Here Is A $5 Fix To Cool Your Raspberry Pi 3

[phoronix]

Phoronix: Here Is A $5 Fix To Cool Your Raspberry Pi 3

Over the past week of running benchmarks on the Raspberry Pi 3 we have seen how warm this new $35 quad-core ARM 64-bit developer board can get and it's significantly hotter than the Raspberry Pi 2...

http://www.phoronix.com/scan.php?page=news_item&px=Raspberry-Pi-3-Heatsink

[JeanPaul145]

From what I understand the new ODROID is about $/€/£ 5,- more expensive than the new Raspberry Pi 3, but performs much better. Why not just go for that then?

[Michael]

From what I understand the new ODROID is about $/€/£ 5,- more expensive than the new Raspberry Pi 3, but performs much better. Why not just go for that then?

Some people like the Pi for the community, they need that specific board design / compatibility, etc.

[Jan Werner]

Hey Michael,
There are already several benchmarks of Pine+ on openbenchmarking.
Please DO use heatsink for Pine, latest BSP from @longsleep and compile options that specify architecture / optimization level.
To see how wildly different results you get: http://openbenchmarking.org/result/1...GA-1603082GA36
and we welcome you to join our discussion on Pine forums: http://forum.pine64.org/showthread.php?tid=389

[dh04000]

From what I understand the new ODROID is about $/€/£ 5,- more expensive than the new Raspberry Pi 3, but performs much better. Why not just go for that then?

If poor performance was the sole-something a raspberry pi user cared about, they wouldn't be using a raspberry pie in the first place. Likely, not even an ARM board.

[dh04000]

Hey Michael, I think you got your math a bit off, "...that small passive heatsink is able to drop the SoC temperature almost in half."

The measurement of a non-heated surface was 72F (room temp). The measurement of the cpu withut heat sink was 207F. The delta temperature, ie: the thermal mass of the cpu due to internal heating, was 135F. The heat sink dropped this to 127F, which is a delta of 55F. This is an improvement of ~60%, not "dropping the temp almost in half".

[dsx724]

Hey Michael, I think you got your math a bit off, "...that small passive heatsink is able to drop the SoC temperature almost in half."

The measurement of a non-heated surface was 72°F (room temp). The measurement of the cpu withut heat sink was 207°F. The delta temperature, ie: the thermal mass of the cpu due to internal heating, was 135°F. The heat sink dropped this to 127°F, which is a delta of 55°F. This is an improvement of ~60%, not "dropping the temp almost in half".

We provided Michael with the data. We took the time it took to reach 100°C and then measure the temperature again at the same duration with a heatsink. It was 50°C in the same amount of time. The heatsink prevented the system from going to 100°C. However, as with all passive systems, the steady state temperature creeps up really slowly until it reached a steady state of 78°C.

With active cooling, which we really recommend for the Pi 3, the temperature remained at most 50°C.

Steady State (22°C Ambient)	Active Cooling	Passive Cooling
No Heatsink	64°C	100°C (Shutdown)
Heatsink	49°C	78°C

Also, the two boards we have do not measure temperature accurately after 80°C. When our other readings indicate 100°C, the SoC is still reporting 83°C. This prevents the software from kicking in and throttling to 600MHz which it should do at 85°C.

[dh04000]

We provided Michael with the data. We took the time it took to reach 100C and then measure the temperature again at the same duration with a heatsink. It was 50C in the same amount of time. The heatsink prevented the system from going to 100C. However, as with all passive systems, the steady state temperature creeps up really slowly until it reached a steady state of 78C.

With active cooling, which we really recommend for the Pi 3, the temperature remained at most 50C.

Steady State (22C Ambient)	Active Cooling	Passive Cooling
No Heatsink	64C	100C (Shutdown)
Heatsink	49C	78C

Also, the two boards we have do not measure temperature accurately after 80C. When our other readings indicate 100C, the SoC is still reporting 83C. This prevents the software from kicking in and throttling to 600MHz which it should do at 85C.

Oh, I see. The measurements in the article weren't steady state. Yes, steady state is the most informative as it is directly related to the heat sinks ability to remove heat and transfer it to the air.

[Tomin]

Am I the only one that thinks writing Celsius degrees (or Fahrenheit for that matter) without degree symbol is annoying?
Here, have some °°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°
On Finnish keyboard layout you can write ° with Ctrl+Alt Gr+Shift+0 (fourth layer), but I don't know how to do that on UK or US layout.

[dsx724]

Am I the only one that thinks writing Celsius degrees (or Fahrenheit for that matter) without degree symbol is annoying?
Here, have some °°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°°
On Finnish keyboard layout you can write ° with Ctrl+Alt Gr+Shift+0 (fourth layer), but I don't know how to do that on UK or US layout.

Fixed. I agree with you but theres no way to type that on US keyboard on the default keyboard. Maybe it's just laptops.