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
You saying they (noctua) listen to engineers rather than marketing people?

I guess that's why they produce products that actually actually work and look like vomit.

Something like that I guess, yeah.

But don't blame the look on the engineers, that certainly came from some marketing guys.
 
  • Like
Reactions: Goatee

Skripka

Cat-Dog Perch Manager
May 18, 2020
443
543
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...

That is because TDP has wildly different meanings to different people in different companies. AMD's 'TDP' numbers aren't even measurements--they're artificial numbers calculated from randomly chosen constants. And that is before how 'new' CPUs behave differently compared to the old stuff.
 

tinyitx

Shrink Ray Wielder
Jan 25, 2018
2,279
2,338
A quite dated thread (orginal in Chinese) with an actual test (CPU: i5-2500K and Coolermaster Vortex Plus).
Interesting read.

 
  • Like
Reactions: REVOCCASES

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,057
3,331
www.revoccases.com
That is because TDP has wildly different meanings to different people in different companies. AMD's 'TDP' numbers aren't even measurements--they're artificial numbers calculated from randomly chosen constants. And that is before how 'new' CPUs behave differently compared to the old stuff.

Yes, that's what I mean. Noctuas list makes much more sense both for Noctua and the Customers. Most manufacturers just printing a TDP rating on the carton and then people start wondering why they can't keep their CPUs cool...
 

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,057
3,331
www.revoccases.com
A quite dated thread (orginal in Chinese) with an actual test (CPU: i5-2500K and Coolermaster Vortex Plus).
Interesting read.


Nice find. Now just need to find out what kind of heatpipes this cooler used. Not sure they already used sintered heatpipes in 2010.
 
Last edited:

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,057
3,331
www.revoccases.com
Here comes another test with the IS40 and 3800X:

In my last test I have run the IS40 at 84W which is about 15% below its official rating. Now I wanted to see what happens if you run it more close to 100% (100W).

For this test I was more curious about the influence on CPU performance (clock speeds) because this is what finally matters, right?

So here we go... Settings: fan @ 100%, Ambient: 18C, Kombustor CPU Burner 16/16 Threads, CPU Standard Settings


Test 1: Heatpipe Orientation "0 degrees"
CPU Temperature: 82C +/-1C
CPU Power: 96W +/-2W
Clock: 4.074MHz +/-5 MHz




Test 2: Heatpipe Orientation "90 degrees"
CPU Temperature: 82C +/-1C
CPU Power: 96W +/-2W
Clock: 4.073MHz +/-5 MHz




Test 3: Heatpipe Orientation "90 degrees, bends facing up"
CPU Temperature: 82C +/-1C
CPU Power: 96W +/-2W
Clock: 4.073MHz +/-5 MHz




Test 4: Heatpipe Orientation "90 degrees, bends facing down"
CPU Temperature: 82C +/-1C
CPU Power: 96W +/-2W
Clock: 4.073MHz +/-5 MHz




Test 5: Heatpipe Orientation "upside down"
CPU Temperature: 86C +/-1C
CPU Power: 88W +/-3W
Clock: 4.024MHz +/-5 MHz





Summarizing I noticed only slight differences in thermals and CPU clocks for orientation 1 to 4, all within a tolerance of +/-1C and +/-5 MHz during the test. Nothing I personally would care about. With auto settings the CPU does a good job to keep maximum possible clock speeds and holds a stable temperature & power consumption. Again, only with the cooler upside down the CPU starts to throttle noticeable, so this orientation does clearly not work well. Note that with other coolers you might see different results. Looking at the thread @tinyitx linked above, I think it all comes down to the quality of heatpipes used and I am surprised that a budget cooler like the IS40 seems to perform well in all common orientations.
 
Last edited:

tinyitx

Shrink Ray Wielder
Jan 25, 2018
2,279
2,338
Summarizing I noticed only slight differences in thermals and CPU clocks for orientation 1 to 4, all within a tolerance of +/-1C and +/-5 MHz during the test. Nothing I personally would care about. With auto settings the CPU does a good job to keep maximum possible clock speeds and holds a stable temperature & power consumption. Only with the cooler upside down the CPU starts to throttle noticeable, so this orientation does clearly not work well. Note that with other coolers you might see different results. Looking at the thread @tinyitx linked above, I think it all comes down to the quality of heatpipes used and I am surprised that a budget cooler like the IS40 seems to perform well in all common orientations.
Again, Test 3 does not show a temp or clock difference. This reconfirms the previous good news.
And I reiterate that an end user does not really know what type of heat pipes is used in a display card or in an CPU cooler until/unless he cuts it up, which is, of course, not feasible. So, how to tell without cutting it up?

The ID Cooling IS40 shows that quality and price do not necessarily directly correlate. IS40 is a budget cooler in terms of price, for sure. But, ID Cooling does not appear to skim on the quality of heat pipes.
Maybe one is so used to equating price of a cooler (eg Noctua's) with quality and so when one comes across a lower priced cooler (eg ID Cooling brand), we tend to think of it as sub-quality? Perhaps, it is just that ID Cooling has lower profit margin than Noctua's and is selling their coolers at a lower price point? /offtopic
 

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,057
3,331
www.revoccases.com
I reiterate that an end user does not really know what type of heat pipes is used in a display card or in an CPU cooler

I believe most medium to higher end coolers (and graphic cards) from well known brands use sintered heatpipes these days. But yes, you are right, you'll never know for sure unless you test it like I did.

Or if you are lucky you can find some marketing stuff like that on the manufactures website:

"Combined with grooved and copper powder sintered interfaces, the composite heatpipe design efficiently improve thermal transfer of heat between two solid interfaces and increase cooling capacity up to 30%"

 
  • Like
Reactions: tinyitx

tinyitx

Shrink Ray Wielder
Jan 25, 2018
2,279
2,338
"Combined with grooved and copper powder sintered interfaces, the composite heatpipe design efficiently improve thermal transfer of heat between two solid interfaces and increase cooling capacity up to 30%"
I am not a heatpipe expert and cannot comment much. But, I suppose, this heatpipe design must be special enough for Asrock to bring it up and mention it in their ad. If it is common among other manufacturers, then Asrock is just stating the obvious.
 

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,057
3,331
www.revoccases.com
I am not a heatpipe expert and cannot comment much. But, I suppose, this heatpipe design must be special enough for Asrock to bring it up and mention it in their ad. If it is common among other manufacturers, then Asrock is just stating the obvious.

I remember a couple of years ago all VGA manufacturers putting info like that about their heatpipes on their websites but it seems nowadays it's difficult to find. This makes me think it has already become standard to use sintered heatpipes on graphic cards. 🧐
 

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,057
3,331
www.revoccases.com
Just found this on Reddit...

Unfortunately it's missing details about test conditions etc... from my POV all of these high end cards should be using sintered heatpipes. Maybe the length of the heatpipes makes a difference?

Anyways, interesting to read that bottom facing I/O orientation seems to work better according to the comments.

 

robbee

King of Cable Management
n3rdware
Bronze Supporter
Sep 24, 2016
855
1,335
@REVOCCASES that comment about the IO on the bottom is me. I'll do some tests tonight by putting my case upside down so the IO ends up on top to see if I can confirm what the reddit OP mentioned. Feel free to ask me for specific setup or tests.
 

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,057
3,331
www.revoccases.com
@REVOCCASES that comment about the IO on the bottom is me. I'll do some tests tonight by putting my case upside down so the IO ends up on top to see if I can confirm what the reddit OP mentioned. Feel free to ask me for specific setup or tests.

Yeah, I was just thinking of you and your new case. I wanted to test it myself in my MID2, but since I'm using an AIO on top and only have a RTX2080S .. 😅

I would go for Kombustor / Furmark and let temps settle in with normal orientation, then turn it around and see what happens.

Looking forward to the results. :)
 

robbee

King of Cable Management
n3rdware
Bronze Supporter
Sep 24, 2016
855
1,335
Ok, I benchmarked my Asus 3080 TUF OC in some orientations. Ran Furmark for about 10 minutes per orientation per test, which was enough for the temperature to stabilize. My case doesn't have any side panels yet so it's pretty much open air.

Here are the results with stock fan profile:

OrientationTemperatureFan speedFan RPM
I/O on bottom67°C61%1320 RPM
I/O on top76°C75%1960 RPM
Horizontal, fans facing up68°C62%1360 RPM

Next, I tried a fan speed normalized test to test if the heatpipe was really the bottleneck. If this were to be true, the I/O on top orientation would not become a whole lot hotter with a slower fan speed, no? Let's take a look:

OrientationTemperatureFan speedFan RPM
I/O on bottom65°C65%1500 RPM
I/O on top81°C65%1500 RPM

But it does. And this makes the difference between both orientation painfully obvious.

So I think we can conclude that, with this heatsink design and this amount of power draw, heatsink orientation makes a lot of difference!
 

riba2233

Shrink Ray Wielder
SFF Time
Jan 2, 2019
1,630
2,102
www.sfftime.com
Ok, I benchmarked my Asus 3080 TUF OC in some orientations. Ran Furmark for about 10 minutes per orientation per test, which was enough for the temperature to stabilize. My case doesn't have any side panels yet so it's pretty much open air.

Here are the results with stock fan profile:

OrientationTemperatureFan speedFan RPM
I/O on bottom67°C61%1320 RPM
I/O on top76°C75%1960 RPM
Horizontal, fans facing up68°C62%1360 RPM

Next, I tried a fan speed normalized test to test if the heatpipe was really the bottleneck. If this were to be true, the I/O on top orientation would not become a whole lot hotter with a slower fan speed, no? Let's take a look:

OrientationTemperatureFan speedFan RPM
I/O on bottom65°C65%1500 RPM
I/O on top81°C65%1500 RPM

But it does. And this makes the difference between both orientation painfully obvious.

So I think we can conclude that, with this heatsink design and this amount of power draw, heatsink orientation makes a lot of difference!

Thanks, awesome test! To me io at the bottom makes more sense since it is also better for cable routing.
 
  • Like
Reactions: REVOCCASES

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,057
3,331
www.revoccases.com
Ok, I benchmarked my Asus 3080 TUF OC in some orientations. Ran Furmark for about 10 minutes per orientation per test, which was enough for the temperature to stabilize. My case doesn't have any side panels yet so it's pretty much open air.

Here are the results with stock fan profile:

OrientationTemperatureFan speedFan RPM
I/O on bottom67°C61%1320 RPM
I/O on top76°C75%1960 RPM
Horizontal, fans facing up68°C62%1360 RPM

Next, I tried a fan speed normalized test to test if the heatpipe was really the bottleneck. If this were to be true, the I/O on top orientation would not become a whole lot hotter with a slower fan speed, no? Let's take a look:

OrientationTemperatureFan speedFan RPM
I/O on bottom65°C65%1500 RPM
I/O on top81°C65%1500 RPM

But it does. And this makes the difference between both orientation painfully obvious.

So I think we can conclude that, with this heatsink design and this amount of power draw, heatsink orientation makes a lot of difference!

Great test, thank you!

Seems I/O facing downwards is the way to go in vertical cases, not only for cleaner cable management but more importantly for better thermals.
 

thewizzard1

Airflow Optimizer
Jan 27, 2017
333
248
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.
 

tinyitx

Shrink Ray Wielder
Jan 25, 2018
2,279
2,338
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.
The above test was done on a GPU. I do not get how varying a CPU wattage limit brings anything interesting to the table. Would you elaborate please?