Cooling Noctua NH-P1 passive cooler. To have or not to have?

[Phuncz]

I'm amazed it doesn't tip over ! It probably needs a heavy GPU on the other side for that I guess. I love the build !

 

[tinyitx]

I'm amazed it doesn't tip over ! It probably needs a heavy GPU on the other side for that I guess. I love the build !

Tipping over is not going to happen.

Quick back-of-envelope calculation below.

BTW, it is hot today. Observatory says it will hit 34C. Right now it is 32C outside. I think I will turn on the air conditioning a bit to get a more comfortable condition to test the NH-P1. Numbers should come out in a couple hours.

 

[tinyitx]

Brief testing summary
Testing software: Metro Exodus benchmark run on a continuous loop of 10 rounds. Ultra Preset (at 2560x1600)
9900K in default settings
Ambient temp = 27C
CPU Idle temp = 36C

CPU Max temp = 73C

Next, I put on the NF-A12x25 LS PWM fan (which is recommended by Noctua if one wants additional cooling capacity) to see how much more headroom will there be. The fan is installed blowing horizontally toward the motherboard. This fan was bought as an insurance. When fanless fails, this fan will come in. It is always good to have a Plan B.
At 20% (minimum rotational speed), CPU max temp = 68C
At 100% (~1200 rpm), CPU max temp = 59C

So, it seems the additional headroom is not too small. IIRC, the fan will add another 30W cooling capacity on top of the NH-P1's.

But, I am quite happy with fanless. The fan will be on standby mode indefinitely.
So much for now. Kind of tired....

 

[rfarmer]

Brief testing summary
Testing software: Metro Exodus benchmark run on a continuous loop of 10 rounds. Ultra Preset (at 2560x1600)
9900K in default settings
Ambient temp = 27C
CPU Idle temp = 36C

CPU Max temp = 73C

Next, I put on the NF-A12x25 LS PWM fan (which is recommended by Noctua if one wants additional cooling capacity) to see how much more headroom will there be. The fan is installed blowing horizontally toward the motherboard. This fan was bought as an insurance. When fanless fails, this fan will come in. It is always good to have a Plan B.
At 20% (minimum rotational speed), CPU max temp = 68C
At 100% (~1200 rpm), CPU max temp = 59C

So, it seems the additional headroom is not too small. IIRC, the fan will add another 30W cooling capacity on top of the NH-P1's.

But, I am quite happy with fanless. The fan will be on standby mode indefinitely.
So much for now. Kind of tired....

73C is a very reasonable temp and you can easily live with that, nice to have insurance just in case.

 

[REVOCCASES]

Brief testing summary
Testing software: Metro Exodus benchmark run on a continuous loop of 10 rounds. Ultra Preset (at 2560x1600)
9900K in default settings
Ambient temp = 27C
CPU Idle temp = 36C

CPU Max temp = 73C

Next, I put on the NF-A12x25 LS PWM fan (which is recommended by Noctua if one wants additional cooling capacity) to see how much more headroom will there be. The fan is installed blowing horizontally toward the motherboard. This fan was bought as an insurance. When fanless fails, this fan will come in. It is always good to have a Plan B.
At 20% (minimum rotational speed), CPU max temp = 68C
At 100% (~1200 rpm), CPU max temp = 59C

So, it seems the additional headroom is not too small. IIRC, the fan will add another 30W cooling capacity on top of the NH-P1's.

But, I am quite happy with fanless. The fan will be on standby mode indefinitely.
So much for now. Kind of tired....

Results look really good so far. I may still change my mind and get one...

If you find the time, could you please do me the favor to run "MSI Kombustor CPU Burner" stress test with the P1 in passive mode and monitor average power draw & temperature in HWinfo?

Many thanks.

 

[tinyitx]

73C is a very reasonable temp and you can easily live with that, nice to have insurance just in case.

Yes. 73C is quite acceptable to me. I use a gaming benchmark instead of a synthetic test like OCCT or Intel XTU as this reflects more a realistic daily usage.

Results look really good so far. I may still change my mind and get one...

If you find the time, could you please do me the favor to run "MSI Kombustor CPU Burner" stress test with the P1 in passive mode and monitor average power draw & temperature in HWinfo?

Many thanks.

This Kombuster should give higher max temp.
But, I will do it.
Ja, nein problem.

By ''average power draw'', you mean the ''CPU Package Power''. Right?

 

[REVOCCASES]

This Kombuster should give higher max temp.
But, I will do it.
Ja, nein problem.

Thank you! Unfortunately I don't have Metro Exodus in my games library - but still I'm trying to get an idea if I could run my 3800X passive cooled during gaming with reasonable temps.

I find Kombustor CPU Burner actually pretty useful as it will not stress your CPU as hard as Prime95 or similar torture tests. From my experience the (thermal) results are more comparable with running an actual, CPU demanding game.

By ''average power draw'', you mean the ''CPU Package Power''. Right?

Exactly.

 

[tinyitx]

Thank you! Unfortunately I don't have Metro Exodus in my games library - but still I'm trying to get an idea if I could run my 3800X passive cooled during gaming with reasonable temps.

I find Kombustor CPU Burner actually pretty useful as it will not stress your CPU as hard as Prime95 or similar torture tests. From my experience the (thermal) results are more comparable with running an actual, CPU demanding game.



Exactly.

I used to use OCCT a lot since my early PC years. Its auto generating graphs is a nice-to-have feature. But, in recent years, I tend to use these synthetic tests only for the purpose of testing the stability of an overclock while to use a gaming benchmark to check actual day-to-day usage thermals.

I admit I have little experience with Kombustor. Anyway, I have downloaded and installed the latest version. Will do some testing tonight.

 

[biopunk]

 

[tinyitx]

@REVOCCASES

Kombustor CPU Burner test (2560x1600 if that matters)
Ambient temp = 27C

Time-------CPU Package Power (Ave)-----CPU Package Temp (Max)
10min--------------70W----------------------------74C
15min--------------76W----------------------------75C
20min--------------80W----------------------------75C

I stopped at 20-minute mark as the temp seems to be in a steady state.

 

[REVOCCASES]

@REVOCCASES

Kombustor CPU Burner test (2560x1600 if that matters)
Ambient temp = 27C

Time-------CPU Package Power (Ave)-----CPU Package Temp (Max)
10min--------------70W----------------------------74C
15min--------------76W----------------------------75C
20min--------------80W----------------------------75C

I stopped at 20-minute mark as the temp seems to be in a steady state.

Thank you very much! That looks actually very promising for the use case I have in mind.

 

[REVOCCASES]

Another test (in German):

Golem.de: IT-News für Profis

 

[tinyitx]

Now that the CPU has its new toy to play with, the CPU's neighbour is not happy. How so?

With the NH-C14S, the 14cm fan blew down towards the motherboard. And the top M.2 SSD received some air from the downdraft (even though the cooling benefit not as big as I would like).
Now, with the NH-P1, this side benefit is gone. M.2 SSD temp creeps up silently to a level that I am not (Edit) happy with. Something needs to be done.

One solution is to install the NF-A12x25 LS PWM fan to blow down again. But this would just totally defeat the original purpose of switching from C14S to P1. The prime directive (ie go fanless) is violated.

Another solution is to ziptie a small (say, 4cm) fan to the underside of the P1 to blow air directly towards the M.2 SSD. But this also violates the prime directive.

Now, I am thinking of replacing the stock Asus heatsink cover that came with the Asus Z390i Gaming by a beefier 3rd party heatsink.
So far, these 2 candidates seem to qualify for the job.
a) Sabrent Rocket

b) Cryorig Frostbit

I suppose, the 9900K has its badass passive cooler, the Samsung 970 Pro wants its own too.
Anyone has used either one before?

 

[tinyitx]

Thank you very much! That looks actually very promising for the use case I have in mind.

You are welcome.
BTW, care to leak some details about what you have in mind?

 

[rfarmer]

Now that the CPU has its new toy to play with, the CPU's neighbour is not happy. How so?

With the NH-C14S, the 14cm fan blew down towards the motherboard. And the top M.2 SSD received some air from the downdraft (even though the cooling benefit not as big as I would like).
Now, with the NH-P1, this side benefit is gone. M.2 SSD temp creeps up silently to a level that I am happy with. Something needs to be done.

One solution is to install the NF-A12x25 LS PWM fan to blow down again. But this would just totally defeat the original purpose of switching from C14S to P1. The prime directive (ie go fanless) is violated.

Another solution is to ziptie a small (say, 4cm) fan to the underside of the P1 to blow air directly towards the M.2 SSD. But this also violates the prime directive.

Now, I am thinking of replacing the stock Asus heatsink cover that came with the Asus Z390i Gaming by a beefier 3rd party heatsink.
So far, these 2 candidates seem to qualify for the job.
a) Sabrent Rocket

b) Cryorig Frostbit

I suppose, the 9900K has its badass passive cooler, the Samsung 970 Pro wants its own too.
Anyone has used either one before?

Both look like they could do a decent job, let us know which you choose and how it works.

 

[REVOCCASES]

You are welcome.
BTW, care to leak some details about what you have in mind?

My last fully passive build was in 2008 with a XFX 7600 GT Fatal1ty in some big ATX case - I like to see if I can put something (for today's standards) similar powerful together in SFF. The biggest challenge will probably be the GPU. Palit makes a passive cooled GTX1650 but it's more or less impossible to get - so I need to think about alternatives...

 

[BaK]

Both look like they could do a decent job, let us know which you choose and how it works.

As long as they cool down the controller and not the NAND: https://smallformfactor.net/forum/threads/my-second-streacom-db4-build.16405/post-244723

 

[tinyitx]

As long as they cool down the controller and not the NAND: https://smallformfactor.net/forum/threads/my-second-streacom-db4-build.16405/post-244723

Yes, Nand prefers 'warmer' while the controller prefers 'cooler'. But the question is, how cool is 'cool' and how hot is 'warm'?
The controller can take care of itself by throttling while the Nand cannot. Maybe, this is why Samsung Magician only reports the Nand temp. If one wants to check the controller temp, one needs to use HWinfo or alike software.

Anyway, I more or less rely on Samsung Magician telling me whether the Nand is within its 'normal' operating temperature. So far, I have seen my 970 Pro creeps into mid to high 40's (High rating) and even low 50's (Too High rating) when the system is idling or under light loading (eg browsing). I am not comfortable about this Samsung's comment on the nand temp.

I am leaning towards getting the Cryorig Frostbit. It seems the primary heatpipe is oriented in such a way that will cool the controller more than the nand, ie more cooling to the controller and relatively less cooling to the nand. So, this sounds like they have thought about the 'controller cool + nand warm' issue.
This pic shows how the heatpipes work:-

 

[GuilleAcoustic]

Oh nice ! What's this chassis ? Also, does the NH-P1 protrudes past the IO shield ?

 

[thelaughingman]

Oh nice ! What's this chassis ? Also, does the NH-P1 protrudes past the IO shield ?

it's the ZS-OC v1 open case. and I believe the P1 does not protrude past the IO shield.