My Windows 11 Pro PC has become nearly unusable after weeks or months of increasingly frequent crashes. Chrome and Opera may crash shortly after opening, the keyboard sometimes disconnects and reconnects, and the system can blue-screen while idle or within minutes of booting.
The system has an Intel Core i5-12600K, Radeon RX 7800 XT, ASRock B760M Pro RS WiFi motherboard, 32GB of G.Skill DDR5-6000 memory, and about 1TB of internal storage. There is no CPU or GPU overclocking or undervolting, and XMP is disabled.
Recent crashes include BAD_POOL_CALLER, CRITICAL_PROCESS_DIED, SYSTEM_SERVICE_EXCEPTION, SYSTEM_THREAD_EXCEPTION_NOT_HANDLED, DRIVER_IRQL_NOT_LESS_OR_EQUAL, and DRIVER_VERIFIER_DMA_VIOLATION. Different dumps have mentioned cbfilter24.sys, FLTMGR.SYS, amdkmdag.sys, dxgmms2.sys, Netwtw14.sys, and ntoskrnl.exe. I previously disabled a Fortect filesystem filter after cbfilter24.sys appeared in a dump, and an external hard drive is currently disconnected after a DMA-related crash occurred soon after it was connected.
Windows Memory Diagnostic found no errors, CHKDSK reported no bad sectors or file-record problems, and I used DDU before reinstalling the AMD graphics driver. Driver Verifier is currently disabled. However, SFC repeatedly reports corruption it cannot repair, DISM fails with error 0x800f0915 because repair content cannot be found, and Windows' in-place repair option fails with error 0x8007000d.
Disabling the graphics card did not completely stop the crashes, although amdkmdag.sys and dxgmms2.sys have appeared in several dumps. I am trying to determine whether the cause is faulty RAM, unstable memory settings, the GPU or another driver, storage or Windows corruption, the motherboard or USB controller, or a combination of problems. What should I test next before replacing parts at random?
4 Answers
Check the internal drive's SMART health and detailed CrystalDiskInfo values, not just whether CHKDSK found bad sectors. Storage problems can cause application crashes, damaged system files, and CRITICAL_PROCESS_DIED. Leave the external HDD disconnected while testing, and reconnect it only after the base system is stable; also update the BIOS and chipset, network, USB, and graphics drivers from the hardware manufacturers rather than relying on random driver packages.
Start by updating the motherboard BIOS and loading its default settings. Even with XMP disabled, check what memory speed and timings the BIOS is actually applying. If crashes continue, test each RAM stick separately, using one module at a time in the recommended slot. A clean result from Windows Memory Diagnostic does not completely rule out intermittent memory or memory-controller problems.
The various driver names may just be victims of the same underlying memory, storage, or kernel corruption rather than several unrelated bad drivers. Collect several actual minidump files from the Windows Minidump folder and analyze them with a proper debugger; a list from a basic dump viewer is not enough to identify the root cause. Also review the SFC log to see which files cannot be repaired.
The failed SFC and DISM repairs make a clean Windows installation a useful dividing line. Back up your data, disconnect unnecessary USB devices and secondary drives, then install Windows fresh on the internal drive with only essential hardware attached. If the fresh installation still produces unrelated BSODs, that strongly points to hardware or firmware rather than Windows itself.
Because the system is unstable, copy the dumps while in Safe Mode if necessary and keep backups of important data before changing BIOS or storage settings.