Wall wart adequacy
#1
I've been using the newly-arrived PBP, usually plugged in. In some sessions, the charge indicator drops below 100%, but still says it's charging. If that's true, it suggests that the wall-wart is unable to charge the machine at the rate it uses power. (Modest use, nothing dramatic.)

Is this a software problem, a failing unit, or is it an indication that a more puissant charger might be a good idea?
#2
This is a known issue, and it's not the wall wart's fault. The charge controller in the PBP is apparently inadequate for what the PBP can do under full load. Without modifying the hardware, the best solution is to throttle the CPUs and/or dim the display. There are threads here with more details.
#3
Also note that the development team enabled over-clocking of the CPUs. This of course gives better performance. But, it appears that the external power circuit was not designed for the extra power load. So the extra power comes from the batteries when higher loading of the CPUs is used.
--
Arwen Evenstar
Princess of Rivendale
#4
(09-27-2020, 04:47 PM)Arwen Wrote: Also note that the development team enabled over-clocking of the CPUs. This of course gives better performance. But, it appears that the external power circuit was not designed for the extra power load. So the extra power comes from the batteries when higher loading of the CPUs is used.

The small cores are not overclocked. The big ones are overclocked by 0.2GHz. I doubt that matters. That's just one of the many imperfections of the Pinebook Pro.
#5
(09-27-2020, 06:38 PM)Der Geist der Maschine Wrote:
(09-27-2020, 04:47 PM)Arwen Wrote: Also note that the development team enabled over-clocking of the CPUs. This of course gives better performance. But, it appears that the external power circuit was not designed for the extra power load. So the extra power comes from the batteries when higher loading of the CPUs is used.

The small cores are not overclocked. The big ones are overclocked by 0.2GHz. I doubt that matters. That's just one of the many imperfections of the Pinebook Pro.

I don't know that I want to, so no big deal... but is there a way to un-overclock the CPUs? Is there are writeups concerning this issue? I can search here but .... it gets a little hard to find things.

I think that a lot of these issues would be good to port over to the wiki, a sub-wiki, or a different 'power users' 'extras' or similar section of the wiki... theres many things that could use extra discussion and information. Tongue Especially for us newbies.
-----
 pAULIE42o
.  .  .  .   .  . .
/s
#6
(09-27-2020, 06:43 PM)Paulie420 Wrote:
(09-27-2020, 06:38 PM)Der Geist der Maschine Wrote:
(09-27-2020, 04:47 PM)Arwen Wrote: Also note that the development team enabled over-clocking of the CPUs. This of course gives better performance. But, it appears that the external power circuit was not designed for the extra power load. So the extra power comes from the batteries when higher loading of the CPUs is used.

The small cores are not overclocked. The big ones are overclocked by 0.2GHz. I doubt that matters. That's just one of the many imperfections of the Pinebook Pro.
is there a way to un-overclock the CPUs?

The quick and dirty way is capping scaling_max_freq https://www.kernel.org/doc/Documentation...-guide.txt

The proper way would be setting non-overclocking core frequencies in the device tree. My Pinebook Pro review (see Website icon below) scratches that (search for rk3399-opp.dtsi)
#7
(09-27-2020, 06:38 PM)Der Geist der Maschine Wrote: The small cores are not overclocked. The big ones are overclocked by 0.2GHz. I doubt that matters. That's just one of the many imperfections of the Pinebook Pro.

Right, small ones are not overclocked. As to the big ones overclock - the biggest problem is that they raised voltage on big ones to 1.3V. This seems to escalate the problem with charging. 1.3V is also, afaik, over the safe limit (1.25V) recommended by Rockchip.

My pbp is running happily overclocked and undervolted (compared to original overclock) for months now - at 1.7/2.18 GHz and 1.15/1.25 V (little/big respectively). It generates less heat and the charging problems seem to be reduced. Of course ymmv because of the usual "silicon lottery".

And it makes hell of a difference.
#8
(09-27-2020, 06:38 PM)Der Geist der Maschine Wrote:
(09-27-2020, 04:47 PM)Arwen Wrote: Also note that the development team enabled over-clocking of the CPUs. This of course gives better performance. But, it appears that the external power circuit was not designed for the extra power load. So the extra power comes from the batteries when higher loading of the CPUs is used.

The small cores are not overclocked. The big ones are overclocked by 0.2GHz. I doubt that matters. That's just one of the many imperfections of the Pinebook Pro.

Surprise surprise, the cores are not overclocked on latest manjaro, anymore:

Code:
$ uname -a
Linux manjaro 5.7.19-1-MANJARO-ARM #1 SMP Wed Sep 2 20:43:09 +03 2020 aarch64 GNU/Linux

$ cat /sys/devices/system/cpu/cpu?/cpufreq/cpuinfo_max_freq | tr "\n" "\t"
1416000 1416000 1416000 1416000 1800000 1800000

Not that this changes much wrt to these battery load/unload cycles.
#9
(09-27-2020, 12:12 PM)Perl6_user Wrote: I've been using the newly-arrived PBP, usually plugged in. In some sessions, the charge indicator drops below 100%, but still says it's charging. If that's true, it suggests that the wall-wart is unable to charge the machine at the rate it uses power. (Modest use, nothing dramatic.)

Is this a software problem, a failing unit, or is it an indication that a more puissant charger might be a good idea?

I have the same issue. My battery dropped to 0% during the last 30 minutes of a YouTube video at full volume and brightness. I'm still at 0% while typing this and I also have ncspot playing music and yay is compiling a package right now as well. Compiling very slowly, but at least my music isn't skipping. The battery can just charge when I go to sleep! Big Grin
#10
Rising the supply voltage of cores above recommended limits is just insane.

With that fantastic product I stopped wondering a long time ago, but obviously I am the only one not making a benefit out of a huge failure point... Blush

(09-27-2020, 04:47 PM)Arwen Wrote: Also note that the development team enabled over-clocking of the CPUs. This of course gives better performance. But, it appears that the external power circuit was not designed for the extra power load. So the extra power comes from the batteries when higher loading of the CPUs is used.

Overclocking a development system is just insane.
Your 'technical' argumentation will not convince any competent engineer.


Forum Jump:


Users browsing this thread: 1 Guest(s)