I've been troubleshooting a very specific crash pattern for the past few weeks. The system randomly freezes to a black screen, and the keyboard becomes completely unresponsive—Caps Lock no longer toggles its LED—so it appears to be a full hardware or kernel lockup rather than a normal graphics driver crash.
The problem happens almost exclusively during the first 15–30 minutes after a cold boot, while the computer is idle, when launching Steam, or right as Windows finishes uninstalling an update. If the system gets through that initial period, it becomes completely stable for the rest of the day. I can run games and heavy stress tests for more than 10 hours without a crash.
The system uses an Intel Core i5-14600KF with Hyper-Threading disabled, an AMD Radeon RX 7700 XT, an ASUS PRIME H610M-K D4 motherboard, 32GB of Kingston Fury Beast DDR4-2666, a Cooler Master Hyper 612 APEX cooler, and a Chieftec 700W power supply. The motherboard has no VRM heatsinks.
I've already run SFC and DISM, removed a recent Windows quality update, updated the BIOS and GPU drivers, reseated the GPU and memory, changed the RAM slots, and tested the graphics card with FurMark and 3DMark. The GPU remains stable under sustained load, although AMD Adrenalin resets my custom fan curve after a crash and reports an unexpected system malfunction.
I've now replaced the Chieftec PSU with an EVGA 850GQ 850W Gold unit and disabled XMP. A Gigabyte B760M GAMING X DDR4 motherboard is also on the way. Could the H610 board's VRM or low-power-state behavior explain why the crashes only occur while the system is cold or idle?
2 Answers
The cold-and-idle pattern also points to a low-power-state or firmware stability issue rather than a GPU that fails under load. For testing, use completely default BIOS settings with XMP, ReBAR, Hyper-Threading changes, undervolting, and motherboard power-saving tweaks disabled. If the board exposes CPU C-state or ASUS power-management options, temporarily test with those at conservative defaults. The PSU replacement is useful, but the H610 board may be struggling with the 14600KF even if the system passes heavy stress tests.
The H610 motherboard is a strong suspect, especially paired with a 14600KF and no VRM heatsinks. Cold VRMs can behave differently during startup and low-load transitions, and the fact that the machine becomes reliable after warming up fits that possibility. Testing the new B760 board and the replacement PSU should help narrow it down. If the crashes disappear with the B760, the original board’s power delivery is probably the culprit.
That’s what I’m hoping. It worked normally for more than a year, but most crashes happen when I launch Steam shortly after a cold boot. Once it survives the first 15 minutes, it usually runs all day without trouble.
Aside from XMP, ReBAR, and disabling Hyper-Threading, the BIOS settings were already at their defaults. I’ve now disabled XMP as the next stability test.