Drivers & Software

[Compsplay]

Here's a description and then a timeline of my issue:

My PC started having issues with crashing while idling (screen turned off). I don't remember when I updated my GPU drivers exactly, but I do remember updating it to Adrenalin Edition 24.6.1 not too long before. However, it was fine for at least a few days before the issues started popping up and then devolved into OS instability.

Specs:

CPU: AMD Ryzen Threadripper 2950X

M/B: ASUS ROG Zenith Extreme Alpha, BIOS 2701

GPU: AMD Radeon RX 6900 XT

Latest GPU Driver Version: 24.7.1

07/15:

GPU driver is 24.6.1, BIOS is 2101. The monitors are black, and PC seems to have crashed while idling overnight (I routinely leave the computer on overnight to idle after locking it). Jiggling the mouse or typing on the keyboard doesn't wake it up. This is the first day it starts to happen. Turning the PC back on, it initializes to POST code 97, with text "Test NVRAM". It remains stuck here unless I shut it down. Decide to shut it down, turn off PSU, turn on the PSU, and try again. This time it boots.

07/16 - 07/18:

Issue persists, but degrades from only occurring overnight to happening if left for a few hours. I suspect a PSU failure and contact my PSU manufacturer about a possible replacement. I still haven't received an update at the time of this posting. I can still play video games and do intensive work, but if I leave the computer without inputting anything for 1-2+ hours, it seems to crash, and it doesn't output any video. I read some posts about USB Selective Suspend and think that might be an issue, so I go into my power settings to enable it. This does not resolve the issue.

07/19:

I read some posts and decide that maybe the BIOS is too old or out of date, even though this issue did not persist for the past year or so when I have the current BIOS. Still, no harm in updating it, so I download and upgrade the BIOS from 2202 -> 2701. The problem appears to go away as the PC did not crash the next morning, my RAM is able to boost to 3000 MHz (its designated spec) for the first time in years, despite not being able to before. However, I quickly change it in the following days back to 2800 MHz (which is what it normally is) following these crashes.

07/20:

I jiggle the mouse, and the PC does wake up, only to crash in the time it takes for it to idle. I come back to blank monitors, and I power cycle the PSU. I get some new POST codes when I try to boot it without power cycling the PSU first, like 92, and 94. This makes me think that the GPU driver might be out of date, so after turning it on I update the driver to 24.7.1. I do this by having the AMD Software reset to factory settings (checking the box) and the installing it, after which I reboot the PC at the prompt. This doesn't fix the issue. I've updated my drivers this way for a long time and I haven't had this issue before, but maybe this is causing them to be corrupted.

07/21 - 07/23:

Issue persists, but now the PC crashes after leaving it for 20-30 minutes on idle. I check the power settings and it just seemed to crash whenever the monitor turns off, even though I set the PC to not turn off (and this issue has never happened to me before). Power cycling the PSU turns out to not always solve the problem, as sometimes the PC boots into POST code 92. As a result, to be safe I enable hibernation and put the PC in hiberate before I go to sleep. This seems to work around the problem. 

07/24:

I run complete Windows Defender scan on my drives. My PC is able to complete this successfully over the course of 4 hours without issue, and detected no problems. It had no issues and I thought 

07/25 (Today):

The PC has BSODed multiple times. At the start of the day, when I tried to wake it up from hiberation, the PC appeared to wake up, but nothing showed up, demonstrating that it crashed while hibernating (that is my guess). Checking event viewer, it appears to have "shut down unexpectedly at 03:41". After that, I turn it on again and it worked normally, until it stopped opening applications. As in, I would click on them, and nothing would happen. I check Task Manager and I watch as Discord simply hangs in the background, while my web browser and other applications start up and shut down immediately. Here are the BSOD errors, courtesy of WhoCrashed (free software for reading dump files):

1. 15:10:58: CRITICAL_PROCESS_DIED
2. 17:33:23: SYSTEM_SERVICE_EXCEPTION (fu
3. 17:43:58: PAGE_FAULT_IN_NONPAGED_AREA
4. 18:06:36: REFERENCE_BY_POINTER
5. 18:17:45: DRIVER_IRQL_NOT_LESS_OR_EQUAL
6. 18:17:45: DRIVER_IRQL_NOT_LESS_OR_EQUAL (there were two of these dumps)

All of these were minidumps, except number 5, which was a full memory dump. I have tried using "dism /online /cleanup-image /restore-health" and "sfc /scannow" as an administrator, but the former just gives me error code 87 and the latter will reach 100%, then say that it cannot complete the requested task. I've tried turning off D.O.C.P. memory overclock and downclocking my RAM to 2800 MHz, but that has proved unstable as well. My CPU has not been overclocked for a few years, though this was not due to stability issues and due to performance in some games dropping because I had it locked at 4100 MHz but benefitted from single cores boosting higher. 

Update: While writing this, I decided to downclock my RAM to 2400 MHz, and running an "sfc /scannow" scan did not seem to find any integrity violations. I still have D.O.C.P memory overclocking disabled. It has not crashed, and I will update if there are further issues. Updating the BIOS may have caused an issue with the RAM stability, but I'm not sure, though this happened in the past when I first built the PC in 2020. I also disabled CSM in my BIOS settings, since I don't remember enabling it and Windows 10 does not require it being enabled to boot into it. Disabling CSM did not resolve my stability issues, but I wanted to mention changes I made to the BIOS.

[radu1006]

I never got BSOD, but other driver crashes I get all the time with any of the 2024 drivers.

I think the aggressive power management for idle causes these issues. Especially if you use HDR.

The only viable solution is to use 23.12.1 drivers for me.

[Compsplay]

07/26 Update:

I reset the CMOS on my motherboard yesterday, so that the PC ran at default settings; no memory speed faster than 2133 MHz, no CPU overclock. I then used AMD Cleanup Utility to remove all the drivers, and redownloaded AMD Adrenalin Edition to install 27.4.1. No BSODs or anything like that. However, when I left the PC locked and idle overnight (intentionally to test), I woke up to a screen that received no display signal when I tried to wake it up, just like when the problem started. I turn off the PC, and then turned it back on without resetting the PSU. The PC boots just fine and gets past POST, but there is no display out to the monitor. My GPU is confirmed working as the fans are spinning. I've tried unplugging and replugging the HDMI and DP cables to my monitors, but that doesn't resolve the issue. I don't know what else to try or what else might be the problem. 

I want to add that when I boot the PC, since 2701 there would be a brief flash of the ROG logo when the PC turns on, plus the phrase "Press DEL or F2 to initialize boot menu" or something like that if the GPU has an output, but it doesn't do that. Furthermore, when I disconnect a peripheral (like a keyboard or mouse) and plug it back in, the RGB lights do not come back on after some time (though if I do it shortly after the mobo stops displaying POST codes, they still turn on, indicating that the motherboard is able to detect them and send power to them). When my PC boots, it does all the checks, and out of the codes it sticks to B4 briefly before going to 92 during the VGA checks, and then A2 is the last code when checking the HDD/SSD before the PC (apparently) boots into Windows. 

I turned off the PSU and turned it back on, and my GPU finally has a display out to the monitor.

[goodplay]

Have you web searched the 'Test NVRAM', maybe also post a query on the asus rog forum?