I built this PC about a month and a half ago, and it has been randomly crashing once every few days. The crashes happen during gaming, file transfers, voice chat, idle use, and even while Windows is downloading updates. The screen appears black rather than blue.
System specs:
- Motherboard: ASUS Prime B850-Plus WiFi
- GPU: ASUS Prime RTX 5060 8GB
- CPU: Ryzen 5 9600X
- PSU: MSI MAG A750GL PCIE5 750W
- SSD: WD 1TB SN5100 M.2
- RAM: Neo Forza Trinity RGB 16GB DDR5-5200 CL40, with EXPO enabled
- Cooler: DeepCool AG400 Plus
The first crashes seemed connected to the MediaTek MT7922 Wi-Fi driver. Updating it to an older version reduced the frequency, but did not stop the crashes. More recently, crashes have often happened while extracting large archives or transferring tens of gigabytes to an old external hard drive. Transfer speeds sometimes drop from around 38 MB/s to 5 MB/s, although disk usage remains near 50%.
The crash messages have included MEMORY_MANAGEMENT (0x1A), SYSTEM_THREAD_EXCEPTION_NOT_HANDLED (0x7E), SYSTEM_SERVICE_EXCEPTION (0x3B), and errors mentioning ntoskrnl.exe or afd.sys. CrystalDiskInfo reports no SSD problems, and Windows Memory Diagnostic found no RAM errors. All drivers and Windows updates are current. The system is running Windows 11 version 26H1, build 28000.2804.
Could the Windows version, Wi-Fi driver, EXPO memory profile, external drive, or another hardware component be responsible? Should I install an older Windows release, or is there a better way to identify the actual cause?
2 Answers
The file names shown in a crash report usually are not the real culprit. ntoskrnl.exe is the Windows kernel, and afd.sys is part of the networking stack; both can appear when another driver or unstable hardware causes memory corruption. The most useful next step is to collect the crash dumps from C:WindowsMinidump. Several dumps can be analyzed together and may identify a specific driver or pattern. If the folder is empty, configure Windows to create small memory dumps and reproduce the issue.
I would start by removing variables before blaming the motherboard. Install a stable mainstream Windows release rather than an unusually new preview-style build, and perform a clean installation of the chipset, network, graphics, and storage drivers from the hardware manufacturers. Also disable EXPO temporarily and run the RAM at its default settings. Memory Diagnostic is not a reliable guarantee that DDR5 is stable, especially under load. If crashes continue at stock settings, test each memory module separately and check the external drive and its USB cable or enclosure, since the transfer-related failures could involve that device or its driver.
I had assumed the kernel was directly responsible because it appears so often, but I’ll collect the dump files and test with EXPO disabled before replacing anything.