Cooling the Pi 5: The truth about a good cooler test and how to stress the CPU

[@coliniking]

Nice video. Thanks for exercising stress-ng. The CPU stressor in stress-ng is a generic compute mix, so it may not reach maximum thermal limits. One may find better results when exercising combined floating point compute, cache and memory operations such as the matrix stressor or the fp stressor using one of the many fp-method options.

[@Volker-Dirr]

Thank you for your feedback. So I am not totally wrong if the master of stress-ng write me such a nice answer. I saw your GitHub repository, but I must admit that I didn't read all your different tests. It was just a bit luck that I discovered this stuff. I just saw other KZbinrs using stress-ng and the measured temp was totally different from timetable generator FET on a Raspberry Pi 5. So i tried to check the reason for it. Here are now only 10 minutes left, so I will turn off my computer now. Happy new year!

[@elmedinnyt]

Sehr cooles Video! Ich habe mich früher immer gefragt warum beim Spielaffe spielen auf meinem damals alten Laptop immer so heiß wurde😂. Dann hatte mir dass mein Vater erklärt gehabt. Aber sehr informatives Video! P.S. Ich habe auch ein neues Video gepostet, ich kann sie wenn sie wollen pushen.

[@Volker-Dirr]

Ich habe es gesehen. Wieder sehr viele Views. Aber nur 17 Sekunden. Da wird es schwierig auf die Watchtime zu kommen. Ehrlich gesagt habe ich dein Video nicht verstanden. Ich kenne weder den Mann, noch das Logo auf der Kappe.

[@stefanp4258]

It is possible to set the fan PWM speed fixed at level 3, would give better comparable results.

[@Volker-Dirr]

Yes, that sounds like a simplification that will make it easier to compare. I am currently not 100% sure if there might be other side affects by that. So maybe, just to be sure, it will be more save to measure it not only with stage 3, but for every stage. Then the max temperature will be interesting. And if it hits the thermal limit at 80°C then maybe the time to reach that will be interesting. If the stage is always changing, like in the default settings, then one variant might be to measure how much percent of the time it will stay in the higher stage and how much time in percent will stay in the lower stage. Not easy to measure all this stuff.