Overheating?

Does the board require a heatsink?
The Polarfire chip does run hotter than I care to touch and after a while I can’t reboot anymore;
it just dies on kernel start:

[    3.976142] random: crng init done
[    3.983647] of_cfs_init
[    3.989419] of_cfs_init: OK
[    3.994362] Waiting for root device /dev/mmcblk0p3...
[    4.051456] mmc0: error -110 whilst initialising MMC card
[    4.319232] mmc0: error -110 whilst initialising MMC card
[    4.602678] mmc0: error -110 whilst initialising MMC card
[    4.932425] mmc0: error -110 whilst initialising MMC card

I just got it and have tinkered around with it for a bit, but have never had anything
attached to any of the io pins, in case you were wondering.
How hot is the Polarfire supposed to get in normal operation, as a baseline for comparison?

At this point I normally just cut the power and let it cool down.
After a short while it starts working again.

PS: I noted this tidbit, it might or might not be related to the problem:
This is logged from the HSS stage 0:

[1.689149] L2 Cache Configuration:
    L2-Scratchpad:  4 ways (512 KiB)
         L2-Cache:  8 ways (1024 KiB)
           L2-LIM:  4 ways (512 KiB)
[1.703645] DDR-Lo size is   32 MiB
[1.707898] DDR-Hi size is 1888 MiB
[1.731751] Design Info:
    Design Name: DEFAULT_A455AEAB364A52CD54CD61
    Design Version: 02.00.3
[1.742167] Attempting to select eMMC ... Failed
[2.643975] HSS_MMCInit() returned 0
[2.648576] User button NOT pressed
Press a key to enter CLI, ESC to skip
Timeout in 1 second
..
[3.658216] CLI boot interrupt timeout
[3.662729] Initializing Mi-V IHC

The interesting bit happens at timestamp 1.74 and 2.64, I think.

I don’t have this board.

What I would try would be to apply a temporary heatsink of some sort. If you have some silicone thermal tape just attach a heat sink from a Pi or just a small block of alum or copper. Even a heatsink from NVMe, it might be longer just don’t short something out. This will help to refine the issue, maybe or maybe not a thermal issue based upon the results.

Always useful to have something like this

88Pcs Heatsink Assortment Kit

1 Like

While I appreciate your suggestions, I was more interested in a general discussion
about the expectations as to how hot the board would get under normal operating conditions.

( I have already put a small fan close to the Polarfire and will continue to monitor it. )

Should the general consensus turn out to be you need a heatsink then I’m confident
that a new feature request for one will be forthcoming. :wink:

I’d be curious to hear if @Vauban and/or @jkridner would like to
weigh in with some notes, time permitting…

You should not need a heat-sink with PolarFire in general.
I have seen this emmc problem pop-up with some boards and did not get to the bottom of it yet. I have an updated HSS from Microchip that I intend to use to debug this but did not have the time to do much about it yet. I am not saying temperature has nothing to do with it but I have used one of these troublesome boards showing the problem, debugged for an hour, problem went away for the rest of the day then came back the following morning. That’s when I decided to switch to more recent Microchip software.
Please try with a heat-sink anyway and let us know if the problem persists. That will give me one more data-point.

Been running a small fan close to it and it’s keeping below where I can comfortably keep a finger on it.

Still early days, but so far it’s been idling along quite happily. I’ll keep you posted.

Sadly I don’t have any way to reliably measure it’s temperature, which is why I posted this feature request:

$sudo apt install gnome-system-monitor

You can look at CPU activity and see what is going on, if the cores are running 100% that would generate plenty of thermal energy. If everything is at an idle, well, your guess might be way better than mine on this.