As we previously stated the ALF 2’s water block was rather decent, but Asetek (who makes the vast majority of AIOs on the market) was not worried… as Arctic was still working with the same assumptions Asetek (and CoolIT, and Cooler Master, and… and… and) all made. Namely an AIO waterblock should be as small as possible. Which makes zero sense when one thinks about it logically. After all, even if one was to double the Z-height of the ALF 3’s “large” combination block it still would be tiny compared to even entry-level “tower” air coolers.
In either case this “smaller is better” assumption meant that a lot of compromises had to be made when it came to the design of the pump. Basically, with the ALF 2 and (basically) all the rest, an AIO’s water pump is centrally located and is mounted in a horizontal orientation. As such water is sucked in from the inlet tube, forced to do a 90-degree bend, pushed through the pump, then another 90degree bend down to the micro-fin cooling plate level, then another 90-degree bend, forced through the copper cold plate, then another 90-degree bend and up and into the outlet tubing. That is a lot of 90-degree bends. Each 90-degree bend in a loop increases friction, which in turn makes it harder for the pump to push the water through the radiator and the cold plate.
Compared and contrast that with the ALF 3. The water exits the inlet port, immediately enters (and cools) the water pump, then is pushed down to the micro-fin cold plate height, then and only then does it encounter a 90-degree bend… where it then flows over said microplate, slams into a second 90-degree bend… but then exits the water block via the outlet tube. Which is fewer 90-degree bends. This means less pressure is required to push the water through all that tubing, water channels, and water block.
Equally important is by rotating the pump 90 degrees on its axis Arctic was able to use a bigger pump. Bigger in that it has more surface area for the impeller blades. Bigger in that it has… well… bigger bearings. Bigger in that the pump engine can have more copper windings. All of which means (potentially) lower noise, (potentially) longer lifespan, and (potentially) better overall performance. Continuing in this vein is the fact that Arctic does not make a bunch of different-sized pumps for their ALF series. Rather, they make one. One pump that has been optimized for (aka big enough and strong enough to handle) the 420-class ALF 3… and is simply recycled for the smaller ALF 3 variants. This neatly explains why a piddly little 280-class AIO has a pump that can pull 4 watts of 12v power (for point of reference a DC-LT pulls 5 watts and has been used in many a custom loop over the years). We sincerely doubt anyone will care about an extra watt of power (or two), but everyone will care when that (relatively) oversized pump is used in an AIO with a radiator that is 2/3rds as long as the 420 is built to handle. Since it is the rad where most of the “back pressure” comes from that is a good definition of the word “overkill”. Overkill in a good way. Overkill in a way that allows you the owner of the ALF 3 280 to run it slower than it “needs” to in the 420 or even 360… which in turn means either a higher flow rate (when needed) or lower noise (when not).
Now if they would only stop worrying about block height so much, and put a big boi DDC equivalent pump into their loop, that would as a bonus push the inlet/outlet ports up higher, we would be truly happy. We make mention of this as this new and massively improved water block does have one major downside to its oversized dimensions. Namely, it is so big it overflows the CPU area and right into the PCIe slot zone. Worse still, the ALF 3 does not respect PCI-SIG ‘no-go’ zones with regard to z-height restrictions. To be precise it intrudes upon the topmost PCIe slot’s area with what feels like not (just) aggressiveness in its intrusion…. it feels almost personal. Almost as if PCI-SIG stole the ALF 3’s cat and they took that personally.
Yes. That is a bit hyperbolic. Yes. The most likely reason is not spite, and rather when the ALF 3 was still on the drawing board that topmost PCIe slot was pretty much useless and rarely used… and that was probably the reasoning behind ignoring PCIe standards. Sadly in between the days of yesteryore and today that zone has become extremely important… as the topmost PCIe slot has been modded into a PCIe 5.0 x4 M.2 2280/2210 slot. AKA the typical system’s “C Drive” M.2 port. AAKA is the port where the hottest running M.2 drive resides and needs as much cooling as it can get. Sadly, it is not going to get as such as it should when it lives in ALF 3’s house.
To be precise, with the AMD ‘offset’ baked directly/natively into the ALF 3, the inlet/outlet ports invade the M.2 zone by a whopping ~16mm. To be fair, even if it was just disrespecting PCI-SIG by a mere 1mm, the end result would be the same: decreased 3rd party M.2 cooling solution compatibility. Significantly reducing it as this overhanging of said M.2 port will restrict M.2 heatsinks to ~28mm in height (and 25 is a more comfortable hard “safe” limit).
To put this restriction in perspective, a Sabrent Rocket/Colossus sits about ~25mm above the M.2 PCB; an Acidalie/Elecgear is ~30mm; a Thermalright HR10 is about 42’ish; the Thermalright HR-09 Pro is about 75mm. These four represent some of the most popular aftermarket M.2 heatsinks in North America right now. Of the four only the Sabrent will work with the ALF 3. This translates to a good 8 to 10 additional degrees on a PCIe 5 x4 SSD when pushed hard compared to the HR10, add another 5 or so degrees when compared to the HR-09 Pro… and add another 5 or more when dealing with actively cooled variants of said chonky bois.
Make no mistake it is not just 3rd party heatsinks that will be affected by this arbitrary z-height limitation. For example, the MSI Spatium M570 Pro’s stock heatsink will not fit underneath the ALF 3’s block. In the coming years, more and more enthusiast grade SSD’s will come with heatsinks that make the M70 Pro look small… as PCIe 6 M.2s are expected to have a TDP in the 25-30 watt range and require massive passive cooling abilities.
Thus, for some, this will not be a deal-breaker for right now or even 2025. After all, most M.2 drives come with lower than 28mm height heatsinks, and most motherboard’s “stock” M.2 heatsinks are equally “low profile”. The operative word is most… as “the times they are a changing”.
Motherboards are starting to come standard with excellent, robustly built M.2 heatsinks that are also rather tall. ASUS’ RoG CrossHair Hero X860E is an excellent example of this necessary trend. A trend that will trickle down from the uber-high end to the more moderately priced motherboards in the coming years. We feel very comfortable in stating that if forced to choose between the two, few will opt to keep the ALF 3 over their new ~1K CAD motherboard. We seriously doubt many will even pick the ALF3 over a $300 motherboard. Thus turning this from a potential deal breaker… into the ALF 3 quickly migrating on to experienced builders ‘do not buy’ list in the coming years. Hopefully, before that doomsday scenario occurs a future ALF 3 revision will have… less “romantic intent” towards PCI-SIG standards and allow builders to use the best of the best M.2 heatsinks and/or whatever motherboard their customer wants.
