Swapping out the stock TR Frozen Notte 360 fans with Super Flower Megacool - Best tests to run?

Before starting this, I expected to see a significant improvement in the performance of the TR FN 360 AIO because the SF MC fans are that much better. However, there were a few questions I had regarding radiator size, flow and density, and the pump speed. The noise from the pump is negligible, so I ran it at 100%. I also tried two different fans on exhaust. See below for specs.


It definitely seems like the pump's power makes a difference. I ran my tests 3 times with Superposition and 3DMark, monitoring with HWInfo64 and, unless those two tests are not the best tests with which to evaluate the cooling results, it didn't matter whether I used the stock TR fans or the SF ones. I thought maybe it was because I had that Everflow on exhaust, so I put a Noctua A12x25 and still got basically the same results. I had the pump running at 100% for all tests, the exhaust on silent and the AIO fans on turbo.


I've got a lottery-losing 12700k, TR BCF and TR Frozen Notte 360 front-mounted, tubes up on both ends. My XFX Speedster Merc 319 6800 XT is mounted at a 45 degree angle facing up on an EZDIY GC mount sitting on the PSU shroud.


At no time does the CPU throttle, but the GPU hot spot exceeds 100o C.


Prior to running the test series (SP, SW, PR & TS), I ran SP 3 times to heat up the loop.

Published specs:

Pump:

The pump runs at 5,300 RPM, >=1,500 mL/min, 1.6M max lift. Again, running at 100%.

AIO fans:

The TR fans have a max of 2,000 RPM, 72.37 CFM and 2.87 mmAq.

The SF fans have a max of 3,300 RPM, 151.87 CFM and 7.14 mmAq.

Exhaust:

The Cooljag Everflow R121232BUAF has a max of 3,000 RPM, 133.91 CFM and 6.03 mmAq.

The Noctua NF-A12x25 has a max of 2,000 (mine goes up to 2,100) RPM, 102.1 CFM and 2.34 mmAq


I ran the SF with the Cooljag, and I ran the TR fans separately with the Cooljag and the Noctua.


The results were basically identical. There was NO SIGNIFICANT difference for the CPU. This suggests to me that if you've got a strong enough pump, that's more important than the fans, combined with the right rad. This is not what I expected to see! However, I have 7 more 360mm AIOs that I'm going to repeat this test with, so we shall see. The next one will be the one with the worst pump & rad.

Possible confounding factors:

• The fans are on the inside of the rad, pulling air in through it instead of pushing air in.


• The front can hold 140mm fans, so there're gaps on either side of the rad as a result of it being 120mm wide.


• The side and top mounting positions are open aside from the filters they come with.


• Perhaps the positive pressure combined with the openings provides enough places for the heat to escape.


• Perhaps these are not the best tests since they depend heavily on the GPU and, when combined with the other factors, provide misleading results.

If anyone has any theories, I'd love to hear them, especially since the STS video clearly showed that push-pull made a difference.

Would I be better off running Prime95 and/or something else?

NEXT TEST:


I'm not sure which of these has the weakest pump and rad combo. Any thoughts? My guess is that it's the Azza or Sama. Does anyone know? Some only have RPM,

Enermax Aquafusion ADV 360

DeepCool LS720

EK Nucleus CR360

Sama CY360*

Azza Cube 360*

be quiet! Pure Loop 360


I ordered the LF3 but it won't be coming until Monday most likely. If STS can get better results by doing a push-pull on a 280, then I should be able to see results with better fans...?

Conundrum


I have a conundrum that I can't explain (I have 2 theories but I'm probably wrong): my tests are repeatedly coming up with deltas below 0.











I've tried 4 different devices to read the ambient temperature and 3 of them give me almost the same readings, while the 4th is a few degrees higher so I don't even use it. Still, I get results up to 6o C below the lowest ambient readings. I stuck a thermocouple in the back of my case above the PSU shroud and it's showing about a 7o C higher temp than ambient.