SBC Thermal Performance
Real thermal benchmarks for single board computer cooling. Every configuration runs the same 60 minute sustained CPU stress test with one-second temperature sampling.
7boards tested
19cooling configurations
20+hours of stress testing
Biggest improvement
-37.0°C
52Pi ICE Tower Plus on the Raspberry Pi 5
Coolest result
47°C peak
Banana Pi R4 with its best cooler
Hottest stock board
87°C peak
Raspberry Pi 5, throttles under load
Sort:
Banana Pi
R4
47°C best69°C worst
Configs tested2
Stock coolingNone
Top coolerGeneric BPI R4 Active Cooler-22.0°C
View comparison →
Radxa
Dragon Q8B
51°C best
Configs tested1
Stock coolingActive
View comparison →
Raspberry Pi
4
63°C best78°C worst
Configs tested4
Stock coolingNone
Top coolerEDATEC ED-Pi4Case-OB-15.0°C
View comparison →
Raspberry Pi
5
50°C best87°C worst
Configs tested8
Stock coolingNone
Top cooler52Pi ICE Tower Plus-37.0°C
View comparison →
Raspberry Pi
Compute Module 0 (CM0)
70°C best
Configs tested1
Stock coolingNone
View comparison →
Raspberry Pi
Compute Module 4 (CM4)
55°C best76°C worst
Configs tested2
Stock coolingNone
Top coolerEDATEC ED-CM4COOLER-21.0°C
View comparison →
Raspberry Pi
Compute Module 5
84°C best
Configs tested1
View comparison →
How We Test
Each cooling configuration is tested using a standardised methodology: 60 seconds of baseline idle temperature, 60 minutes of stress-ng CPU stress, and 60 seconds of cooldown. Temperature improvement is calculated by comparing peak load temperature against the same SBC with stock/bare board cooling.