Random BSODs and shutdowns following a CPU upgrade, yet successful stress tests passed.
Random BSODs and shutdowns following a CPU upgrade, yet successful stress tests passed.
Hello! I recently switched from the Ryzen 7 2700X to the Ryzen 9 5900X. My machine began shutting down unexpectedly without any BSOD, just a sudden power-off. Initially, I suspected RAM problems, so I ran MemTest86 and it completed successfully with no errors. Then I tested the CPU using Prime95, which also showed no issues; temperatures stayed steady around 90°C with no fluctuations. I considered the PSU as a possibility, even though it should handle the new setup well. I performed a PSU stress test via OCCT, but found nothing abnormal. I wasn’t convinced about the GPU either, so I tried OCCT’s 3D adaptive as well, still without results. All tests passed, but my computer still crashes randomly—sometimes while idle and other times when launching a game. It seems to happen more often with certain programs, though I can’t pinpoint a pattern. A BIOS update didn’t resolve the problem either. Here are the specs: CPU – AMD Ryzen 9 5900X, GPU – MSI Gaming GeForce RTX 2060 Super, 8GB RAM, Corsair Vengeance RGB PRO, 32GB, motherboard – Asrock X570 Phantom Gaming, PSU – C orsair CX650m 650W, 80+ Bronze, OS – Windows 11 x64, BIOS version 5.6 (latest stable). Crash dumps and Perfmon report are attached. Examining the dump files, they appear minimal. The most recent crash involved afd.sys triggering a page fault, possibly due to a null pointer (0xfffffffffffffd0 or similar), but the call stack looked normal. This behavior changed each time, making it inconsistent with BSODs as well—different drivers, modules, and processes involved. I’m not sure what else to try, so any advice would be appreciated! crash.zip Edited October 20, 2024 by Aad3 Added comment on attached zip file
Are you using the identical Windows setup? If yes, consider reinstalling the chipset drivers along with any related system drivers.
Reviewed the case's airflow, particularly around the VRM area. Verified the CMOS battery voltage level.
It's a Thermalright AQUA ELITE 240 V3, sufficient for the 5900. You've got two top fans, one rear fan, and two AIO fans in front, ensuring strong airflow. The CMOS battery reads 3.26V. Motherboard temperatures are stable, with the CPU and SB around 10°C to 40°C.
Software, it seems you're missing a multimeter with me right now, so I can't verify it. Could a low CMOS battery be responsible for this kind of instability? I recall sometimes having minor problems like lost RTC and BIOS settings without a battery, though that was a while ago and I haven't tested it on current boards.
The oldest dump files came from a past mistake by Microsoft regarding kernel symbols, making it hard to debug PCs with this patch. Symbols clarify what commands drivers perform; without them you receive raw memory data that’s hard to interpret. Most systems have their own symbols, so they usually work fine—though some users still face issues. The most recent one involved a generic error, mentioning Afd.sys, a common driver that often appears misleadingly in diagnostics. It was a general memory problem, but single dump files rarely give accurate results.
EDIT: For clarity, a red herring is an incorrect clue. I noticed you mentioned Driver Verifier, but it’s not recommended for everyday use. Most BSOD cases don’t need it. It’s mainly useful for developers testing drivers. Running it against other programs wastes time since most errors are harmless. Only specialized tasks like event tracing have any real value. For reference, NordVPN and Focusrite were the only driver verifier crashes reported.
I understand the situation, and including those details was meant to cover possible interpretations. It's really hard to proceed without symbols, especially given the limited info in a minidump about the loaded images. It looks like memory corruption might be involved, based on how unpredictable the faults are. I initially thought it could be hardware failure, but since Memtest performed well, I’m more inclined toward a software problem. The fact that it simply shuts down without a crash adds to my confusion. I’ve come to accept this reality. I still think a professional driver should pass standard tests, given how unreliable drivers can be, and they usually take the OS with them in case of issues. However, what I’ve learned over time is that driver development isn’t up to the task. Most third-party drivers fail these checks, which only makes things worse.