I built a new PC that has been randomly crashing for nearly two months. The crashes produce different errors, including IRQL_NOT_LESS_OR_EQUAL, INTERRUPT_EXCEPTION_NOT_HANDLED, KERNEL_SECURITY_CHECK_FAILURE, SYSTEM_SERVICE_EXCEPTION, AMD display-driver failures involving amdkmdag.sys, a MediaTek Wi-Fi driver failure involving mtkwl6ex.sys, a USBXHCI failure, and one WHEA Event 18 reporting a processor-core machine-check exception. I also had an earlier black-screen live dump through the AMD display path.
The system is running Windows 11 25H2 with a Ryzen 5 7500F, a used Gigabyte Radeon RX 5700 XT OC, an Asus B650M AYW-WIFI motherboard, one 16 GB DDR5-4800 CL40 memory module, a TeamGroup 1 TB Gen4 NVMe SSD, an 850 W Cooler Master MWE Gold V3 power supply, and a 1 TB Seagate hard drive. The monitor and graphics card are reused; the rest of the system is new.
I have tried many AMD graphics-driver versions, clean chipset and Wi-Fi driver installations, a clean Windows installation, MemTest86 for five passes, OCCT tests for the CPU, RAM, VRAM, GPU, and power supply, and Cinebench. None of those tests reported an error. The crashes happen randomly and usually while playing Battlefield 1 or other games, although Dota 2 has never triggered one. The USB, Wi-Fi, and audio-related crashes began after a BIOS update. BIOS settings are otherwise stock, except that I forced the graphics slot to PCIe Gen 3. I have not tried FurMark because I am concerned about damaging the used GPU.
Could this be a faulty GPU, CPU, motherboard, PCIe connection, storage device, driver, or something such as EA Anti-Cheat, EA App, SignalRGB, or the latest BIOS? What is the best way to identify the failing component?
3 Answers
The single 16 GB memory stick is not automatically defective, but it prevents dual-channel operation and is worth replacing or testing with a matched two-stick kit on the motherboard's memory compatibility list. Run the system at conservative JEDEC settings, reseat the GPU and power cables, and test the graphics card in another known-good system if available. If the crashes follow the RX 5700 XT, that is strong evidence against Windows or the game launcher. If the GPU works elsewhere, the motherboard, CPU, PCIe slot, or power delivery becomes more likely. A normal one-hour stress test does not rule out intermittent faults, especially heat, transient power, or PCIe errors.
The variety of bugchecks makes one specific driver less convincing. A bad or unstable hardware path can corrupt memory and make whichever driver happens to be running report the crash. The WHEA Event 18 is especially important because it points toward a CPU or PCIe-level hardware error, although a single event is not conclusive. The used graphics card, motherboard PCIe slot, CPU, and storage should all remain suspects. Check the full WHEA event details and inspect the actual minidumps rather than relying only on the displayed faulting module.
Start by collecting the files in C:WindowsMinidump and configuring Windows to create Small Memory Dumps if the folder is empty. Multiple dumps can show whether the crashes consistently involve the AMD GPU, Wi-Fi adapter, USB controller, or unrelated kernel corruption. Also test with unnecessary hardware and software removed: disconnect the hard drive, uninstall SignalRGB and similar hardware-control utilities, temporarily disable the onboard Wi-Fi and Bluetooth, and use a basic wired network connection if possible. Since the crashes began after the BIOS update, compare with a known stable BIOS version using the board manufacturer's documented rollback procedure.
Related Questions
Lenovo Thinkpad Stuck In Update Loop Install FilterDriverU2_Reload