Pick AIOs, set the heat load, choose what they should have in common — fan rpm, noise, or coolant temperature — and read the rest off the overlay. Same model as the database. Not measurements.
Same noise: each AIO's stock fans at the rpm where they reach that level, then its coolant. Same coolant: the rpm each needs for that rise above intake air at the heat load, then its noise. A fan under its minimum rpm is held there.
A radiator moves heat in proportion to the difference between coolant and incoming air: Q = q · (Tcoolant − Tair).
q depends on the air actually passing through the core. Airflow per fan is CFMmax · derate · rpm/rpmmax
with the radiator derate (0.62, or 0.72 for pressure-optimized blades) times the core-thickness factor. Per 120 mm segment at 27 mm:
q = 8.5 W/K · (airflow / 45 CFM)0.6; heat transfer over fins scales with roughly the 0.6 power of velocity.
| Core | Airflow factor | Fin area factor |
|---|---|---|
| 27 mm | 1.00 | 1.00 |
| 30 mm | 0.97 | 1.09 |
| 38 mm | 0.90 | 1.15 |
| 45 mm | 0.84 | 1.22 |
Thick cores gain fin area but lose airflow; at low rpm the net is close to zero. With fans off, natural convection is taken as 1 W/K per segment.
The band is ±15 % on q: fin density, coolant flow and preheated case air. Pump speed matters little above roughly 1 l/min.
Noise is the stock fan's curve plus 10 · log₁₀(fan count).
Air entering the radiator is case air, usually 5 to 10 K above room. Die temperatures sit above the coolant by the block and TIM delta, typically 25 to 45 K under load. Nothing here is measured; the same model runs in RadiatorTemp.