My PC has been randomly crashing or freezing for several weeks, usually while gaming but occasionally during lighter use. In demanding games, the monitors go black, audio may continue briefly, and the fans suddenly ramp up to 100%. Sometimes only the game crashes, but more often the whole system becomes unresponsive or reboots. There is no consistent workload or timing: it can run for hours, or fail within 5–10 minutes.
I've checked Event Viewer, which only shows Kernel-Power Event ID 41 after the unexpected shutdown. I've also tried disabling XMP and lowering memory speeds, replacing the RAM, changing GPUs, testing different graphics driver versions, repeatedly using DDU for clean installations, disabling overlays, updating the BIOS, adjusting the page file and power settings, and trying various registry changes.
The problem began months ago after upgrading from an RX 6700 XT to an RX 7800 XT, and it continued after switching to an RTX 5070 Ti. I'm wondering whether faulty memory or a motherboard/slot issue could still be responsible, since one of my G.Skill DDR4 kits showed warning signs in Windows Memory Diagnostic. I plan to test the memory more thoroughly and possibly use its lifetime warranty.
System specifications:
- CPU: Ryzen 7 5800X
- GPU: ASUS TUF RTX 5070 Ti OC
- Motherboard: Gigabyte B550 AORUS Elite V2
- RAM: G.Skill Trident Z 32GB DDR4, 2×16GB, 3200 MHz CL16
- PSU: ASUS TUF Gaming 850W 80+ Gold
- CPU cooler: Thermalright Phantom Spirit 120 EVO
- OS: Windows 11 Pro
What would be the best way to isolate whether the RAM, motherboard, GPU, power delivery, or another component is causing these crashes?
1 Answer
The memory is worth investigating first, especially since Windows Memory Diagnostic reported errors. Test each stick separately, then test each motherboard slot separately. Run a more thorough memory test such as TestMem5 with the Absolute configuration rather than relying only on the built-in Windows test. Try both the default memory settings and XMP, because a kit that fails only with XMP may indicate instability rather than a completely dead module. If errors follow one stick, replace or warranty the kit; if they only appear in one slot, the motherboard or CPU memory controller may be involved.
I ran the built-in memory test several times and saw definite warning signs. The newer kit hasn’t shown the same errors so far, but I’ll test it one stick and slot at a time with TestMem5 before ruling memory out.