I recently bought a used PC with an RTX 5070, Ryzen 5 9600X, ASRock B650M-H/M.2+ motherboard, 32 GB DDR5-5600, a Corsair AIO, and a 650 W Seasonic power supply. It ran well when I tested it at the seller's house, but after bringing it home I started getting extreme stuttering in games. GTA V Enhanced and CS2 report decent average FPS, yet frametimes are consistently around 20–50 ms with occasional spikes near 300 ms, and 1% lows are extremely poor.
I have already reinstalled Windows, used DDU for several graphics driver versions, updated chipset and other drivers, tested the memory sticks individually, reset the BIOS and CMOS, tried both EXPO and JEDEC memory settings, confirmed Resizable BAR, checked the SSD, disabled the integrated GPU, and ran GPU and CPU stress tests without reported errors. The BIOS has also frozen several times during startup.
LatencyMon reports sustained latency around 1500–2500 microseconds, with occasional peaks around 16,000 microseconds. The main reported drivers are nvlddmkm.sys, Wdf01000.sys, dxgkrnl.sys, and ntoskrnl.exe.
The most suspicious issue is CPU clock behavior. Shortly after startup, the CPU runs around 4.5–5 GHz, but after a few minutes it can fall to around 2 GHz or even 600 MHz, including during stress tests. The CPU die temperature remains around 40–50°C, while HWiNFO reports the CPU VDD_MISC VRM sensor gradually reaching approximately 99°C. The clock speed drops as that temperature reaches 99°C, and HWiNFO logs HTC Thermal Throttling and PROCHOT EXT Thermal Throttling.
Could the motherboard VRM be overheating or triggering protection despite the CPU itself being cool? Is the basic motherboard design, AIO airflow, cooler installation, BIOS version, or something else likely responsible?
4 Answers
Before replacing parts, remove and reinstall the AIO block, apply fresh thermal paste, and verify that the pump is connected to the correct header and is actually running. Also check that the radiator and case fans are oriented correctly. A cooler that shifted during transport can cause problems, although the reported VRM temperature makes motherboard airflow and power delivery especially worth testing.
The LatencyMon results may be a consequence rather than the original cause. Graphics and kernel drivers often appear at the top when the CPU is suddenly downclocking, because the GPU is left waiting for the processor. It is still reasonable to test USB, audio, network, and other device drivers, but first solve the repeatable CPU/VRM throttle. A fresh USB installation can rule out software, though it is unlikely to fix a hardware sensor-triggered clock drop.
The CPU temperature being low does not rule out a power-delivery problem. The clock drop happening when the VDD_MISC VRM sensor reaches about 99°C, together with PROCHOT EXT, strongly suggests a VRM or voltage-protection throttle rather than ordinary CPU-core overheating. That board has very limited VRM cooling, and an AIO may provide less airflow over the motherboard area than a tower cooler. Try running with the side panel removed and a room fan aimed directly at the VRM area. If the throttling disappears or takes much longer, the board’s VRM cooling is probably the main issue.
Update the motherboard BIOS to the latest stable version and load default settings afterward. Avoid changing several settings at once while testing. Confirm the CPU power connectors are fully seated and inspect the board for anything unusual. Since the BIOS has frozen during startup, the motherboard deserves more suspicion than the graphics card, especially if the system still throttles in a clean operating-system installation.
You can also temporarily set a more aggressive case-fan curve and test Eco Mode or a lower CPU power limit. If reducing CPU power stops the throttling, that further points toward the motherboard power delivery rather than the CPU cooler.