What case, or none?
Which location in the case, and what are the immediate obstacles(filters, radiators, drive cage, grille, etc) present to the fan?
What fan speeds were compared?
Any noise normalized(to your comfort zone) comparisons done?
This is inside a Corsair 5000x with a Noctua U12A providing the barrier to airflow by mounting it on the left or right, with the air then passing to another Noctua A12x25 mounted at the rear.
All case fans are set to minimum no matter what the temperature is.
As I have already indicated, the iPPC was not tested, hence "wildcard". They're just on the case as front intake (top and bottom).
Fans on the cooler are connected to the mobo CPU headers of the Asus TG H670-Pro wifi D4, two sticks of DDR4, 3 M.2 SSDs, 1 XFX Speedster Swft319 Rx 6800 XT CORE GC, Corsair RM850x PSU, Those fans are set to the default turbo mode in BIOS.
Ambient temperature is recorded in Celsius for each test and deducted from the results to get delta.
Temperatures recorded aside from that are: CPU , CPU Package, motherboard, VRM thermistor, and graphics card (except the 1st test as BIOS doesn't show that), including minimums and maximums, and variance. The average is calculated, as is the temperature range.
There are 3 tests per configuration of 1 or 2 fans, although this poll is only relevant to the positioning of the 4 types of fans.
1, BIOS results (control to suggest lowest possible temperatures achievable while on). Again, this doesn't include the GC as that is not monitored by my mobo.
- Windows 11, with Corsair iCUE open and on the dashboard, Asus Armoury Crate (default), and HWInfo64 to simulate a light load.
- Windows 11, with HWInfo64 and FutureMark's 3DMark Speed Way Benchmark run to simulate a heavy load.
*Of course, startup items are also resident in 2 and 3.
I also did two control scenarios in which the left glass panel was left off, one with both A12x25s, the other with an A12x25 on the left (intake) and T30 on the right (exhaust) because I also wanted to see how pairs work differently on a single tower.
Data is recorded from HWInfo64 for 2 and 3. I turned on logs for the 5 data points and recorded the mins and maxes from each chart.
Data, including variance, was entered into OpenOffice Calc on 3 sheets. Again, average temps and the temperature range for each sensor was calculated by formula. For BIOS, both C and F degrees were recorded as it showed both, but HWInfo64 only showed Celsius, so I had a formula calculate it.
To achieve scores and ranking, I sorted each sheet by each Celsius and averaged result and filled in the rank column. If multiple fans achieved the same temperature, they were given the same rank. Once all ranks were determined, the ranks for CPU and CPU Package were added together to get the CPU score, and the ranks for the mobo, VRM and GC were added together with the CPU Score to get the overall score. Scores were then colored with the best being purple and the worst being white (after red). Sheets were sorted by the CPU score to determine which works best in each position.
My goal is to determine best function (in or out), not noise levels, so I'm not testing noise in any way, nor am I considering fan speeds, wattage, etc. This is purely about pushing and pulling air through a barrier. Thus, my goal is different from all the fan videos I've watched.
I realize this is methodology not ideal, as I don't have any special equipment, but one works with what one has. I have to start somewhere, and that means using what I've got. As long as I test everything consistently with the same methodology, the results will be fairly useful. If some people feel that these results are inadequate, that is understandable as they weren't achieved under laboratory conditions but, again, I don't have those resources. I'm not rich like Linus, nor do I have a massive number of subscribers and donors. Thus, consistently performing the tests in the same way is very important.
I plan to test other fans, and also test a twin-tower cooler (Peerless Assassin) because the results I got when I put two of the fans on the U12A were sometimes surprising. This is more than conjecture and opinions (your answers in the poll will be that, even if you look at SP and AF) because it is backed by data acquired in a consistent manner, albeit under imperfect conditions that Steve @GN would stick his nose up at.
;P Finally, I am not motivated by getting rich (although it'd be nice) and, if companies send me hardware directly I'll be happy to accept it but I won't use it for testing because some companies are unethical and will send "special" units to foil the testing. I also will not be influenced by donations from private or public sources - my results will be the only arbiter of the ranking and will be uninfluenced by the requests of sponsors, companies and individuals to "help" a company by falsifying the results and declaring a product to be of a greater or more special quality than it really is. That is my guarantee, my oath, to everyone. I'd rather shut down entirely than become a con artist influencer like the ones that Coffeezilla and others expose.
Eventually, I will design and build custom equipment to measure airflow and static pressure, dB(A), and temperature, but that will take resources I don't yet have.
It is funny that @USAFRet
said "bottom of the list". You'll see why after enough people do the poll.