US Liquid Cooling editorial · Reviewed · 4 min read
An AIO can report a normal pump speed while the CPU still overheats. The number tells you about a rotation signal; it does not prove that coolant is circulating properly, the cold plate is making good contact, or the radiator is getting enough air.
Start with the pattern of the temperature rise. A machine that suddenly races toward its thermal limit in BIOS needs a different investigation from one that runs a little warmer during a long render. Repeatedly launching a stress test will not help identify a cooler that cannot handle a light load.
Is the CPU actually overheating?
Use the temperature limit published for your exact processor, rather than a universal rule such as “anything over 80°C is bad.” Also record whether the CPU is throttling, shutting down, or simply boosting to a high temperature during heavy work.
Note the room temperature, CPU package temperature, workload and approximate package power. If your cooler exposes a liquid-temperature sensor, record that too. Compare the machine with its own previous behavior before comparing it with someone else's screenshot.
NZXT's cooler diagnosis guide separates pump, mounting, fan and radiator problems. That is a more useful starting point than assuming a pump RPM reading clears the whole cooling system.
Check what the RPM number belongs to
Open the cooler manual and follow the cable that provides the displayed signal. A header named CPU_FAN may be reporting a radiator fan, a pump tachometer, or a shared cable's selected signal.
Different products also report speed differently. ARCTIC documents a scaled pump readout on some Liquid Freezer III models. Consequently, a number that looks unexpectedly high or low compared with a different cooler is not, by itself, evidence of failure.
With the computer shut down and disconnected from power, check the required power, control and monitoring connections. On some AIOs the motherboard lead only reports speed while a separate connection supplies the pump. A working screen or RGB lighting does not answer all of those wiring questions.
Use the heat pattern to narrow the investigation
These observations are clues, not definitive tests:
| Observation | What to investigate next |
|---|---|
| CPU becomes extremely hot soon after a cold start | Cold-plate contact, correct mounting hardware, pump power and circulation |
| CPU and reported coolant temperature rise during sustained work | Radiator airflow, fan control, dust and workload relative to cooling capacity |
| Problem appeared immediately after moving or servicing the PC | Disturbed cables, tubing position, mounting and trapped air |
| Fans run, RPM looks normal, but overheating continues | Internal restriction or cooler failure remains possible |
Do not use one warm hose and one cooler hose as a complete diagnosis. Touch is a poor measuring instrument, and some temperature difference is expected when a loop transfers heat.
Inspect the mounting before buying another cooler
If the problem began with installation, confirm the socket bracket, standoffs and backplate against the actual product manual. Check that any protective film was removed and that the mounting sequence was followed. Tightening harder is not a substitute for using the right hardware.
If a remount is necessary, inspect the thermal-paste contact pattern while the machine is powered off. Reapply paste according to the cooler and processor guidance. Avoid turning this into repeated remounts without a hypothesis: write down what was wrong, what changed and the resulting temperature behavior.
CORSAIR's powered-on but overheating checklist provides a useful manufacturer reference for this stage.
Separate radiator airflow from circulation trouble
Inspect dust filters and radiator fins with power disconnected. Check that fans are installed in the intended direction and respond to their controller. A fan spinning slowly at the desktop does not prove the load profile is correct.
Make one change at a time. For example, first correct an obstructed filter, then repeat a workload the machine can safely sustain. Record the same measurements. That gives you evidence; changing the pump profile, fan profile, BIOS power limits and paste simultaneously does not.
Stop testing if the temperature rapidly approaches the processor's limit, protection triggers, or liquid is visible. Leave a sealed AIO closed unless its manufacturer explicitly provides a service procedure. An internal obstruction is normally a support or replacement issue, not an invitation to open the pump housing.
What should you send to support?
Send the cooler model, CPU, motherboard, purchase date, cable arrangement, a short recording of the symptom, and a small temperature log. Include whether the issue happens in BIOS and whether the system previously worked normally.
For replacement planning, read how long an AIO cooler lasts. If wiring is unclear, use our AIO pump header guide. Solve the failure pattern first; radiator size is a separate buying decision.
Sources & further reading
- Is my Kraken cooler failing? – NZXT Support Center ↗
- AIO Cooler:What to do if your AIO cooler powers on but your unit still overheats – Corsair ↗
Explore the company directory → · All liquid-cooling guides →