I'm lost. I've done almost everything I can think of, but for the past 1 week or so my computer has started crashing and rebooting several minutes after locking it (Windows 11).
7800X3D
B650E PG Riptide WiFi
2x16 G.Skill Flare X5 6000 CL30
AMD 6950 XT
Corsair RM850x
I've had my 6950 XT for approaching 2 years now, first in my 8700K build and then moved to my 7800X3D when I built it in April. While temps have been awful (it idles at 60-65 and hits hotspot 105 under heavy load after a few minutes if I don't undervolt and reduce the boost speed), it has never had any issues other than that.
Now though, a variety of things happened over the course of a few days.
- I installed the Windows 11 AMD performance patch (KB5041587)
- I updated motherboard BIOS from 2.10 to 3.06
- I updated Phantom Gaming motherboard utility app because after the Win 11 update it was blocking components of the old version
- I updated Adrenalin to 24.7.1
I don't know exactly when the issue started as when these updates happened I was turning on the computer, using it, then shutting it down and don't believe there were any situations where I locked it. However, once I did lock the machine after a few minutes when the screen powers off the system dies. When locked, it stays on the Windows lock screen for a minute, then the monitor power offs. At this point, Everything is normal. For the next few minutes if I try to unlock the computer by waking the screen (note: Windows is set for sleep never but 25 minutes screen turn off, but it goes off way before this, like 2-3 minutes), it will come back on. Any longer than a few minutes and it's gone and one of two things happens.
1) It either stays as is but is completely unresponsive. The computer still appears to be running as case fans are still spinning and RGB is on, but it doesn't react at all to keyboard/mouse input and the monitor is still off. If I power on and off the monitor (LG ultrawide 3440x1440, connected with DP1.4) it just shows no signal then goes off again as if it's not even connected to anything. Secondary monitor (Lenovo ultrawide 3440x1440, connected with HDMI 2.0) does the same.
Or
2) The ran RGB turns off. Monitor is totally black. Fans are still spinning, but the system very clearly begins a bootlooping process. Without ever stopping, the fans slightly spin up or down and then the motherboard debug lights come on as if the computer has just been powered on. They go off, then nothing happens except for the fans continuing to spin for the next minute, and the process the repeats. During this time, the monitors are completely off as if they are not even connected. It never recovers from this and I have to hard power it off.
In both of these cases, when I hard power off the system and boot it back up, Adrenalin has reset my 6950 XT tuning, even if I've only made a minor change like adjusting the fan curve slightly but not touching voltage, clocks, power etc. The last events in Event Viewer before the system dies is always a kernal-power "The system session has transitioned from 0 to 1" (numbers anywhere from 0 to 7).
Sleep, hibernate, fast boot, fast startup are all off. I've tried Balanced and High Performance power profiles in Windows. I've set the ASPM (active state power management) and ULPS (ultra low power state) registry entries to 0. I turned off ASPM in the BIOS. I reverted back to BIOS 2.06. I reset BIOS to defaults and tried running without XMP/EXPO and with PBO off. I uninstalled Phantom Gaming completely. I reverted back to Adrenalin 24.3.1 which I know was working back when I built the system. I uninstalled Adrenalin with DDU removing all settings, then reinstalled various version 24.3.1, 24.4.1, 24.7.1, 24.8.1 from scratch. I've got a Corsair RM850x which should be enough. It's using 2 separate PCIe cables, one for each 8 pin on the card.
The only thing that has worked was physically removing the 6950 XT from the system and just running on the 7800X3D igpu. I can game for hours with the 6950 XT at full load in Helldivers 2, Cities Skylines 2, or Once Human, but as soon as the card goes into a low power state after locking Windows it just dies.
The only things I haven't done at this point are:
- tried a different GPU (don't have one)
- reinstalled Windows
- updated GPU vbios (is this even a thing for the reference card? I see some MIB cards have vbios switches but don't see anything about vbios versions from AMD even being available). Would rather just go through warranty if this is the case since the card already seems to have issues with power states I don't want it to have an issue during a firmware update
