This kind of erratic behaviour was/is visible on the GS5 port from @piggz from 4.4 … I’ve been trying to track it down for some time. I did:
zypper install sysstat
pidstat -I 1 10800 > /var/log/pidstats.log 2>&1 &
To collect exact stats. but that’s a log of log data to work with.
I am not seeing that at all. It is identical with 4.5.0.16 and consumption is now below where I can be bothered to care any more. I see no changes whatsoever and consumption is consistent and low in normal “standby”.
@Speedy-10
Yes, it looks it sleeps well, but it also looks like yours has constant CPU usage (10-15% estimated) until you plug in the charger. I find that odd.
@eson Your phone also looks like it sleeps ok’ish, and the CPU usage is also quite interesting (to say the least) until you plug in the charger.
@poetaster
Thank you for the tip. I will look into it when i have time.
It isn’t necessarily weird to have high(reported percentages) CPU usage when a CPU is (near) idle.
It could very well be reporting a low(maximum) clock speed in that state and/or using only the ‘little’ cores.
I think metrics like this are only confusing stuff(al least in this case when it doesn’t seem to matter for draining mA) . My guess is that the drainage ‘problem’ has little to with CPU usage/states, but is more datacom/modem related.
In my case I know that Wifi is using a lot of power. Setting Situations app to disable wifi when screen is disabled saves 15% a day.
Edit:
The reason why I think it’s more datacom/modem related is because that part is fishy overall on the 10III . In my case(T-Mobile NL business ) I had to use SIM slot two, one just didn’t reliably work.
Now it’s OK. I can get two days with multiple exchange accounts synchronizing , Android including Signal and Whatsapp running(MicroG enabled) and moderate usage. Could be better ,but good enough for me.
I understand what you’re saying, and you might be right, but at least in my case this kind of behaviour did not exist before 4.5.0.
“Standby” or idle power consumption of the 10 III is MANY TIMES HIGHER than that of the XA2 Ultra. As I wrote so many times, the XA2U consumes merely 9-10 mA when sleeping, whereas the lowest the 10 III is able to ever go is 32-35 mA, and usually not below 40-45 mA. Since July 2022 when I got it, I have NEVER seen in 10 III’s power consumption logs any single occurence of a figure lower than >30 mA, whereas my XA2 Ultra goes down to 9-10 mA every single freaking time it goes to sleep.
I already posted a screenshot from the XA2 Ultra, let me quote it:
As you can easily calculate, it is a difference of some 4-5 TIMES vs. the 10 III, which is simply HORRIBLE considering that the 10 III has a much more power efficient SoC, much more power efficient display, and so on.
The XA2 Ultra doing nothing but with the screen turned on (and only Battery Buddy running and showing the stats) eats some 50-60 mA, whereas the 10 III in the same conditions eats >100 mA. And so on.
This is what matters - raw power consumption figures, and not some fancy graphs. These raw mA figures clearly show how pathetically unoptimized its power consumption is if it eats that much more energy than XA2U while having that much less power hungry hardware.
Therefore, as long as the 10 III eats SEVERAL TIMES more energy than the XA2U (whereas it should be eating LESS, due to having more power efficient SoC and display) I just can’t agree that the 10 III’s power consumption is anywhere near of what can be called “normal”, “low”, or anything like that.
Last but not least, as it was also already mentioned, all reviews of the 10 III running its genuine Sony Android praise it for spectacular battery life, and place it on top of all rankings. See e.g. how it is on the 24th place of all phones ever ranked by GSMarena, with majority of the phones ranked higher than only because of having considerably bigger batteries (usually 5000-6000 mAh), which means that with its battery capacity it is actually a real champion. Well, on Sony’s Android - but not on SFOS. I would never manage to reach 137 hours on SFOS, even with literally everything disabled and the device totally idling, let alone conducting any test procedures. I’m afraid I wouldn’t even reach half of it.
Whether it is because of SFOS itself or AOSP binaries, I don’t know. I am only stating a fact.
Your long, and most likely correct, comparative rant about the XA2 Ultra is interesting, but unrelated to any negative change in 10iii behaviour across point releases, which I didn’t observe.
As you point out, the 10iii appears objectively less efficient than an older SoC based phone with SFOS and this is the unfortunate state of play which everyone would like addressed.
Given my experience with the Volla22/GS5 I think you are correct. I haven’t tried using situations yet, but I noticed that turning off wifi (leaving 4G on) overnight saved about 10% in 6-8 hours. I’ll try situations. Thanks!
Thank you for repeating this, it should not be necessary because its quite easy to understand, but somehow it is necessary to repeat it over and over again.
I just played around with the cpu frequency governors. Setting the lowest scaling frequency to 300000 and the governor to powersave so that it always uses the lowest frequency, made nearly no difference in power consumption.
So i think the power drain is not cpu related, maybe other parts of the phone, like the modem or some sensors are missing some power saving features…
In all pictures above, I can see, that CPU-usage is nearly 0% during charging. This is perhaps a too simple idea:
If one could put the phone in the same state like during charging with a fitting command (e.g. systemctl xxx ??), this could reduce battery consumption (when the phone is not charging).
In the next step, one could try to use Situations to put the phone in this status to desired times
This wouldn’t help. As you could see in the pictures sleep mode is also disabled.
Retested the idle current with 4.5.0.18, and it remains as before ~0.8%/hr
Flightmode, gps off, android stopped. I turned battery saver on this time. I also restarted the phone in flightmode as suggested by @jojomen
edit: tried disabling NFC too - no effect.
edit: Ran with Wifi,BT,Cellular on for 12hrs, 80->71% = 0.75%/hr i.e the same drain flightmode on or off.
I have Skype and VKPortalog running in Android, and I don’t (informally) perceive much excess drain.
BTW, just ran a long flightmode test of the Android 12 X10iii. WhatsApp and Skype are running.
4d18h 60%->39% = 0.2%/hr
Do I really have to use Android now. This is so sad. Thanks for your test.
No, just keep on charging 
Sadly i think there is nothing we could do about it.
Probably not relevant, but on the GS5 with 4.5.0.16 which is not really stable I see a lot cpu activity when the device is sleeping that is systemd-logind and co.
Same on my X10III. +20
Just upgraded to 4.5.0.18. No change. Still, the main ‘battery’ issue I’ve seen, which seems to be the consistent issue’ is WIFI. Turn that off, and all’s a lot better.
Maybe, wifi isn’t sleeping properly because of some hook related to input checks? Turning off wifi certainly does save battery power.
Sorry, but that you didn’t observe what I do, doesn’t make my observations in any way “unrelated” to your lack of such observations.
OK, but what does such a test have to do with real life usage conditions? I own this phone in order to use it, not as a museum exhibit. Therefore, what seriously bothers and worries me is first of all its DREADFUL power consumption not when everything is disabled (as in such a state I hardly ever have it) but with everything turned on, i.e. as in normal in-use state: mobile data on, wifi on, flightmode off. And in such conditions (and still not even actively used), in the very best case it consumes 1% every 40-50 minutes (i.e. up to 1.5% per hour) , which is simply horrible. Power consumption during voice calls is even more horrifying, quite often 1% every 8-12 minutes.
But OK, today when going to sleep I turned off literally everything but mobile network (for voice calls). Mobile data was off, WiFi off, GPS off, Android off, etc. After 8 hours it lost 8%. Just to compare, my previous daily driver the BlackBerry Passport was losing in the same conditions and in the very same place at my bedside, some 2-3% (of a 1000 mAh smaller battery), i.e. some 1/4th - 1/5th of what the 10 III eats.
Sadly this is the only thing that saves the 10iii from being unusable, it’s big battery…
There must be something totaly wron with the power saving. It is not a problem with some CPU settings. Playing around with governors and tweaking their settings doesn’t give any noticeable gains when something else is draining so much power.
Has anyone built and installed Sonys AOSP and checked the power usage in Android?
