• Discord?! We've had a Discord server for a while, but now we've opened it up to folks who aren't members of our fantastic community forum. Join our discussions by clicking the "Discord" button on the side menu!!

CPU Noctua L12 Single versus Dual Fan Performance

Mark13

SFF Lingo Aficionado
Original poster
Sep 5, 2020
113
128
I decided to do some top mounted fan testing on the Noctua L12 Ghost Edition cooler. I first replaced the included below the cooler B9x25 fan with the newer A9x25 fan before starting as it has both higher RPM range and higher static pressure. I initially wanted to compare the performance of top mounted fans in a dual downdraft configuration, but after noticing the top mounted fans reduced the maximum RPM of the bottom A9 fan, I also decided to compare the performance of the top mounted fans both with and without a lower fan. I also tried two front mounted (RAM side) fans as (case) fans with only a top mounted Scythe Wonder Snail fan for comparison.
Tested:
Noctua L12 Ghost Edition cooler with lower A9x25 fan only
Scythe Wonder Snail
Scythe Kaze Flex Black
Noctua A12x25
Phanteks T30
Noctua A12x15
Arctic P14

I included the Arctic P14 fan as it fit very nicely on top of the L12 cooler when mounted in my 5.9 liter Yiluka YLK01 case. The counter-rotating Scythe fans slowed the A9x25 fan the most.
Using a dual fan configuration with this cooler may well not be worth it. The Noctua A12x25 used alone was the quietest. At 100% PWM.

MakerBeam ITX Open Test Bench
Aorus B550 ITX MB; Ryzen 7 4750G; 32 GB Corsair LPX 3200 RAM; Metal block over NVMe drive not installed; Thermal Grizzly Carbonaut Thermal Pad
No Graphics Card installed
Noctua L12 Ghost Edition Cooler; Lower Fan: Noctua A9x25 (When Installed); Upper Fan (or Case Fan) see below
Open C of cooler faced towards unused graphics card slot. Cooler fins run front (RAM side) to back (IO side) of MB.
Blender Classroom Render & Unigine Valley Ultra Settings
Room Temp 21C; CPU & Case Fan(s) PWM 100%; Temp Sensor TZ10 read 16.8C on all test runs; (Case) fan placed blowing air on RAM side of MB
CPUID HWMonitor Readings:

Upper Fan​
Scythe WS​
Scythe WS​
Scythe KFB​
Scythe KFB​
Noc A12x25​
Noc A12x25​
Phan T30​
Phan T30​
Noc A12x15​
Noc A12x15​
Arctic P14​
Arctic P14​
Scythe WS​
Scythe WS​
Lower Fan​
A9x25​
A9x25​
A9x25​
A9x25​
A9x25​
A9x25​
A9x25​
Case Fan​
Arctic P14​
Scythe KFB​
Blender Render Time​
10:30.08​
10:22.45​
10:21.40​
10:23.94​
10:19.92​
10:23.15​
10:21.00​
10:22.00​
10:20.13​
10:25.49​
10:21.15​
10:24.45​
10:23.26​
11:00.70​
10:21.98​
Temp Sensor 0​
43​
38​
37​
37​
38​
37​
38​
37​
37​
39​
41​
37​
39​
37​
35​
Temp Sensor 1​
44​
40​
40​
39​
41​
39​
40​
39​
40​
40​
43​
39​
41​
39​
38​
Temp Sensor 2 CPU​
74​
68​
66​
67​
67​
67​
67​
66​
66​
68​
69​
66​
67​
66​
65​
Temp Sensor 3 PCI​
21​
21​
21​
21​
21​
21​
21​
21​
21​
21​
21​
21​
21​
21​
21​
Temp Sensor 4​
44​
40​
39​
39​
41​
39​
40​
39​
39​
40​
42​
39​
41​
38​
38​
Temp Sensor 5​
42​
37​
37​
36​
39​
37​
38​
36​
37​
38​
41​
36​
39​
37​
35​
CPU Pkg Max Temp​
74.6​
68​
66.9​
67​
67.8​
67​
67.5​
66.5​
67.1​
68.6​
69.9​
66.8​
67.6​
66.4​
65.1​
CPU Pkg Max Wattage​
89.01​
88.8​
88.85​
89.09​
88.87​
89.06​
88.89​
89.93​
88.85​
89.01​
88.78​
88.82​
88.89​
88.91​
88.83​
NVMe Top 970 Evo +​
43​
38​
37​
37​
38​
37​
38​
37​
38​
39​
41​
37​
39​
41​
37​
I-GPU Max Temp​
64​
57​
58​
57​
58​
57​
58​
57​
57​
58​
60​
57​
58​
56​
55​
CPU Lower Fan Max RPM​
2045​
1934​
1928​
2002​
2002​
1991​
1985​
CPU Top Fan Max RPM​
2331​
2337​
1646​
1634​
2057​
2057​
1979​
2020​
1824​
1767​
1614​
1662​
2327​
2327​
Case Fan Max RPM​
1721​
1573​

After conducting the tests I realized that I also needed to reverse the (case) fan direction and add these results to the updated table:

Upper Fan​
Scythe WS​
Scythe WS​
Scythe KFB​
Scythe KFB​
Noc A12x25​
Noc A12x25​
Phan T30​
Phan T30​
Noc A12x15​
Noc A12x15​
Arctic P14​
Arctic P14​
Scythe WS​
Scythe WS​
Scythe WS​
Lower Fan​
A9x25​
A9x25​
A9x25​
A9x25​
A9x25​
A9x25​
A9x25​
Case Fan​
Arctic P14​
Scythe KFB​
Scythe KFB​
Blender Render Time​
10:30.08​
10:22.45​
10:21.40​
10:23.94​
10:19.92​
10:23.15​
10:21.00​
10:22.00​
10:20.13​
10:25.49​
10:21.15​
10:24.45​
10:23.26​
11:00.70​
10:21.98​
11.0793​
Temp Sensor 0​
43​
38​
37​
37​
38​
37​
38​
37​
37​
39​
41​
37​
39​
37​
35​
36​
Temp Sensor 1​
44​
40​
40​
39​
41​
39​
40​
39​
40​
40​
43​
39​
41​
39​
38​
39​
Temp Sensor 2 CPU​
74​
68​
66​
67​
67​
67​
67​
66​
66​
68​
69​
66​
67​
66​
65​
65​
Temp Sensor 3 PCI​
21​
21​
21​
21​
21​
21​
21​
21​
21​
21​
21​
21​
21​
21​
21​
21​
Temp Sensor 4​
44​
40​
39​
39​
41​
39​
40​
39​
39​
40​
42​
39​
41​
38​
38​
39​
Temp Sensor 5​
42​
37​
37​
36​
39​
37​
38​
36​
37​
38​
41​
36​
39​
37​
35​
36​
CPU Pkg Max Temp​
74.6​
68​
66.9​
67​
67.8​
67​
67.5​
66.5​
67.1​
68.6​
69.9​
66.8​
67.6​
66.4​
65.1​
65.5​
CPU Pkg Max Wattage​
89.01​
88.8​
88.85​
89.09​
88.87​
89.06​
88.89​
89.93​
88.85​
89.01​
88.78​
88.82​
88.89​
88.91​
88.83​
88.91​
NVMe Top 970 Evo +​
43​
38​
37​
37​
38​
37​
38​
37​
38​
39​
41​
37​
39​
41​
37​
37​
I-GPU Max Temp​
64​
57​
58​
57​
58​
57​
58​
57​
57​
58​
60​
57​
58​
56​
55​
51​
CPU Lower Fan Max RPM​
2045​
1934​
1928​
2002​
2002​
1991​
1985​
CPU Top Fan Max RPM​
2331​
2337​
1646​
1634​
2057​
2057​
1979​
2020​
1824​
1767​
1614​
1662​
2327​
2327​
2319​
Case Fan Max RPM​
1721​
1573​
1558​
Case Fan Direction​
Intake​
Intake​
Exhaust​
 
Last edited:

rfarmer

Shrink Ray Wielder
Jul 7, 2017
2,338
2,407
Interesting test. I have the NH-L12 GE in my Ghost, I am using the NF A9x25 Chromax in mine. I mainly got it because it was chromax but the added rpm and static pressure over the NF B9x25 was a bonus. Wish I had room in the Ghost for a top mounted fan. I am really surprised how well the Scythe Wonder Snail performed, especially compared to the Phanteks T30.
 

Mark13

SFF Lingo Aficionado
Original poster
Sep 5, 2020
113
128
Interesting test. I have the NH-L12 GE in my Ghost, I am using the NF A9x25 Chromax in mine. I mainly got it because it was chromax but the added rpm and static pressure over the NF B9x25 was a bonus. Wish I had room in the Ghost for a top mounted fan. I am really surprised how well the Scythe Wonder Snail performed, especially compared to the Phanteks T30.
That is part of the reason I wanted to run this test. Of course the Wonder Snail is a bit louder at 100% PWM and a couple of hundred RPM faster. I intend to repeat these tests again shortly in a new case that I am waiting on. Will post those results as well.
 

Analogue Blacksheep

King of Cable Management
Bronze Supporter
Dec 2, 2018
683
503
Interesting, I currently have an NF-A12x15 as well as a NF-A9x25 on my L12 Ghost cooler in my Revoccases RCC-MIT1 case.

Managed to get the 120mm fan brackets by emailing Noctua for some.
 
  • Like
Reactions: Mark13 and rfarmer

Mark13

SFF Lingo Aficionado
Original poster
Sep 5, 2020
113
128
Interesting, I currently have an NF-A12x15 as well as a NF-A9x25 on my L12 Ghost cooler in my Revoccases RCC-MIT1 case.

Managed to get the 120mm fan brackets by emailing Noctua for some.
Yes, Noctua will provide fan clips for 25mm thick fans on the L12GE at no charge with proof of purchase - got mine from them as well.
 

hrh_ginsterbusch

Average Stuffer
Nov 18, 2021
61
19
wp-devil.com
Seeing the Phanteks T30 in there: Did you run it in performance mode? Cause that kinda tops out at roughly 2000 rpm, which would explain the numbers. Without performance mode, it could brute-force anything through a wall, with its max 3000 rpm.

cu, w0lf.
 

Mark13

SFF Lingo Aficionado
Original poster
Sep 5, 2020
113
128
Seeing the Phanteks T30 in there: Did you run it in performance mode? Cause that kinda tops out at roughly 2000 rpm, which would explain the numbers. Without performance mode, it could brute-force anything through a wall, with its max 3000 rpm.

cu, w0lf.
No, I just used the standard mode that tops out a bit over 2,000 RPM. I realize this higher range might be useful for some, but not for me when on the desk next to my head.
 
  • Like
Reactions: hrh_ginsterbusch

brushrunner

What's an ITX?
New User
Aug 6, 2022
1
0
I would have loved to see you push the Phanteks to 2400rpm or even 3000rpm, just to see if the heatsink can handle that additional cooling or if it gets limited by its size

Have you tried dual a12x15's as well? would love to see how that compares against a single a12x25 on top