Cooling Test - does heatpipe orientation have a noticeable influence on thermals?

[REVOCCASES]

Might also be fun to test this with a variable (but static in test) load - How well does the change in orientation affect a 50W processor versus a 75W versus a 100W processor? This can be done fairly easily by modifying the processor's current limits in BIOS.

I kinda did this in my previous tests. Not by limiting the current in BIOS but using different static load scenarios resulting in different power draws of the CPU.

For my Ryzen 3800X I found that there is no performance impact as long as the temperature stays below 81C. Once it reaches that value the CPU (BIOS / motherboard) starts to reduce clock speeds / voltages. So my conclusion is: any orientation is fine (performance wise) as long as you (your cooler) can keep your Ryzen below 81C.

Not sure where that limit is for Intel CPUs or different graphic chips. Maybe we can still collect some more test results here.

 

[thewizzard1]

@tinyitx @REVOCCASES Loading the heatsink above it's normal thermal capacity, to demonstrate it's (unreasonable, yes) capacity for failure at an orientation - The reason heatpipes lose performance at some orientations is because the liquid boiling off at the 'hot' end recondenses at the 'cold' end and flows back via capillary action, wicking against gravity even through warm copper. Too hot of copper, the liquid has problems overcoming gravity and capillary action and can even pool via gravity at the 'cold' end of the heatsink, which will cascade thermal failure of the heatsink at a certain thermal load, by lack of working fluid at the 'hot' end.

It might require more power, anything less than the rated load of a heatsink might only have a small degree (pun intended) of difference - Much larger thermal loads (100w, 150W, 200W even?) will really show what effect orientation has at its worst. The processor doesn't matter, I just know my own tests (testing a Clevo laptop heatsink), with Intel I was very easily able to tailor my thermal load via the BIOS.

 

[REVOCCASES]

@tinyitx @REVOCCASES Loading the heatsink above it's normal thermal capacity, to demonstrate it's (unreasonable, yes) capacity for failure at an orientation - The reason heatpipes lose performance at some orientations is because the liquid boiling off at the 'hot' end recondenses at the 'cold' end and flows back via capillary action, wicking against gravity even through warm copper. Too hot of copper, the liquid has problems overcoming gravity and capillary action and can even pool via gravity at the 'cold' end of the heatsink, which will cascade thermal failure of the heatsink at a certain thermal load, by lack of working fluid at the 'hot' end.

It might require more power, anything less than the rated load of a heatsink might only have a small degree (pun intended) of difference - Much larger thermal loads (100w, 150W, 200W even?) will really show what effect orientation has at its worst. The processor doesn't matter, I just know my own tests (testing a Clevo laptop heatsink), with Intel I was very easily able to tailor my thermal load via the BIOS.

Now I got your idea.

At some point heatpipes will stop to transfer heat and rather act like an insulator. Would indeed be interesting to see at which wattage this happens considering different orientations.

But I guess your CPU will first reach it's thermal limit and start to throttle before reaching this point.

BTW, here is a video from GN how heatpipes are made: