So my games cause a system hang after a while, COD Warzone will run smooth or sometimes it is -10 fps or the whole system hang.
The problem: Black screen with sound and then after a few seconds the fans go noisy, I have to shut it down with the power button.
I tried:
Undervolting even by 50 mV causes crash with graphic demanding apps, increasing the power limit in Radeon Settings even by +10% causes it to hang inmediatly when trying a benchmark (when doing it with Afterburner it causes the same), OCCT Power test runs fine but crashes some times and if I up the power limit it opens for 1 second and hangs.
I put the Ram frequency and voltage to auto (in case the ram oc was the problem) and the problem persists.
In Unigine Heaven the card is between 140w and 155w, most of the time at 140w.
My system specs:
Ryzen 5 3600
2 x 16gb Hyper X Fury 2400mhz overclocked to 3200mhz
MSI RX580 Gaming X 4gb
Antec HCG 620M psu
Gigabyte B450M S2H
Just tried Aida64 system stress test, it hanged too.
What are the CPU & GPU temperatures when you stress test the GPU?
What happens when you use OCCT to stress test the PSU? Are the Outputs within 5% tolerances?
GPU reaches 71c and CPU 80c, how do I check those outputs?
When stressing only the GPU with OCCT 3d test:
63c for CPU
71c for GPU
1,41 V CPU VCORE
0.95V GPU
12.024 V, 5.050 V
119 W (mostly 134) for GPU
836,2 Mhz gpu clock
Your Temperatures are good for both the CPU and GPU and your PSU 12 & 5 vdc voltages are good also.
The software will show the PSU outputs during the Stress tests or running a Monitoring Software while running the Stress tests.
The best way to check if a hardware is defective (RX 580) is either putting the GPU card in another compatible computer and see if the same thing occurs or you replace the RX 580 with another GPU card in your computer and see if the same thing occurs.
Most GPU Manufacturers have a 3 year Warranty on their GPUs. You can always open a Ticket with the GPU Manufacturer and see if they believe the RX 580 needs to be RMAed if still under Warranty. If not under Warranty they may still give you an indication what is wrong.
Try installing a previous RX 580 AMD Driver and see if the problem reoccurs or changes or is fixed from here: https://www.amd.com/en/support/previous-drivers/graphics/radeon-500-series/radeon-rx-500-series/rade...
NOTE: I would put everything back to default settings in BIOS and in Radeon Settings and set back your RX 580 back to default for troubleshooting purposes. If you start making changes either in BIOS or in Radeon Setting and changing the RX 580 parameters ( under or overvolt as an example), one of those changes could be causing the problem you are having.
Also 3rd party software like MSI Afterburner may cause conflicts with AMD Radeon Settings. best to temporarily uninstall Afterburner until you figure out what is wrong.
Yeah I just touched the GPU settings to see if it fixes anything, I'm always running it on default settings, also tried with default BIOS settings and problem persisted...
Is this a card that had been working and now is not? Or is this a card that has always done this?
Polaris does not like to be underpowered. Most cards are factory overclocked and in my experience need the Power Limit set to +50 to be stable as the default power ceiling is for reference designs. Typically you will want a custom fan curve that will ramp up to 100 and much sooner than the factory default. My 2 Polaris cards get unstable above about 74 degrees.
If this card was working right and now isn't.
a. Did this start after a driver update?
b. If the card is more than a couple years old the thermal past may be bad and time to replace or RMA the card
or
c. You do have power supply going bad.
Thanks for replying, this started happening like a month ago, I tried 20.5.1, 20.4.2 and now 20.3.1 and some 2019 drivers, all show problems, also changed thermal paste a few weeks ago.
When I up the Power Limit even by 10% it crashes immediatly in benchmarks, when I lower it, it works better but I lose performance.
Honestly it sounds like a defective card. I would try to RMA it. Unfortunately depending on who made the card, changing the paste can void warranty. Read the warranty and if it says anything like that maybe not mention you did it.
how is your PSU holding out?
have you tried a clean install of windows to rule out a messy setup?
Yes, I tried 2 clean installs, how can I check how my PSU is holding out?
I have the PSU since 2015, bought it second hand, it came out in 2012, I have the GPU since 2018.
If you can buy one retail locally or Amazon, some place that will take a return, the easiest thing is to swap it out and see what happens.
I should add that with the age of your power supply it could be going bad. However the issue you are having is a pretty common one with bad GPU's too.