In the single-board computer field, the Raspberry Pi 3 has been the benchmark for high performance devices: A newcomer SBC has to provably outperform the Pi 3 to build a compelling case on superior performance. One big asterisk in these comparisons is the thermal conditions during comparison. When run hard, a default stock Raspberry Pi 3 can drive itself up to the self-imposed thermal limit of 85 degrees Celsius, at which point it throttles performance.
The BeagleBoard people has released their smallest form factor yet – the PocketBeagle. Performance of their standard sized BeagleBoard are not thermally constrained like the Pi 3. But their claims of BeagleBone level of performance in a small form factor leads to the question whether packing into a tiny space also increased the thermal load. And if so, perhaps a heat sink will help performance as it would on a Raspberry Pi 3.
For this test, we ran a PocketBeagle at 100% CPU using the ‘stress’ tool available as part of its Debian distribution and measured its temperature with a non-contact infrared thermometer over a period of six minutes. The ambient air temperature was 28° C. Two runs were performed, and their average value plotted out against time (in number of seconds).
The temperature rise visibly started to taper off at six minutes. Gauging by the trend, it may approach but is unlikely to surpass 60° C. The Octavo OSD335x datasheet page 28 listed the operating temperature range of 0 to 85°C. We have plenty of thermal headroom on a PocketBeagle without use of any heat sinks when operating at normal room temperature.
Perhaps a PocketBeagle would need a heat sink in some high-temperature environments, but for normal use, anybody trying to sell a heat sink to improve performance is probably just selling snake oil. The thermal profile here is not anything like a Raspberry Pi 3.