Possible random crashes linked to GPU issues
Possible random crashes linked to GPU issues
My Setup: CPU model AMD Ryzen 7 3800X
Motherboard ASUS Prime X570-Pro (Bios 4408)
RAM 2x Corsair 16GB DIMMs at 3200MHz
Storage: 2x NVMe drives (1TB primary, 240GB); 3x SATA (2x 6TB, 1x 120GB SSD)
GPU Gigabyte 1660Ti (6GB)
Monitors: 3 full HD displays at 60Hz
Issues: Occasional random crashes. They might appear suddenly, often freezing the screen—displaying just a few colors or a distorted version of what’s on screen. Sometimes it triggers BSODs (Kmode Exception not managed; IRQL not low; kernel security check failed). Other times it restarts instantly. Most BSODs seem linked to graphics drivers. I tried boosting fan speed but the problems persisted. Recently updated drivers helped temporarily, though crashes continued.
I’ve checked temperatures and ran diagnostics for the CPU. The crashes still happen despite those checks.
Freezing patterns: Sometimes after Windows loads and the desktop appears (even before login). Other times during software tasks—VirtualBox, PHPStorm, browsers, email, Discord. Occasionally while gaming (especially 7 days to die). Even when idle.
No freezing when BIOS is active and everything functions normally. The crashes usually stop once Windows boots fully or after a restart.
Possible causes: Graphics card issues during troubleshooting—maybe I left the PCIe bracket loose, or forgot to secure it properly. Inspected both sides of the card for damage, but found nothing visible (even under magnification). PCIe slot itself appears intact.
My troubleshooting steps:
- Restarted after crashes and restarted again.
- Performed a cold boot (waiting 30+ minutes before trying).
- Disabled Fast Boot and set Startup Repair to off.
- Updated drivers system-wide.
- Removed non-essential hardware (extra USB ports, Wi-Fi/BT card) and drivers.
- Reinstalled graphics drivers.
- Re-seated the GPU in the slot.
- Forced GPU to 3rd generation PCIe speed.
- Checked Event Viewer at crash time—no useful data.
- Followed advice from trusted guides about BSOD codes.
Results: Some days system runs smoothly for over an hour; other days it crashes repeatedly, often within a minute of loading the desktop. Yesterday I experienced video playback issues (black frames, stuttering) while watching a YouTube video. Using Task Manager’s Resource Monitor, I looked for performance links but found nothing—CPU, RAM, GPU usage were normal. I couldn’t retrieve a perfmon report; it keeps crashing around completion time.
Here’s the link to the minidumps: https://www.dropbox.com/s/gvwb5ctaki2hot...p.zip?dl=0
Any guidance would be greatly appreciated.
Have you tried using DDU to uninstall GPU drivers? It might be a good option.
Typical causes for the "Kernel Security Check Failure" error include damaged system files, malfunctioning hardware, or obsolete drivers. https://softwarekeep.com/help-center/how...windows-10 Most often, IRQL NOT LESS OR EQUAL / KMODE EXCEPTION HANDLED alerts appear when there are issues with hardware, drivers, or antivirus software. https://answers.microsoft.com/en-us/wind...20dc478cd0 The IRQL error is a very low-level problem stemming from conflicts between devices connected to your PC. Since it involves interactions among internal and external components, the mainboard, Windows OS, and installed drivers, resolving it can be challenging. Learn more about IRQL on Wikipedia. https://neosmart.net/wiki/irql-not-less-or-equal/ An IRQL is a hardware-independent method that helps Windows prioritize interrupts from the processor. On supported architectures, it directs signals to an interrupt controller, which then manages priority levels. Higher-priority interrupts take precedence, while lower ones wait until the current task completes. Windows assigns unique IRQL values for each processor in a multi-core system, allowing APCs and kernel operations to be interrupted appropriately. https://en.wikipedia.org/wiki/IRQL_(Windows) To address Kmode Exception Not Handled issues, consider these steps: Disable fast startup, update drivers via Windows Memory Diagnostics, and run the update tool. https://www.iobit.com/en/knowledge-how-t...10-122.php
I completed another DDU uninstall/reinstall, and the system feels more reliable. I’ll check further to confirm; otherwise, a fresh Windows installation might be necessary. Appreciate the advice.
Here are some clearer points summarizing your situation:
You experienced multiple crashes during Windows 10 installation and updates. After a clean install of Windows 10, you upgraded to Windows 11 but still faced issues. During setup, the system prompted for a Microsoft account login, which caused unexpected restarts. A crash occurred while logging in to Steam, followed by another crash in PowerPoint due to a system service error.
You suspected RAM problems after these events, especially since symptoms included random freezes and crashes. You ran a Windows Memory Diagnostic, which detected hardware issues but crashed midway. A Memtest86 test showed no problems. You disabled XMP profiles on your RAM and tried overclocking, but the issue persisted.
You removed one RAM dim from your system and re-enabled XMP, which led to another crash with an unexpected kernel trap. After restarting and running a memory diagnostic again, everything worked fine.
To speed up testing each RAM module, you could:
- Use a dedicated memory tester or software that checks individual modules.
- Check for physical damage or loose connections.
- Monitor temperatures and ensure proper cooling.
- Consider running a more detailed diagnostics tool like CrystalDiskInfo or MemTest86 with hardware-specific profiles.
What steps have you tried so far?
The most effective method to check RAM involves testing the machine with one stick at a time. Memory testers frequently overlook problems in DDR4 and newer systems. Regarding dump files, three-quarters of the blame goes to storage-related drivers. The remaining cases seemed to point to RAM, but because of the page file, storage and RAM can appear similar. I’ve noticed false positives in Windows Memory Diagnostic before, so I tend to suspect storage issues more often. If you experienced a crash using just one stick, swap it out and try the other. Should both fail when used alone, consider installing the OS on another drive. Unfortunately, there aren’t reliable tools for assessing SSD health easily. Modern SSDs use SMART diagnostics, which was our usual approach before. With today’s drives, these useful metrics are missing, making general health ratings unreliable.