<div dir="ltr">On Tue, Apr 3, 2018 at 8:17 PM, James Cameron <span dir="ltr"><<a href="mailto:quozl@laptop.org" target="_blank">quozl@laptop.org</a>></span> wrote:<br><div class="gmail_extra"><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">On Tue, Apr 03, 2018 at 05:28:31PM -0400, Adam Holt wrote:<br>
<span class="">> On Tue, Apr 3, 2018 at 5:01 PM, James Cameron <[1]<a href="mailto:quozl@laptop.org">quozl@laptop.org</a>> wrote:<br>
><br>
>     On Tue, Apr 03, 2018 at 10:33:12AM -0400, Adam Holt wrote:<br>
</span>>     > On Mon, Apr 2, 2018 at 12:36 PM, Adam Holt <[1][2]<a href="mailto:holt@laptop.org">holt@laptop.org</a>> wrote:<br>
>     ><br>
>     >     On Mon, Apr 2, 2018 at 10:44 AM, Adam Holt <[2][3]<a href="mailto:holt@laptop.org">holt@laptop.org</a>><br>
<div><div class="h5">>     wrote:<br>
>     ><br>
>     >         1) Recap from 2016: heatsinks on the original RPi3 CPU are not a<br>
>     >         cure-all when the RPi 3 is enclosed in a case that lacks<br>
>     ventilation on<br>
>     >         a hot day.  But CPU heatsinks Do Work when taking the plastic top<br>
>     off<br>
>     >         the of the original RPi 3.  The CPU throttling problem<br>
>     "immediately"<br>
>     >         goes away on such hot days...bringing the temperature back down<br>
>     below<br>
>     >         80C...as measured by the command:<br>
>     ><br>
>     >            vcgencmd measure_temp<br>
>     ><br>
>     >         2) It's snowing today right outside my window, so I can't easily<br>
>     >         simulate a hot summer's day -- but can others who live in hot<br>
>     >         environments report back their readings above, when running the<br>
>     new<br>
>     >         RPi3 B+ in various conditions?<br>
>     >           ☆ With heatsink on CPU -- and without?<br>
>     >           ☆ With motherboard fully enclose by a case -- and without?<br>
>     ><br>
>     >     My own results, with all 4 CPU's unloaded, in a chilly room:<br>
>     ><br>
>     >  RPi 3 with-heatsink-on-CPU / RPi 3 B+ / RPi 3 B+ with-heatsink-on-CPU<br>
>     > 44-46C / 46-48C / 45-47C case's plastic top removed (w/o wind or active<br>
>     > ventilation)<br>
>     > 46-49C / 48-51C / 48-52C case's plastic top attached (contains 100+ small<br>
>     holes<br>
>     > on 1 end, allowing very little ventilation)<br>
>     ><br>
>     >     After I ran "yes > /dev/null &" 4 times, to fully load all 4 cores of<br>
>     the<br>
>     >     CPU:<br>
>     ><br>
>     >  RPi 3 with-heatsink-on-CPU / RPi 3 B+ / RPi 3 B+ with-heatsink-on-CPU<br>
>     > 80-82C / 70-71C / 71-72C case's plastic top removed (w/o wind or active<br>
>     > ventilation)<br>
>     > 82-84C / 75-78C / 79-82C case's plastic top attached (contains 100+ small<br>
>     holes<br>
>     > on 1 end, allowing very little ventilation)<br>
>     ><br>
>     > RESULT: attaching a heatsink to the RPi 3 B+ CPU does not help.  It might<br>
>     even<br>
>     > make things a bit worse, hmm.<br>
><br>
>     Yes, your heatsink is no good.<br>
><br>
>     The B+ CPU has a heatsink or heat spreader already, that silver<br>
>     coloured bevelled structure with the black dot and Broadcom logo.<br>
><br>
>     What is the shape, size, and attachment method for your added<br>
>     heatsink?<br>
><br>
> It's the standard Canakit 7-fin aluminum heatsink shown here, attached to the<br>
> CPU with its own basic 3M self-adhesive sticker:<br>
><br>
</div></div>> [4]<a href="http://www.bestofjay.com/w/" rel="noreferrer" target="_blank">http://www.bestofjay.com/w/</a><br>
> raspberry-pi-3-overclock-heat-<wbr>test-flirc-case-vs-canakit-<wbr>heatsink/<br>
> [5]<a href="http://11986-presscdn-0-77.pagely.netdna-cdn.com/wp-content/uploads/2016/04/" rel="noreferrer" target="_blank">http://11986-presscdn-0-77.<wbr>pagely.netdna-cdn.com/wp-<wbr>content/uploads/2016/04/</a><br>
> heat-sinks-installed.jpg<br>
<br>
Given your results, probably wrong heatsink for the job.  No, I don't<br>
know of a better one.<br></blockquote><div><br></div><div>Indeed.  It'd be really great if the Raspberry Pi Foundation would publish basic recommendations here, even if just community/testing suggestions.<br></div><div><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Theory; the adhesive won't make good contact with outer bevel, and<br>
passive airflow above that outer bevel will be reduced.<br>
<br>
Theory; bubble in adhesive if the heatsink was not angled slightly as<br>
you pushed it down.<br>
<br>
Theory; the thermal control system is very different to the previous<br>
version, and this invalidates your test method.  The system uses<br>
thermal mass and core frequency scaling, and you've changed the<br>
thermal mass.  It's non-linear.<br>
<br>
Report your kernel version; the corresponding Raspbian release has<br>
changes for the B+.<br></blockquote><div><br></div><div>Kernel is 4.14.30-v7+ after 2018-03-28's Raspbian update ("apt update; apt dist-upgrade; reboot").<br></div><div><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
See<br>
<a href="https://www.raspberrypi.org/blog/raspberry-pi-3-model-bplus-sale-now-35/" rel="noreferrer" target="_blank">https://www.raspberrypi.org/<wbr>blog/raspberry-pi-3-model-<wbr>bplus-sale-now-35/</a><br>
for more technical detail on the thermals; note how the core frequency<br>
varies and the thermal mass is a resource.<br>
<br>
Suggest you use sysbench for at least 15 minutes before reading the<br>
temperature.  The "yes" test is very weak.<br></blockquote><div><br></div><div>Thanks.  Any recommended parameters for sysbench to battle-test all RPi 3 B+ subsystems simultaneously?<br></div><div><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<span class="">> FWIW these same heatsinks do lower the temperature of the original RPi 3 by<br>
> many degrees IF the CPU 's under medium load AND the top of the plastic case is<br>
> removed :)<br>
<br>
</span>But irrelevant, as the CPU is a different model, has a flat surface,<br>
and does not have a thermal control system that uses thermal mass.<br>
<br>
Please remember; things change, and this breaks your knowledge.  You<br>
must learn again.<br>
<div><div class="h5"><br>
><br>
>     > In any case: this result is completely different the original RPi 3<br>
>     (where past<br>
>     > experiments have shown that a heatsink-on-CPU greatly lowers its<br>
>     temperature,<br>
>     > When The Case Is Open!)<br>
>     >  <br>
>     ><br>
>     >     CONCLUSION: the RPi 3 B+ appears to be a real winner in CPU-intensive<br>
>     >     conditions, even in a "wintry" conditions (room is about 15C, and it<br>
>     >     continues to snow right outside the window).  The RPi 3 CPU is<br>
>     supposed to<br>
>     >     self-throttle at 80C, until it just about turns itself off at 85C. <br>
>     I'd<br>
>     >     assume the RPi 3 B+ CPU does the same?  But do not know for sure. <br>
>     Thanks<br>
>     >     to all who can add any similar data points, in warmer climate/<br>
>     conditions.<br>
>     ><br>
>     >     CLARIFS:  Both RPi's were running a near-final prerelease of<br>
>     >     Internet-in-a-Box 6.5 on Raspbian Lite to explore real-world<br>
>     conditions.  <br>
>     >     I waited 10+ min in all above 8 experiments before taking "steady<br>
>     state"<br>
>     >     temp readings.  Still, fluctuations in CPU activity (and temperature)<br>
>     >     arise, even long after I ran "killall yes" to end the most intensive<br>
>     CPU<br>
>     >     activity.<br>
>     ><br>
>     >         3) Prelim thermal analysis of RPi 3 B+:<br>
>     ><br>
</div></div>>     >         [3][6]<a href="https://cdn-images-1.medium.com/max/1600/1*0jU89Yu_" rel="noreferrer" target="_blank">https://cdn-images-1.<wbr>medium.com/max/1600/1*0jU89Yu_</a><br>
>     6miI-CODB<br>
>     >         MuHAw.png<br>
>     >         [4][7]<a href="https://medium.com/@ghalfacree/benchmarking-the-" rel="noreferrer" target="_blank">https://medium.com/@<wbr>ghalfacree/benchmarking-the-</a><br>
>     raspberry-pi<br>
>     >         -3-b-plus-44122cf3d806<br>
>     ><br>
>     >         --<br>
>     >         [5]<br>
>     >         [6]Unsung Heroes of OLPC, interviewed live @ [7][8]http://<br>
>     <a href="http://unleashkids.org" rel="noreferrer" target="_blank">unleashkids.org</a><br>
>     >         !<br>
>     ><br>
>     >         --<br>
>     >         Unsung Heroes of OLPC, interviewed live @ [8][9]http://<br>
>     <a href="http://unleashkids.org" rel="noreferrer" target="_blank">unleashkids.org</a> !<br>
>     ><br>
>     ><br>
>     > References:<br>
>     ><br>
>     > [1] mailto:[10]<a href="mailto:holt@laptop.org">holt@laptop.org</a><br>
>     > [2] mailto:[11]<a href="mailto:holt@laptop.org">holt@laptop.org</a><br>
>     > [3] [12]<a href="https://cdn-images-1.medium.com/max/1600/1*0jU89Yu_6miI-" rel="noreferrer" target="_blank">https://cdn-images-1.<wbr>medium.com/max/1600/1*0jU89Yu_<wbr>6miI-</a><br>
>     CODBMuHAw.png<br>
>     > [4] [13]<a href="https://medium.com/@ghalfacree/benchmarking-the-" rel="noreferrer" target="_blank">https://medium.com/@<wbr>ghalfacree/benchmarking-the-</a><br>
>     raspberry-pi-3-b-plus-<wbr>44122cf3d806<br>
>     > [5] [14]<a href="https://medium.com/@ghalfacree/benchmarking-the-" rel="noreferrer" target="_blank">https://medium.com/@<wbr>ghalfacree/benchmarking-the-</a><br>
>     raspberry-pi-3-b-plus-<wbr>44122cf3d806<br>
>     > [6] [15]<a href="https://medium.com/@ghalfacree/benchmarking-the-" rel="noreferrer" target="_blank">https://medium.com/@<wbr>ghalfacree/benchmarking-the-</a><br>
>     raspberry-pi-3-b-plus-<wbr>44122cf3d806<br>
>     > [7] [16]<a href="http://unleashkids.org/" rel="noreferrer" target="_blank">http://unleashkids.org/</a><br>
>     > [8] [17]<a href="http://unleashkids.org/" rel="noreferrer" target="_blank">http://unleashkids.org/</a><br>
><br>
>     > ______________________________<wbr>_________________<br>
>     > Server-devel mailing list<br>
>     > [18]<a href="mailto:Server-devel@lists.laptop.org">Server-devel@lists.laptop.<wbr>org</a><br>
>     > [19]<a href="http://lists.laptop.org/listinfo/server-devel" rel="noreferrer" target="_blank">http://lists.laptop.org/<wbr>listinfo/server-devel</a><br>
><br>
>     --<br>
>     James Cameron<br>
>     [20]<a href="http://quozl.netrek.org/" rel="noreferrer" target="_blank">http://quozl.netrek.org/</a><br>
>     ______________________________<wbr>_________________<br>
>     Server-devel mailing list<br>
>     [21]<a href="mailto:Server-devel@lists.laptop.org">Server-devel@lists.laptop.<wbr>org</a><br>
>     [22]<a href="http://lists.laptop.org/listinfo/server-devel" rel="noreferrer" target="_blank">http://lists.laptop.org/<wbr>listinfo/server-devel</a><br>
><br>
> --<br>
> Unsung Heroes of OLPC, interviewed live @ [23]<a href="http://unleashkids.org" rel="noreferrer" target="_blank">http://unleashkids.org</a> !<br>
><br>
> References:<br>
><br>
> [1] mailto:<a href="mailto:quozl@laptop.org">quozl@laptop.org</a><br>
> [2] mailto:<a href="mailto:holt@laptop.org">holt@laptop.org</a><br>
> [3] mailto:<a href="mailto:holt@laptop.org">holt@laptop.org</a><br>
> [4] <a href="http://www.bestofjay.com/w/raspberry-pi-3-overclock-heat-test-flirc-case-vs-canakit-heatsink/" rel="noreferrer" target="_blank">http://www.bestofjay.com/w/<wbr>raspberry-pi-3-overclock-heat-<wbr>test-flirc-case-vs-canakit-<wbr>heatsink/</a><br>
> [5] <a href="http://11986-presscdn-0-77.pagely.netdna-cdn.com/wp-content/uploads/2016/04/heat-sinks-installed.jpg" rel="noreferrer" target="_blank">http://11986-presscdn-0-77.<wbr>pagely.netdna-cdn.com/wp-<wbr>content/uploads/2016/04/heat-<wbr>sinks-installed.jpg</a><br>
> [6] <a href="https://cdn-images-1.medium.com/max/1600/1*0jU89Yu_6miI-CODB" rel="noreferrer" target="_blank">https://cdn-images-1.medium.<wbr>com/max/1600/1*0jU89Yu_6miI-<wbr>CODB</a><br>
> [7] <a href="https://medium.com/@ghalfacree/benchmarking-the-raspberry-pi" rel="noreferrer" target="_blank">https://medium.com/@<wbr>ghalfacree/benchmarking-the-<wbr>raspberry-pi</a><br>
> [8] <a href="http://unleashkids.org/" rel="noreferrer" target="_blank">http://unleashkids.org/</a><br>
> [9] <a href="http://unleashkids.org/" rel="noreferrer" target="_blank">http://unleashkids.org/</a><br>
> [10] mailto:<a href="mailto:holt@laptop.org">holt@laptop.org</a><br>
> [11] mailto:<a href="mailto:holt@laptop.org">holt@laptop.org</a><br>
> [12] <a href="https://cdn-images-1.medium.com/max/1600/1*0jU89Yu_6miI-CODBMuHAw.png" rel="noreferrer" target="_blank">https://cdn-images-1.medium.<wbr>com/max/1600/1*0jU89Yu_6miI-<wbr>CODBMuHAw.png</a><br>
> [13] <a href="https://medium.com/@ghalfacree/benchmarking-the-raspberry-pi-3-b-plus-44122cf3d806" rel="noreferrer" target="_blank">https://medium.com/@<wbr>ghalfacree/benchmarking-the-<wbr>raspberry-pi-3-b-plus-<wbr>44122cf3d806</a><br>
> [14] <a href="https://medium.com/@ghalfacree/benchmarking-the-raspberry-pi-3-b-plus-44122cf3d806" rel="noreferrer" target="_blank">https://medium.com/@<wbr>ghalfacree/benchmarking-the-<wbr>raspberry-pi-3-b-plus-<wbr>44122cf3d806</a><br>
> [15] <a href="https://medium.com/@ghalfacree/benchmarking-the-raspberry-pi-3-b-plus-44122cf3d806" rel="noreferrer" target="_blank">https://medium.com/@<wbr>ghalfacree/benchmarking-the-<wbr>raspberry-pi-3-b-plus-<wbr>44122cf3d806</a><br>
> [16] <a href="http://unleashkids.org/" rel="noreferrer" target="_blank">http://unleashkids.org/</a><br>
> [17] <a href="http://unleashkids.org/" rel="noreferrer" target="_blank">http://unleashkids.org/</a><br>
> [18] mailto:<a href="mailto:Server-devel@lists.laptop.org">Server-devel@lists.<wbr>laptop.org</a><br>
> [19] <a href="http://lists.laptop.org/listinfo/server-devel" rel="noreferrer" target="_blank">http://lists.laptop.org/<wbr>listinfo/server-devel</a><br>
> [20] <a href="http://quozl.netrek.org/" rel="noreferrer" target="_blank">http://quozl.netrek.org/</a><br>
> [21] mailto:<a href="mailto:Server-devel@lists.laptop.org">Server-devel@lists.<wbr>laptop.org</a><br>
> [22] <a href="http://lists.laptop.org/listinfo/server-devel" rel="noreferrer" target="_blank">http://lists.laptop.org/<wbr>listinfo/server-devel</a><br>
> [23] <a href="http://unleashkids.org/" rel="noreferrer" target="_blank">http://unleashkids.org/</a><br>
<div class="HOEnZb"><div class="h5"><br>
> ______________________________<wbr>_________________<br>
> Server-devel mailing list<br>
> <a href="mailto:Server-devel@lists.laptop.org">Server-devel@lists.laptop.org</a><br>
> <a href="http://lists.laptop.org/listinfo/server-devel" rel="noreferrer" target="_blank">http://lists.laptop.org/<wbr>listinfo/server-devel</a><br>
<br>
<br>
--<br>
James Cameron<br>
<a href="http://quozl.netrek.org/" rel="noreferrer" target="_blank">http://quozl.netrek.org/</a><br>
______________________________<wbr>_________________<br>
Server-devel mailing list<br>
<a href="mailto:Server-devel@lists.laptop.org">Server-devel@lists.laptop.org</a><br>
<a href="http://lists.laptop.org/listinfo/server-devel" rel="noreferrer" target="_blank">http://lists.laptop.org/<wbr>listinfo/server-devel<br clear="all"><br>-- <br></a><div class="gmail_signature" data-smartmail="gmail_signature"><a href="http://lists.laptop.org/listinfo/server-devel" rel="noreferrer" target="_blank"></a><div dir="ltr"><a href="http://lists.laptop.org/listinfo/server-devel" rel="noreferrer" target="_blank">Unsung Heroes of OLPC, interviewed live @ </a><a href="http://unleashkids.org" target="_blank">http://unleashkids.org</a> !</div></div>
</div></div>
</blockquote></div></div></div>