Think again, the pump does not need to be controlled, the flow hardly affects the temperatures.
Control by power consumption is as bad as control by CPU/GPU temperature.
The temperature of your components depends on the water temperature, which in turn is directly related to the ambient temperature, fan speed and radiator area.
The water temperature changes slowly in relation to component temperature or power consumption.
The best control variable is therefore the water temperature in the system.

Apart from the bad decision it is of course possible to use the power consumption as a control variable, on my 3700X the package power is displayed directly in the aquacomputer hardware monitor, if this is the case with your CPU I can't tell, if it doesn't HWInfo64 can help.

Translated with www.DeepL.com/Translator (free version)

Zitat

But, the water temperature changes only a few degrees Celsius before and after cooling the CPU.

Don't you think it's too hard to control a fan in this range?

No, because the water only heats up slowly this control is the easiest and most sensible.

If you put your CPU under full load, the CPU temperature will jump to the value you can cool down with your loop and your water temperature, from there the CPU temperature will rise slowly in sync with the water temperature without your fans going into a tailspin.
If you control your PC according to the power consumption, it will sound more like a vacuum cleaner, because every load on the CPU will immediately make the fans turn up without any effect on the cooling, because the water only heats up und cools down slowly so these fast speed changes are worthless.
Translated with www.DeepL.com/Translator (free version)

@ no_3: an additional thought on linking fan RPM to processor activity

• a processor spins up
  
• fan RPM immediately increases for unaffected coolant currently in the radiator
  
• the processor generates more heat
  
• the water block draws additional heat into the coolant
  
• depending on loop size/complexity, there is some time delay before the hotter coolant reaches the radiator
  
• the processor ramps down
  
• fan RPM are reduced
  
• the hotter coolant now reaches the radiator, but fan RPM have already dropped
  

Linking fan RPM to CPU activity or status, may actually be detrimental to a coolant loop's performance.

Man's will is his kingdom of heaven.
I can only tell you what works for in water cooling for many years and explain why.

This I have done, if you want to control it differently do it and learn from what you have done ;-)
I am curious about your report.

Zitat von »no_3«

Hello InfoSeeker
Certainly there is that problem too. Fan control that takes into account water flow delay is required to solve this problem.

Fortunately this can be addressed automatically by placing a temperature sensor at the inlet to the radiator, and adjusting fan RPM per coolant temperature rises and falls.

Below is a snapshot of a chart I run to follow the relationship between power in and fan RPM, mostly to check on my fan curve design.
Magenta data is power out from the PSU, and yellow data is resultant fan RPM.
I also tossed coolant/ambient delta and calculated radiator heat dissipation in for kicks.

In this snapshot I was running at around 625 watts, dropped to around 300 watts for a period, then back to the 625 watt level.