Push Vs Pull vs Push-Pull for AIO Liquid Coolers – Does it matter?
|Have you ever noticed that most all-in-one liquid coolers have almost exclusively started shipping with just one set of fans? Like, a 240mm cooler comes with just 2 120mm fans, and only one set of long screws to mount them, even though there’s a perfectly good side to mount another set… What gives? I mean, more fans is more better right? This has to be a conspiracy by BIG FAN to make your coolers less effective and force you to buy bigger and more expensive coolers! Or… maybe it isn’t, and it’s a calculated decision based on a number of nuanced factors, factors I want to explain in this video so let’s get started.
Just to set your expectations, this isn’t a carefully controlled multi-week test run by someone with long hair and years of experience and knowledge in testing this sort of thing. I’m not Steve. I don’t have air conditioning so I can’t actually control the ambient temperature, only measure it, and I purposefully didn’t control things like the fan and pump speed so I could show real world results – what would happen if you stuck some extra fans on your AIO. I did, however, control the CPU which is a Ryzen 9 5900X running with Precision Boost Overdrive set to 185W PPT, 170A EDC and 125A TDC and as well as logging the temperatures I compared the power consumption to make sure that stayed the same across the various runs.
As for the cooler, well that’s this beast, the ARCTIC Liquid Freezer II 360 A-RGB. This comes stock with three P12 PWM A-RGB fans that are hooked up through the same 4 pin PWM fan header as the pump and VRM fan. They are rated for 48.8 CFM of air flow and 1.85mm/H2O of static pressure. It’s a 38mm thick radiator, which is around 10mm thicker than the standard radiators you’d find on say a Corsair H100i, but in theory should be more benefited by more fans. It’s also a fairly high fin density, which again should lend itself to more fans being useful. As for the second side fans, those are the exact same P12s, to keep everything even.
So, does more fans mean more better? Well, in the Gooseberry render it really does look like it! I mean look how much lower the blue push/pull line is than the orange pull, or especially the green push line! Oh, wait sorry I’m being told this is why I recorded ambient temperatures for each run and I should be showing the ambient corrected chart. Apologies, let’s cut to that now.
Ah, that’s much better. The green push and orange pull lines both mix nicely together, and while the blue push/pull line is lower, it’s not by much. In the best case it’s about 1°c lower than the other configurations. In fact, I’m being told we can even zoom in on this to get a closer look, so let’s do that! Now it should be pretty obvious the difference, or really lack thereof. The green and orange lines do clearly intertwine closely, and while the blue line is clearly separated and lower, if you look at the axis you’ll see that the blue line is only an absolute maximum of 1.5°c cooler, but especially throughout the earlier portion is more like 0.5-1°c.
That Gooseberry scene only runs for around 9 minutes on this chip, with this overclock, so what about something longer? Like, Cinebench R23 multi-threaded running on repeat for 30 minutes. That ought to do it. Pause I said, that ought to do it! There. Much better. In this test, interestingly the push and pull lines are reversed with push running a fraction of a degree cooler on the whole. The blue push/pull line is still the coolest, but again if we zoom in. Pause I said zoom in! You can’t find good help these days… You’ll see that the advantage of having more fans is… best case, 2°c. And, if I’m being honest, my ambient temperature measurements were all prior to running either test and I (consistently) ran the Cinebench test AFTER the Blender render, and while I did allow the cooler to return to temperature the ambient very well could have bumped up by a half a degree or so, which may account for some variance there.
Either way, it’s clear that there isn’t a substantial benefit in this setup for running a push/pull configuration as I’d be looking for a clear advantage – something like 5°c or more consistently cooler and you don’t get that here. Now, I can already hear you rushing to the comments to call me a moron – and while I both know you will anyway, and let’s be honest you are probably right to do so – because I didn’t control for fan speed and like OMG that’s the whole point! Here’s the thing. If you’ve watched my review of this cooler, you’ll know that it is absolutely silent. Even with almost 200W being pumped through it over 30 minutes, it’s barely ticking over, so adding more fans – while they will be able to run at a lower RPM thanks to more blades blowing or sucking air through the fins – the added noise of just having twice as many fans counteracts any benefit to the reduced RPM.
The thing is, this behemoth of a cooler is clearly overkill even for this CPU, and it’s not what you are going to be installing in your next gaming PC – ok yes you three people with an insatiable need to have a the ultimate quiet and yet performant PC you can will but the rest of them won’t! For a thinner 240mm radiator, in theory the fans play a bigger part in keeping your parts cool. A big radiator like this has two things going for it, thermal mass and surface area. The larger thermal mass comes from more liquid in the loop to sink the heat away from your CPU and more metal to soak the heat out of the fluid, and the surface area is an entire extra fan’s worth of heatsink fins that the heat can dissipate from. This sort of design, even with this much heat, doesn’t actually need much air moving all that quickly through it to keep things cool.
Compare that to a thinner, 240mm AIO which not only has less fluid, but also has much less surface area for the heat to transfer from water to metal to air, meaning you need more air moving quicker and more forcefully to have it evacuate that heat effectively. That means your fans will be working harder, spinning faster, to maintain the same cooling power. That’s why some 120mm AIOs do still come with two fans in the box, the less surface area you have, the more work the fans have to do to keep air moving over those fins dissipating that heat.
So, is it all some big conspiracy? Are you leaving performance on the table not using all the fans you can? No, not really. In the right scenario you can see some benefit from running push pull, but if your CPU isn’t pumping out all that much heat or you have a monster cooler like this one it’s unlikely to be worth the extra cash – plus look how chonky this thing is with fans on both sides! It’s practically covering half the motherboard, and even with a thinner rad it’s still pretty massive so only do that if you want a touch less fan noise or you are really pushing your cooler to its limits.
It’s also worth noting that practically speaking there really isn’t much of a difference in push vs pull configurations in terms of temps or noise, so my advice there would be mount what’s easiest to get set up and maintain – at the front of a case I’d go pull since it’ll be easier to clean the dust off the front of the radiator instead of having to remove the fans and installation will be easier since you won’t have to fish the long screws through the easily moved fans and then into the radiator which can also move around. If you are putting it at the top, I’d go push for the same reason for installation, plus you get to see the shiny lights clearly rather than only when you are directly below the radiator and can just about see through it.