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

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,057
3,331
www.revoccases.com
Some time ago in one of my threads the question came up if the orientation of the Heatpipes on modern CPU and GPU coolers does have an impact on thermals or not (due to gravity)...


Noctua already did their own tests and confirmed: the orientation of the heatpipes does not matter.



Anyways, @tinyitx and I were curious - so I did a few own tests with my Thermalright Silver Arrow.

General Test Conditions
  • Cooler: Thermalright Silver Arrow
  • TIM: Thermalright OEM
  • CPU: i5-6600
  • Fan Speed: fixed to 500 RPM (+/- 5 RPM)
  • Stress Test: Prime95, Small FFts
  • Ambient: 20C

Test 1: Cooler oriented like it is supposed to be (0 degrees)

CPU Temperature after 30 Minutes: stable 61C +/-1C




Test 2: Cooler flipped 90 degrees

CPU Temperature after 30 Minutes: stable 61C +/-1C



Test 3: Cooler flipped 180 degrees (upside down)

CPU Temperature after 30 Minutes: stable 61C +/-1C



(I also performed a similar test with my RTX2070 and came to the same results.)


My conclusion: There seems to be no noticeable effect on thermals / cooling performance with modern CPU and GPU coolers no matter in which orientation the heatpipes are oriented. Nowadays CPU and GPU coolers usually use sintered heatpipes so that they can be operated in different orientations without measurable impact on cooling performance. The general airflow concept of your build will have a more noticeable effect here.


 
Last edited:

tinyitx

Shrink Ray Wielder
Jan 25, 2018
2,279
2,338
Do you think the relatively whimpy i5 6600 was not really matched well with the beefy 8x6mm monstrosity SA TR4?
The negative gravity effect might still be there but it was just not big enough to come to surface as the SA TR4 has so much cooling capacity. So, even with a reduction of, say, 10% cooling, it still cools the CPU well?
 

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,057
3,331
www.revoccases.com
Do you think the relatively whimpy i5 6600 was not really matched well with the beefy 8x6mm monstrosity SA TR4?
The negative gravity effect might still be there but it was just not big enough to come to surface as the SA TR4 has so much cooling capacity. So, even with a reduction of, say, 10% cooling, it still cools the CPU well?

I think we should have seen at least some effect since I fixed the fan RPM to 500 for all tests so it could reach 60C. I could repeat the test with my 3800X but I don't think there will be a noticeable difference. I mean, I also tested this with my RTX2070 and I could not see any difference in temperatures.

What would be still interesting is testing some very cheap TaoBao tower cooler (which might not use sintered heatpipes).
 

Goatee

King of Cable Management
Jun 22, 2018
738
1,512

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,057
3,331
www.revoccases.com
@REVOCCASES I think I tend to agree with @tinyitx

With heat pipes, you loose capacity as you change orientation. There is a paper here on it: https://www.qats.com/cms/wp-content...ientation_affect_a_heat_pipes_performance.pdf

Given you likely have 8 (given the U design they could be classed as 16 heat pipes) on that cooler each one only has to transport <10 watts of heat on a i5-6600 so the requirement on it is so low I think orientation doesn't matter.

OK, no problem. I will repeat the test.

Should the 3800X be sufficient to create enough heat? Or shall I better test another cooler? I still have the axp90, IS40 and AMD stock cooler.

Maybe the IS40 with 3800X would be a good contrast since this cooler would operate at/above its maximum cooling capacity.
 
Last edited:

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,057
3,331
www.revoccases.com
@tinyitx which CPU and cooler are you running in your current build? Maybe you could flip your case 0 / 90 / 180 degrees like I did and also add some measurements...
 

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,057
3,331
www.revoccases.com
Test #2: 3800X vs. ID-IS40

During this test I learned a thing or two.

If you test with zen2 or zen3 cpu please set the manual clock and voltage, otherwise cpu does what it wants and test won't be very meaningful.

I followed your suggestion and set clock and voltage manually. So far so good but then I noticed that during stress testing the CPU Package Power continuously increased from 80W at the beginning to over 90W after a couple of minutes. Together with the power increase, of course also temperatures continuously increased which made it impossible to use these settings for my test.

After I switched everything back to Auto Settings, Package Power and Temperatures stabilized while maintaining a stable clock of 4.1GHz during the stress test. So, the following results are performed with CPU Auto Settings. Other Settings: fan set to 100%, Ambient: 19C, Kombustor CPU Burner 8/16 Threads (the IS40 can not handle more).


Test 1: Heatpipe Orientation "0 degrees"

CPU Temperature: 77C +/-1C
CPU Power: 84W +/-3W
Clock: 4.1Ghz





Test 2: Heatpipe Orientation "90 degrees"

CPU Temperature: 77C +/-1C
CPU Power: 84W +/-3W
Clock: 4.1Ghz





Test 3: Heatpipe Orientation "90 degrees & flipped"

CPU Temperature: 77C +/-1C
CPU Power: 84W +/-3W
Clock: 4.1Ghz




Test 4: Heatpipe Orientation "upside down"

CPU Temperature: 79C +/-1C
CPU Power: 84W +/-3W
Clock: 4.0Ghz




Summarizing, I did not see any cooling performance difference worth mentioning for orientation 1, 2 and 3. Everything was within a +/-1C tolerance during the tests. But running the IS40 upside down increased temperatures noticeable by two to three degrees compared to the other orientations. I also noticed a slight decrease in Clock Speeds. When I flipped it back into normal position, temperatures dropped steadily back to 77C. Now if this is caused by the gravity factor impacting the heatpipe performance or if it's just because the hot air cant move upwards that easily compared to the other three orientations... I don't know...
 

tinyitx

Shrink Ray Wielder
Jan 25, 2018
2,279
2,338
For Test 4, I agree that it is hard to say if it was due to gravity effect or worse ventilation. Maybe it is a combination of both. But this does not really matter practically as, I imagine, almost no one would install a CPU cooler in such orientation.
What matters is Test 3 as this orientation is seen quite often. But the test shows no performance difference. This is good news!
 
  • Like
Reactions: REVOCCASES

riba2233

Shrink Ray Wielder
SFF Time
Jan 2, 2019
1,630
2,102
www.sfftime.com
Cool, great test, thank you!

Yes, it is normal for cpu power to increase with the temperature if everything else is fixed, I have noticed it with my zen1 cpu as well.
 
  • Like
Reactions: REVOCCASES

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,057
3,331
www.revoccases.com
Yes, it is normal for cpu power to increase with the temperature if everything else is fixed, I have noticed it with my zen1 cpu as well.

I actually didn't know this, so a thing I learned: just leave voltage regulation to the CPU / BIOS unless you want to OC and are ok with higher temperatures
 

riba2233

Shrink Ray Wielder
SFF Time
Jan 2, 2019
1,630
2,102
www.sfftime.com
I actually didn't know this, so a thing I learned: just leave voltage regulation to the CPU / BIOS unless you want to OC and are ok with higher temperatures

Yes, when the silicon gets hotter it has bigger leakage and consumes more power.

These new Ryzens have something similar to nvidia's gpu boost, so they can regulate their voltage and clocks (and thus power) based on the temperature, that is why they can be a bit tricky for measuring thermal performance.
 

robbee

King of Cable Management
n3rdware
Bronze Supporter
Sep 24, 2016
855
1,335
Noctua already did their own tests and confirmed: the orientation of the heatpipes does not matter.

That's weird because I remember seeing something about it in my NH-L12 manual. I went back looking and actually found this:

1610017339706.png

It looks like they advice against the setup you used in test #3 although for you it works well.
 

tinyitx

Shrink Ray Wielder
Jan 25, 2018
2,279
2,338
Doesn't it depend on the type of heatpipes being used ? Here a PDF discussing the wick type vs orientation vs performance topic. The Silver Arrow, for example, uses sintered metal heatpipes, which are more tolerant toward their orientation.
Yes. The main problem is that no (or at least that I know of) manufacturer would state the type of heat pipe used in their product (display card or air cooler). One never knows unless one cuts up a heat pipe. So, one should avoid this adverse orientation, if possible. Or, one can actually do a thermal test to find if there is a performance loss.

Practically speaking, IIRC, the cooling capacity loss is not very big (~10%?). But, psychologically speaking, it just does not feel good when you know your component is not working optimally.
 

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,057
3,331
www.revoccases.com
In general Noctua seem to be cautious regarding performance claims for their coolers. E.g. not giving a general TDP rating but having a compatibility list for all CPUs. Like in the manual they say it MAY and not it WILL reduce performance.

From my feeling Noctua might be right with both statements and orientation 3 may indeed not be optimal but on the other hand it seems you won't notice a real difference under normal circumstances. At least I didn't see it during my tests with the IS40.

I guess if you look close enough and running a cooler near or even over the suggested TDP rating you may notice a more significant difference, at least in synthetic benchmarks.

PS: feel free to repeat the orientation tests with your CPUs/coolers and add your results here. Should be interesting to see if e.g. the NH-L12 really performs different in position 3...