amdgpu SMU driver if version mismatch overrides power limit

I've got 6700S laptop GPU that has 100W power limit provided by ASUS' custom design or whatever. On Windows it uses 100W power limit just fine. But on Linux with **amdgpu** it starts for a couple of seconds with 100W after system boot and by ~25th second it turns it down to 80W which is its vanilla power limit. The value is confirmed by cat /sys/class/drm/card0/device/hwmon/hwmon4/power1_cap. I've tried several kernels: 6.2, 6.5, 6.7. Also I tried several mesa versions: 23.1, 24.1. Also I manually upgraded **linux-firmware** for **amdgpu** specifically - no luck. Dmesg shows this output when 100W to 80W change happens (supposedly): [ 22.496680] amdgpu 0000:03:00.0: amdgpu: RAS: optional ras ta ucode is not available [ 22.513799] amdgpu 0000:03:00.0: amdgpu: SECUREDISPLAY: securedisplay ta ucode is not available [ 22.513802] amdgpu 0000:03:00.0: amdgpu: SMU is resuming... [ 22.513806] amdgpu 0000:03:00.0: amdgpu: smu driver if version = 0x0000000f, smu fw if version = 0x00000013, smu fw program = 0, version = 0x003b2f00 (59.47.0) [ 22.513810] amdgpu 0000:03:00.0: amdgpu: SMU driver if version not matched [ 22.565134] amdgpu 0000:03:00.0: amdgpu: SMU is resumed successfully! Where should I dig to allow this GPU to have power1_cap_max to be 100W? UPD: It seems like this happens when this GPU is being sent to low-power state and then it resumes. By default it boots with 100W limit just fine. Then it's being suspended (or smth) due to inactivity - iGPU is used. When the system needs this GPU back it resumes it and then it overrides the limit!