H100x temperatures

Hi

My 1st post here, so be gentle.

I am an owner of Ryzen 5 3600 cooled by H100x.

This CPU is easily overclockable to 4.5GHz@1,1V at all cores.

I can play games, run memory tests without any issues, but running Prime95 will cause CPU to reach really high temperatures.

Tests are made at 4.2GHz@1.06V to avoid CPU damage.

idle (20W) 37C

4K (110W) 83C

36K (120W) 90C

480K (90W) 67C

CPU temperatures are from Ryzen Master, estimated power from CPUID HWMonitor.

Room temp is 23 deg.

These temperatures are reached almost immediately (ca 5s) after load is applied, then the temperature is slowly rising by 1-2C most likely while the coolant temp is rising and then stable for as long as load is applied (2h tested for 90W and 110W - 90C@120W is too warm for my comfort).

Cooling down to idle temp takes also only a few sec.

Radiator is not too warm, unless I stop the fans, then temperature is slowly rising, so heat is transferred from the block - it is not the issue.

Pump is showing 4000rpm, but I have no idea about real flow and I applied the paste using different methods (1 dot, 5 dots, X).

Block is firmly sitting on CPU and screws are really, really tight.

Ideas?

These temperatures are reached almost immediately (ca 5s) after load is applied, then the temperature is slowly rising by 1-2C most likely while the coolant temp is rising and then stable for as long as load is applied (2h tested for 90W and 110W - 90C@120W is too warm for my comfort).

Cooling down to idle temp takes also only a few sec.

This is your evidence the liquid/radiator heat dissipation part of the cooler is working fine. The "instant on" CPU temp is an indicator of CPU quality (at least in regard to thermal properties), voltage, and conductivity. The "slow rise" after over time is the cooler's job. I would expect about +5-6C for your CPU, but it does vary with test and wattage. Coolant rise should be a watts in vs watts out kind of mechanic, if everything else stays the same.

The H100x won't let you see things like coolant temp and the like, but you can estimate it. As you correctly deduced, the slow rise after is the coolant heating up and you can measure that change for fixed load or static CPU tests like Prime, Linpack, CPU-Z bench.

I think if you are having a discussion about X pattern vs multi-dot, you are chasing 1-2C in any context. I don't have a similar AMD CPU to test, so I can't make comparisons for the initial temperature change. I do think that is the part to focus on. Is my voltage too high? How does the differential between idle CPU temp and that 100% instant temp compare to others? Most people see a 35-50C "differential" between those values, but it is voltage dependent and somewhat unique to each processor.