Having a weird GPU issue that I'm trying to investigate with the help of HWInfo64

[MiniSpip]

What happens: Random slowdowns / micro-freezes in all games, 1-2 seconds. During those micro-freezes, the GPU Core load drops to 0 but memory clock, GPU clock, GPU Power stay level, and at the same time there's a small CPU spike. The spike is more noticeable when looking at the per Core usage.

Which combination of HWInfo64 sensors could help me find out what's happening ? (CPU ? Motherboard ? GPU specific ?)



Important Note: I did not have those issues at all with my previous PC (4790K, same graphics card). On that older machine, I did many, long, gaming sessions without any problem. In a way, the main difference between the 2 machines is CPU/Motherboard.

--------------------------------------------------------
Long Version below.

Temps are OK: ~60°C GPU (fans at 50%) and ~70°C CPU Package while on CPU/GPU intensive games.

Games are installed on a separate (non-OS) NVMe. Previously (see note below), games were on HDD.

Troubleshooting I have tried:

• HAGS On and Off
• Resizebar in BIOS On and Off
• Fullscreen or borderless
• DDU (3 times), drivers re-install with all settings left on default
• Gaming on second OS (dual-boot) with a barebones fresh Win install. Same issue.
• Changing CPU affinity and using BIOS settings to avoid E-Cores

Win10 fast startup is disabled, all GPU stability tests (Furmark, MSI Kombustor, 3DMark...) pass just fine.

Setup:

• GPU: RTX4070 Dual, no overclocking, with drivers: 537.42, installed after DDU
• CPU: 13900k, no overclock
• Mobo: Asus Z790 Hero, latest BIOS 1501
• RAM: G.Skill Trident Z5 64GB (2x32GB) DDR5, XMP 1 enabled (6000 Mhz), no overclock
• PSU: Be Quiet! Dark Power 13 850W (ATX 3.0)
• OS: Windows 10 Pro, build 19045, clean install