MADZyren]

More important data is missing

Motherboard brand, model, bios
PSU brand, model, age, do you have two separate cables from PSU to GPU or one cable splitted into two?
Memory brand, model, how many dimms, clocks, latency, XMP / DOCP
What software is running
Does it happen when stressed or at any moment
Which brand GPU, is it new, bought used, how old
Running PBO?
When you used another PSU, did you change cables as cables are not necessarily compatible between power supplies and you might fry your machine if you swap PSU's without swapping cables as well...

[amdpleasedontblackscreen]

I changed cables with the PSU and now I put the old one back into this pc since nothing happened when I installed the other one. Right now it's a PB500 from Gigabyte. Most of the computer parts are 3 years old.

Motherboard: Asus TUF B450M-PLUS GAMING(the bios is latest ver)

PBO off, bought gpu used, it crashes at any moment when gaming. I did a stress test for 5 minutes and no crash. DOCP off. Ram: GeIL EVO Potenza AMD Edition 32 GB (2 x 16 GB) DDR4-2666 CL19 Memory

[MADZyren]

"I changed cables with the PSU and now I put the old one back into this pc since nothing happened when I installed the other one. "

You can't use another PSU's power cables with another PSU as they can be wired differenctly and in worse case fry your machine.

"Right now it's a PB500 from Gigabyte."

Low end 500W PSU with an eight core CPU and RX580 is pushing it. Could work, might not. I'm not a PSU expert, but 38A  on 12V rail is 456W, not 500W and from what I've understood reading other PSU reviews over time, it is usually expected for 12V rail to be able to deliver the wattage PSU is adverticed being cabable of.

https://ru-gecid-com.translate.goog/power/gigabyte_gp-pb500/?s=1&_x_tr_sl=auto&_x_tr_tl=en&_x_tr_hl=... 

Orig: https://ru.gecid.com/power/gigabyte_gp-pb500/?s=1 

You could try undervolting CPU and GPU to see if it becomes stable when drawing less power. Stress tests pull power evenly all the time, while in gaming, power draw fluctuates. Could be that your PSU simply can't keep up.

Also have you disabled hardware acceleration from internet browsers, steam and all the software you have running, as that has caused issues to some?

Most likely you can enable DOCP as your kit is not running high frequenzies.

[Volkdude85]

650W PSU is minimal.  It "should" work but there is a reason for quotation.  Power supplies are often poorly made thermal trash that can kill you if you modify. 

In Building construction if it is on for 3 hours or more it is considered a "continuous load" you over rate the load by 25% ( roughly don't quote me).  I would go get a 800 watt at least supply and plug it in and see if it fires up.  If it does go big supply if not ask Santa for a new build.   

If it is right after a cleaning try re seating everything with pins. take out the ram check if paste dust bunnies crud got in GPU, CPU, double check steps .. I jammed in ram not quiet right the other day it happens( I have bent pins too(not recently)) I have had a very simple reseat fix lots with a good flashlight holder and patients.  good luck.

[amdpleasedontblackscreen]

I did a stress test for 5 minutes and it doesn't crash when loading up a game and playing it for the same amount of time has a big chance to crash. I will probably test the ram next, gpu has been thoroughly cleaned already and cpu I have to check that too.

[Volkdude85]

first time I pulled a cooler on a old xp cpu I took the cpu out with the cooler and bent the pins. it was glued on!  if it has just been cooled differently then the status quo then your fan motors could load less, XMP could pull higher clock and voltage That one contact over there goes from .5 to .4 Ohm whatever, 650 Watts is close enough to the limit where If it were an elevator, I'd take the stairs.  I have seen on an old 6800gt like circa 2003 pc limp on an old 600 watt with similar crash issues and thrive on a new 850W. PSU.  inrush current is not RMS.  Not all rectifiers are equal.   

[amdpleasedontblackscreen]

Also, I swapped out the gpu with an old GTX 650 card I had in the pc and still crashed in the same game. So it's probably not the gpu or power.

[cpurpe91]

Okay so we can narrow it down to something other than the GPU. Have you run Windows Memory Diagnostics?  Just read you tried memtest. I would have to say it is the board or the CPU at this point. 

Super important: Do not use cables that are not specifically for your PSU MAKE AND MODEL. While you may think they have the same PINOUT they rarely do and can easily ruin your hardware, or even worse, start a fire.

[jhautrey]

I'm having the same crashing issues w different specs but I've tried almost everything I've been told with no fix. I'm almost sure it can't be anything other than the board but I don't know what to do to be sure. Might just have to take it in, which I have been avoiding.

[Volkdude85]

a 650 is like 75 watts at load a rx580 is 185 watts and can spike.  You ever try to power a 100 watt light bulb on a bicycle generator?  Pretty sure most people would die before they could read.  Honest to God I would be AMAZED If it's not your power supply. AMAZED.. but I know 746 watts is a horse power and how a bridge rectifier works and I am saying dude 185 watts / 12 volts = 15 amps and that isn't peak.  Scotty is screamin "he don't have the power"  Corsair 1000 watt plus my dude why because they use good capacitors and "can" have pwm fans that are very quiet out of box.  My 1000 watt "silent pro" wasn't... I fixed it but I know the caps can go off like bombs and can kill me.. I also know I'm never going to sell it.  look for a single 12volt positive rail that can handle all of the load, it will mean a single higher end capacitor. I don't want to worry about rail loads with multi small capacitor 12 volt rails. If you are only pulling 700 watts now you have 300 watts of thermal overhead constantly... probably why my rx580 is purring right now.

[amdpleasedontblackscreen · Accepted solution]

Removed 1 stick of ram making it 16gb in total and no crash yet, I don't know if it was the ram itself or the power the ram uses

[Volkdude85]

That's a bad stick of ram, it happens, I had a nice new set have a stick DOA

glad you have a next step.