Are there issues with strange windows crashes and sudden high GPU temperatures?
Are there issues with strange windows crashes and sudden high GPU temperatures?
Hello. I'm sorry for the lengthy update; I just aimed to gather all details from this unusual situation in hopes of resolving the issue.
Summary at the end.
SPECS:
Device: Desktop
GPU: RTX 4080 Super Founders
CPU: RYZEN 9 7900x
Motherboard: ASUS ProArt X670E-CREATOR WiFi
BIOS Version: 2704
Memory: CORSAIR Dominator Titanium 64GB DDR5 6000MHz C30
Power Supply: Seasonic Prime TX 1000W
Case: Antec Flux Pro
Software: Windows 11 PRO 24H2 (latest update KB5048667 pending)
GPU Drivers: GEFORCE GAME READY DRIVER – WhQL version 566.14
Chipset Drivers: AMD X670E drivers v6.10.17.152
Explanation:
At the start of last month, while playing Marvel Rivals, I encountered a BSOD upon exiting the game. Initially puzzling, it stopped afterward. Then, during a session with Balatro and a YouTube video, my PC abruptly force-restarted. The first crash didn’t produce a dump file, so no clear error was visible. After reinstalling GPU drivers in safe mode and running Furmark with CPU burner for 30 minutes, everything appeared normal until I stopped the tests and Windows BSODed again with the message: Kernel Security Check Failure.
I checked the dump file using Bluescreen View and found the failure to be noskrnl.exe, indicating a memory issue. Running MemTest86 confirmed no errors. I then updated the BIOS to the latest version, re-enabled EXPO, and re-seated my RAM. After that, I ran Furmark and CPU burner tests again. The system restarted without BSOD this time.
However, during a 30-minute Cinebench stress test, the PC hard-restarted immediately after stopping the CPU burner. I resumed Furmark post-restart but still experienced a crash. Post-test, temperatures spiked to 100.6°C on CCD2 briefly, then dropped to 88–95°C, with no thermal throttling reported by HWInfo64.
The issue seemed tied to thermal performance under stress, especially since the CPU reached its max temperature of 95°C. I attempted standard troubleshooting steps—DDU scans, DISM/SFC, RAM reinstall—but the crashes persisted only after stress tests.
Any suggestions or insights would be greatly appreciated!
Following the BIOS update, did you perform a hard reset to guarantee that all previous configuration data was cleared (since sometimes the BIOS resists discarding old settings without this step) and to force the hardware tables to be reinitialized? If not, begin with this process.
BIOS Hard Reset steps
Turn off the device, switch off the power supply and detach the PSU cable from either the wall outlet or the power source.
Take out the motherboard's CMOS battery for roughly three to five minutes. In certain situations, it might be required to remove the graphics card to access the CMOS battery.
While the battery is removed from the motherboard, keep pressing the power button on the case continuously for 15-30 seconds, in order to discharge any remaining charge in the CMOS circuit. Once the five minutes have passed, reinsert the CMOS battery, making sure to place it with the correct orientation as it was removed.
If you had to remove the graphics card, you can now reinstall it, but be sure to reconnect any power cables and your display cable as well.
Reconnect the power supply cable, turn the PSU back on, and restart the system. It should show the POST screen and provide options to enter CMOS/BIOS configuration. Launch the BIOS setup program and adjust the boot settings for either the Windows boot manager or legacy systems, depending on the drive where your operating system is installed.
Save your changes and exit. If the system successfully POSTs and boots, you can proceed further, including returning to the BIOS to configure additional custom settings such as Memory XMP, A-XMP, D.O.C.P profile configurations, fan profiles, or other specific parameters that may have been lost during the CMOS reset.
In certain scenarios, after resetting the CMOS, it might be necessary to load the Optimal or Default values and then save your settings to ensure the hardware tables reset properly in the boot manager.
It is also advisable to consider, if a hard reset is required initially, that if the issue pertains to a missing video signal, trying a different display type could be beneficial. Many systems fail to function correctly with displayport setups. HDMI is a good alternative if you have no display or visual access, or if you receive no signal messages.
Additionally, switching to another monitor when possible can be helpful if there is a display issue. This is common.
Also, which CPU cooler are you employing?
What is the design of your case?
How many case fans are present, and can you describe their exact positions and settings (intake or exhaust)?
If you're using an AIO cooler, where is the radiator positioned and how are the fans arranged—intake or exhaust?
Specifically, which memory slots do your modules occupy? Starting from 1 as the nearest to the CPU up to 4 as the farthest from the motherboard edges. The placement matters because it affects termination and performance.
I've noticed many systems replicate the same configuration as others—placing memory in slots 1 and 3 or 3 and 4 instead of the intended 2 and 4. Some setups cause crashes after shutdowns or stress tests, since cooling fans can slow down when heat buildup occurs unexpectedly. This can happen if the CPU thermal sensor cools too quickly, causing the fan to react, while other parts still absorb heat and trigger a shutdown.
The CPU cooler and overall cooling setup aren't specified. A more powerful cooler is needed for the 7900x. Ensure case ventilation works well with an air cooler if available, and adjust liquid pump speed accordingly. HWInfo64 might miss short-term temperature spikes at the default polling rate of 2000mS; reduce it to 200 or lower. If cooling is limited, use BIOS Curve Optimizer in the PBO section to apply a negative voltage bias (typically -20 to -30). For immediate assistance, consider enabling ECO mode in BIOS, which can significantly reduce temperatures with minimal performance impact.