HWCooling as mentioned along with QuasarZone for on application results, HWBusters/Cybenetics for P-Q curves and independent testing of specs AMCA 210. Figure like 10% uncertainty in the rankings for part to part variation and implementation details. HWBusters has additional error as load noise and P-Q curve convexity are neglected—there’s occasional discrepancies up to ~20% in their data, too, but it’s mostly ok. Gamers Nexus’ choice to review the Peerless Assassin is after Thermalright succeeded it with the Phantom Spirit is an odd one and they still haven’t released anything from the Longwin fan tester they bought.
I have an eco mode 7950X under the latter and thermal throttling’s negligible with TL-C12s at ~850 RPM max. If you’re needing to push the Peerless Assassin above that with a 7900 something’s wrong with the mount or case airflow, there’s a typo and it’s actually an unrestricted 7900X or PPT lifted, ambient temperatures are elevated, or something else is off.
I generally see diminishing returns beyond ~30 m/h³ but, yeah, if thermals aren’t good intake flow’s the first thing I check. Exhaust fans have lowered temperatures 5–10 °C on occasion and, while I haven’t found a lot of RPM sensitivity, in front to back air cooled builds I typically spin the exhausts up somewhat to compensate for hex mesh generally being fairly high flow resistance.
My experience is TL-C12 impellers pretty much tie NF-A12x25s in actual use and, unlike the Noctuas, I’ve never had them whine. Thermalright got bought out a while ago and the new owners seem to have switched to bearings that click and don’t last very long, though. Arctic P12s work well as low cost replacements but likely hum audibly above ~1200 RPM.
Source? The on application measurements mentioned above show little to no difference between the two sizes’ noise-normalized airflows at the population level, so it’s really down to parts selection. For fans that usually get shortlisted to buy it’s around a ±25% range depending what you’re comparing in which direction. QuasarZone’s data also shows a consistent pattern of 140 air coolers underperforming 120s on a noise-normalized basis, which is consistent with my experience.
Spring loaded screws are standard for cooler mounts and the implementation you describe is standard for dual tower air coolers. Noctua’s mounting hardware on their desktop parts that I’ve installed has been poor—lots of sticking and binding where screws on coolers costing a third as much turn smoothly.
Lian Li P28s are pretty much performance interchangeable with T30s and should drop in fine, then, though stretching the clips or minor mods to the fans would be needed.
