hard to follow the vid without knowing the language. It looked like one gpu was stuck at 925core clocks while all the others were 1125 and this "fix" fixed the clock speed being limited. and restored a substantial amount of hashrate with the same power draw.
This video tutorial shows how to lock core clock value we want to.
In many algorithm, nvidia-smi set lower cclock than we set.
HiveOS cannot handle it. So, we have to do custom settings.
core clock: 1110
memory clock: 1500
mining: quai, eth, abel etc.
You can see each GP has different hashrate values. But not.
Use "nvtool --clocks" commands to see each GPU's clocks.
You will realise each GPU has different core clock even if you set all to the same.
In this video, you will learn how to set core clock values properly.
We will bypass nvidia-smi and HiveOS restrictions.
You will not use the highest power value because you will set core clock.
390W is the highest value. But the GPUS use around 360W-370W
in the video, we can use all needed power acording to the core clock - no restrictions !
Without suctom setting:
The default PL is 370W. But nvidia-smi restricts using 370W if you set core clock.
You can see 390W power consumption if you dont set any core clock.
And you will see SW Power Cap Active problem although GPU use 390W.
