Fan configuration test, surprising result

[RacA]

Hi all,

Recently, I have been experimenting with different fan configurations for my PC since my CPU temperatures rose significatly while playing CPU demanding games.
Frankly I got the best results while doing everything "wrong" by the common knowledge about pc cooling, so I wanted to share them with everyone who ever had such a dilemma.
Please note that those results are most likely totally dependant on my case model, but chances are that if you're running a similar setup to mine, this could work for you too.

First, my setup:
CPU: R7 1700 @ 3.9GHz, 1.26V (yes, it's stable, guess that's the lottery win)
RAM: Corsair Vengeance  RGB 16GB 3200Mhz (2x8gb dimms)
GPU: Gigabyte RX 5700XT Gaming OC
MOBO: Gigabyte Aorus B450 Elite
Case: Silentium PC Gladius GD8 (also Tecware Bifrost in some parts of the word)
CPU cooler: Silentium Fera 3 RGB - simple tower cooler - per specs should be able to dissipate 180W heat
Case Fans 6x Silentium PC Sigma PRO fans - nothing pro about them, supposed to be static pressure, 38,5 CFM, 1200rpm

Temperature measurements taken after ~30 min of Battlefield 1, 64 player conquest - Known to be CPU demanding, it actually generates higher temps than OCCT test, but this might be due to additional GPU load in the game.

Configuration 1: 3x front intake 2x top exhaust 1x rear exhaust, CPU - Spikes to 83C; GPU - 71C

Originally I went for the most logical fan layout
Problem with this case is it has a very little clearance for the front fans, back exhast has no obstacles, top fans have bit more room to breathe than front. In result I ended up with negative air pressure.
My CPU temps were bad and final conclusion was that top exhaust fans are stealing air from CPU cooler

Configuration 2: 3x front intake 2x top intake 1x rear exhaust, CPU - Spikes to 76C; GPU - 71C

Common knowledge is that you should always go with natural air convection (hot air rises naturally).
Seems though that even cheapo fans absolutely negate this effect. This has been confirmed by Pudget Systems in this article.
It looks like this config was working way better at the start of the test but temps were gradually increasing to 76C on the CPU and 71C on the GPU.
Easy guess is that single fan was not able to cope with that much air being taken in and airflow in the case was a mess.

Configuration 3: 3x front intake 1x forward top intake 1x rear top exhaust 1x rear exhaust, CPU - Spikes to 74 C; GPU - 69C

Would you look at that, I decided to do the stupidest thing ever (as per every forum post and article I have found) and mounted intake and exhaust on the same (top) case panel.
All sources say that this will cause turbulence in case and that those two fans will suck air from each other. To minimize this effect I have moved them as far apart as possible.
I was hoping that the top fan will feed additional air into the CPU tower and top exhasut in the back will help to suck out hot air directly behind it. Result? Even lower CPU temp, suprisingly GPU temp went down too.

(Those results were also reflected in OCCT stress test but are way more pronounced in a real gaming scenario).

TLDR: The least recommended fan configuration dropped my CPU temp by 9C and GPU temp by 2C.

[Samfisher]

The only reason why config 1 doesn't work is cos the top exhaust fans are moving the hot GPU air through the CPU block.  It's only the preferred config if your GPU is a blower style, or you are using an AIO and there's no CPU block to warm from the GPU.

[RacA]

16 minutes ago, Samfisher said:

The only reason why config 1 doesn't work is cos the top exhaust fans are moving the hot GPU air through the CPU block.  It's only the preferred config if your GPU is a blower style, or you are using an AIO and there's no CPU block to warm from the GPU.

It does make sense, but if you'll look around forums/articles on case fan placements, configuration 1 is always being considered as the best/most optimized.

[alyen]

I consider configuration 1 the best after you remove the top fan that is closer to the front for that type of CPU air cooler setup. If I have enough fan headers, I also make the top fan RPM slightly lower than the rear exhaust fan to allow more air flowing through the CPU cooler.

 

I haven't tested configuration 3, but I can't see it being effective enough to add the extra fan especially considering most people tend to try to make things more quiet. It seems more of an extreme way to keep things cooler with more noise.

[Sir0Tek]

I wonder what would happen when one (the one to the right) or both top fans are removed/disabled and the upper intake in the front as well.

[Kc7vwc]

I'd recommend checking the mount on your CPU cooler. I've got a Ryzen 2700 and with the stock Wraith Spire cooler, I'm barely hitting 68-70°c under load. 

I had an Enermax T50 cooler on it that I've had to remove for other reasons, but that one had Max temps in the 50s. 

Your tower cooler should be doing better than the stock cooler. 

 

My usual workload is World of Tanks/Warships, Overwatch & Rocket League - along with streaming those games. (loving Ryzen for x264 encoding too)

[LogicalDrm]

20 hours ago, Samfisher said:

The only reason why config 1 doesn't work is cos the top exhaust fans are moving the hot GPU air through the CPU block.  It's only the preferred config if your GPU is a blower style, or you are using an AIO and there's no CPU block to warm from the GPU.

This is only partly correct. GPUs effect on CPU cooling is really insignificant. You can test this by switching CPU coolers orientation. I have, and recommend every one to do their own research on best cooling setup.

 

The main reason why 1 doesn't work is because front-most top fan is ducking air directly from top-most front fan. If thermophysics is a thing, removing that top fan would make notable difference.

 

Same reason is behind why 2 and 3 work. CPU is getting most out of fresh air intakes.

[RacA]

50 minutes ago, Kc7vwc said:

I'd recommend checking the mount on your CPU cooler. I've got a Ryzen 2700 and with the stock Wraith Spire cooler, I'm barely hitting 68-70°c under load. 

I had an Enermax T50 cooler on it that I've had to remove for other reasons, but that one had Max temps in the 50s. 

Your tower cooler should be doing better than the stock cooler. 

 

My usual workload is World of Tanks/Warships, Overwatch & Rocket League - along with streaming those games. (loving Ryzen for x264 encoding too)

Cpu cooler mount is good, I chose to test with Battlefield 1 because of those temp spikes that made me worry in the first place. My temps are sitting at 70 now on the OCCT stress test. Also note that cpu is oc'ed to 3.9 all core (even though voltage is relatively low).

[RacA]

16 hours ago, alyen said:

I consider configuration 1 the best after you remove the top fan that is closer to the front for that type of CPU air cooler setup. If I have enough fan headers, I also make the top fan RPM slightly lower than the rear exhaust fan to allow more air flowing through the CPU cooler

Yeah, I think part od the problem was that exhaust in front of cpu cooler was taking air out before it reached the tower.

 

I decided to leave the fan in the front as intake since those front fans are very close to front panel and possibly not pushing enough air. I might test it with that fan off someday.

[gasolin]

What about configuration 1 with a 240mm aio in the top,exhaust and the rear fan at the top as an exhaust fan?