PSU related? AMD driver issues with RX580

markk · 8 Oct 2020 at 19:45

I'm using an old Corsair HX520 to run the following: Ryzen 3700x / Asus Strix B550 Gaming-F / 32GB DDR4 / RX580 and I'm having issues with my computer restarting and giving me WHEA errors when playing Modern Warfare and even just trying to run Valley benchmark. Sometimes it won't let me into Windows and I have to drop into Safe Mode and remove the AMD Drivers to be able to use the machine at all.

This has only started happening 4 days ago and was running fine before - it's probably time to replace the HX520 now but do you think it is the cause of my problems above?

I've tried using different AMD drivers and even ran Prime, CHDSK, Memtest with no errors.

Meddling-Monk · 11 Oct 2020 at 13:00

From what you've said, there's a small possibility it's the PSU. As you have surmised, you do have a fairly old PSU there, given age and power rating it delivers, it may be it is now unable to keep up and provide the stable power needed to keep the system from dropping from stable system voltages, to an unstable one. But the only way to test this is if you have another PSU on hand.

The other possibilities are:

1. WHEA errors are often memory issues. So either your memory is overclocked beyond stability, or the IMC on your CPU is. An easy way to test this one, is to turn down to default settings and see if it still happens. Try memory first, then try again, if it still crashes, try the CPU clocks back to default.

2. Incidentally, because WHEA happens to be memory related typically, it might also be that there's a bad stick of RAM in your system. Do a memory test on individual sticks (I know you have said you've ran this already, but try again if you haven't with individual sticks) and see if it throws something up. If so, get it replaced.

If the system is restarting not only on WHEA errors, then there may be a possibility that it could also be:

3. GPU defect. It might be that there's an issue on part of the GPU's memory that's usually not utilised or stressed. And when fully stressed, finally shows up as a system crash. Although this is typically a system freeze, rather than simply restarting.

4. VRM issue on motherboard. Personally, since most of what your rig is fairly new and didn't have these issues previously, I would put this one very low on the list. But it'll be another defect related issue. But essentially, a PSU change won't solve these two less likely issues.

EsaT · 12 Oct 2020 at 17:22

If you were able to afford that expensive B550 board, then you certainly can afford PSU which is only decade old design.
(that design was literally outdated and low end decade ago)

And for higher end graphics card 650W is sensible starting level.
(courtesy of NVidia's lack of power efficiency advance)

Phanteks AMP has 10 year warranty.
https://www.overclockers.co.uk/phanteks-amp-650w-80-plus-gold-modular-power-supply-ca-09q-pt.html

markk · 12 Oct 2020 at 21:56

Meddling-Monk said:
From what you've said, there's a small possibility it's the PSU. As you have surmised, you do have a fairly old PSU there, given age and power rating it delivers, it may be it is now unable to keep up and provide the stable power needed to keep the system from dropping from stable system voltages, to an unstable one. But the only way to test this is if you have another PSU on hand.

The other possibilities are:

1. WHEA errors are often memory issues. So either your memory is overclocked beyond stability, or the IMC on your CPU is. An easy way to test this one, is to turn down to default settings and see if it still happens. Try memory first, then try again, if it still crashes, try the CPU clocks back to default.

2. Incidentally, because WHEA happens to be memory related typically, it might also be that there's a bad stick of RAM in your system. Do a memory test on individual sticks (I know you have said you've ran this already, but try again if you haven't with individual sticks) and see if it throws something up. If so, get it replaced.

If the system is restarting not only on WHEA errors, then there may be a possibility that it could also be:

3. GPU defect. It might be that there's an issue on part of the GPU's memory that's usually not utilised or stressed. And when fully stressed, finally shows up as a system crash. Although this is typically a system freeze, rather than simply restarting.

4. VRM issue on motherboard. Personally, since most of what your rig is fairly new and didn't have these issues previously, I would put this one very low on the list. But it'll be another defect related issue. But essentially, a PSU change won't solve these two less likely issues.

Thank you - I haven't had chance to try any of the above yet but will do shortly. I don't overclock anymore so it all should be set to default anyway but I will reset it to make sure. I'll try the RAM individually after that - PSU purchase is on the list anyway but I wanted to double check I'm not missing anything first, thanks again.

markk · 15 Oct 2020 at 13:15

I've done more test's on the memory, both together and individually - no errors.
Replaced the PSU with a RM850, reinstalled AMD drivers and it crashed/restarted on the login to Windows screen straight away.
Swapped out the RX580 with an old HD5850 and no problems, so I guess its RMA time on the RX580.

Meddling-Monk · 15 Oct 2020 at 19:22

Ouch. At least it was diagnosed and now taken care of. Always hated it when you run into an issue and it could be multiple various points of failure and you don't have the right stuff to check everything in house easily. That was the same issue on my HD7950 a while back.