SailfishOS 2.0 Nexus 5 CM12.1 Alpha1 | Sfdroid Pre-Alpha/Early Preview - Page 30 - maemo.org

deprecated | # 291 | 2016-11-30, 23:13 | Report

Originally Posted by mautz View Post

I ran into the same problem, when using the new 2.0.4.13 image. No problems when using 2.0.1.11 image and updating to 2.0.5.6.

@ deprecated

Do you use okboard or only the stock keyboard?

You can access the files and folders of sfdroid normally like it used to be on an android only device.

sdcard for example can be found at /storage/sdcard0. Maybe you need root priviliges to access the folders.

I use the stock keyboard, but I think your question might've been directed at the other guy.

I helped him out with his EAP thing after quoting his post.

The Following User Says Thank You to deprecated For This Useful Post:
juiceme

deprecated | # 292 | 2016-11-30, 23:18 | Report

Originally Posted by mautz View Post

Small kernel update...

Kernel v8

Excellent! I've been really enjoying the conservative governor with the mako hotplug driver, but I will definitely give this one a go. Always new toys!

Thanks mautz.

The Following User Says Thank You to deprecated For This Useful Post:
juiceme

claustn | # 293 | 2016-12-01, 13:14 | Report

if the question was for me, yes i'm using OKBoard.

The Following User Says Thank You to claustn For This Useful Post:
juiceme

mautz | # 294 | 2016-12-01, 13:33 | Report

Originally Posted by claustn View Post

if the question was for me, yes i'm using OKBoard.

Yes i meant you

There is a bug in okboard, that causes the transparency issue. Only solutions are use another browser like Webcat or Webpirate or don't use okboard.

The Following User Says Thank You to mautz For This Useful Post:
juiceme

minimec | # 295 | 2016-12-01, 14:58 | Report

During my testing of the new features in the 'mautz' kernels, I started to write a script that generates some basic battery 'poll data'. The last couple of days, the script has evolved and grown, and I guess it's time to share it with you.

The script does basically log the current power consumption with a given interval and (now) generates a report at the end. The report includes some useful information about the current phone settings, allowing you to compare different testing situations easily.

The first version is ready now and is compatible with the 'mautz' kernel_v8, but should also run with the default kernel, besides the fact that some files, the script is checking, are not present. You will have two files, the 'script-routine' and a 'starter', that allows you to launch the routine with individual options. The source code can be found here... https://github.com/minimec/powerstats/tree/master

The report file would look like this... report file
A 'gnuplot' of the file would look like this... gnuplot file

I recommend to download and decompress the '.tar' file with the following commands on the phone, but you can also download the file here...

Code:

curl -L -o powerstats_v1.tar.gz https://www.dropbox.com/s/2wt7hd69la4j116/powerstats_v1.tar.gz?dl=0
tar -pxvzf powerstats_v1.tar.gz

You will get a 'powerstats' file and a hidden '.powerstats-routine' in your '/home/nemo' folder. A 'powerstats-report' folder will be created during first run.

You can get an overview of the options with './powerstats -h'. If you launch the starter with './powerstats', the routine will run with default settings. The 'starter' approach allows you to close the terminal right after launch...

A good first test might be...

Code:

./powerstats -i 2 -m 1 -n 1 -p 40 -t first-try

Code:

[nemo@pris ~]$ ./powerstats --help            
Script to measure current power consumption
on devices running SailfishOS.
 
EXAMPLE:
./powerstats -i 1 -m 30 -n 1 -p 50 -t gps-moving 
 
options:
-h, --help			show brief help
-s, --stop			abort script
-i, --interval=INTERVAL	specify interval (default = 2)
-m, --minutes=MINUTES		specify measure time in minutes (default=60)
-n, --notification=NOTIFICATION	set '1' for notification (default=0)
-p, --processlog=PROCESSLOG	specify limit in mV for process logging (default=60)
-t, --title=TITLE		specify a title for output file (default=idle)
[nemo@pris ~]$

I would love and I hope to see some shared reports, that would allow us to compare the Neus5-SailfishOS experience with different OS versions and settings.

I stay tuned for questions and new ideas from your side.

The Following 4 Users Say Thank You to minimec For This Useful Post:
deprecated, juiceme, mautz, mrsellout

deprecated | # 296 | 2016-12-01, 15:22 | Report

Originally Posted by minimec View Post

During my testing of the new features in the 'mautz' kernels, I started to write a script that generates some basic battery 'poll data'. The last couple of days, the script has evolved and grown, and I guess it's time to share it with you.

Fantastic work, minimec! I'll start generating reports now. I had been just running a watch over ssh after scripting a governor/hotplug driver selector. This will make things much more presentable so we can all share data.

Side note: Are you staying on 2.0.4.14 for the time being?

The Following 2 Users Say Thank You to deprecated For This Useful Post:
juiceme, minimec

mautz | # 297 | 2016-12-01, 17:19 | Report

@minimec

Thank you!! I'll try your script in the next days, at the moment i'm fighting with a new kernel :-D

EDIT:

Just downloaded your tool and tested it with your 'first run' suggestion. Very nice!!!

This gives a much more detailed output than SystemDataScope in terms of Power consumption.

Thank you, awesome!

Last edited by mautz; 2016-12-01 at 19:20.

The Following 3 Users Say Thank You to mautz For This Useful Post:
deprecated, juiceme, minimec

minimec | # 298 | 2016-12-01, 21:59 | Report

Originally Posted by deprecated View Post

Side note: Are you staying on 2.0.4.14 for the time being?

Yeah.. I did see your first positive notes on 2.0.5.6 and do not read any rants by other users like 'mautz' or 'BluesLee'. So I might upgrade at least my testing image quiet soon.

Originally Posted by mautz View Post

This gives a much more detailed output than SystemDataScope in terms of Power consumption.

That looks like a pretty nifty tool. On my side I cannot tell you whether my 'process triggering' is really accurate, and yet it seems to do the trick.

I did a new 'powerstats' analysis. I wanted to know the impact of the 'sailorgram -deamon' on battery life. I did 3x times the same walk, once with 4G/'sailorgram on', then 3G/'sailorgram on', then 4G/'sailorgram off'. In between I made a ~7-10min. pause at home. The result was stunning... I don't not run the 'sailorgram -daemon' permanently anymore, but only connect from time to time... (The high peaks is where I turn the phone on to switch from 4G to 3G and back).

See the results

The Following 2 Users Say Thank You to minimec For This Useful Post:
deprecated, juiceme

mautz | # 299 | 2016-12-02, 04:43 | Report

@ minimec

I did a 5 hour run of your tool last night:

Code:

,---
##   | Phone Info
##   '---
##   Phone model:  Nexus 5
##   Device name:  hammerhead
##   Mod version:  12.1-20160704-UNOFFICIAL-hammerhead [LMY49J]
##   OS version:   SailfishOS 2.0.5.6 (Haapajoki) (armv7hl)
##   Install Type: multirom
##
##   ,---
##   | Current Phone Settings
##   '---
##   CPU Managment:  ()
##   Min/Max Freq:   300000/1497600
##   Voltage Table:  modified
##   CPU Governor:   zzmoove
##
##   Cell Provider:  Drillisch
##   Mobile Data:    disabled/
##   Wireless:       enabled/connected
##   Bluetooth:      disabled/disconnected
##   GPS Receiver:   disabled/disconnected
##
##   ,---
##   | Running Applications
##   '---
##   /usr/bin/jolla-clock
##
##   ,---
##   | Power Statistics (Duration: 300min.) [Start: 2016-12-01-22h03m41s]
##   '---
##   below 20mV: 0% (12 polls)
##   #
##   below 30mV: 60% (5395 polls)
##   ############################################################
##   below 40mV: 24% (2157 polls)
##   ########################
##   below 50mV: 4% (337 polls)
##   ####
##   below 60mV: 3% (237 polls)
##   ###
##   below 70mV: 1% (117 polls)
##   #
##   below 80mV: 3% (258 polls)
##   ###
##   below 90mV: 1% (133 polls)
##   #
##   below 100mV: 1% (55 polls)
##   #
##   below 110mV: 1% (89 polls)
##   #
##   below 120mV: 1% (53 polls)

This is my SystemDataScope output, seems that your script is a little bit CPU hungry

Is it possible to let the sort command run less often?

Concerning your Sailorgram observations...in my opinion SFOS isn't a very battery friendly OS. Everything causes battery drain, nearly every application causes too much CPU load when the phone is put to sleep, even the message app...

If you have a good 4G reception in your area you could leave 4G always on, i didn't notice a difference between leaving the phone in 4G mode or using 3G and only turn 4G on for mobile data usage.

At the moment i have to charge the Nexus every 2 days with my normal usage of the phone. I'm thinking of putting a battery of the LG G2 into the nexus, to get some extra juice...

The Following 4 Users Say Thank You to mautz For This Useful Post:
deprecated, jellyroll, juiceme, minimec

minimec | # 300 | 2016-12-02, 11:55 | Report

Originally Posted by mautz View Post

This is my SystemDataScope output, seems that your script is a little bit CPU hungry

I honestly don't understand that SystemDataScope output at all. I am reading that the CPU was running at ~25% during the whole test. That seems strange. The 'powerstats-routine' process would show up in 'top' with about '0.3%CPU/0.1%MEM' usage on my phone, with peaks of '0.6%-0.9%CPU', when logging an active process.

Originally Posted by mautz View Post

Is it possible to let the sort command run less often?

The 'sort' command is part of the 'logging active process' implementation. You can easily get get rid of it by setting option '-p' to a high number, like '10000', or by changing the default value for the variable 'PROCESSLOG' in the 'powerstats' starter file. I will implement an option to disable process logging in the next version.

Originally Posted by mautz View Post

Concerning your Sailorgram observations...

As for 'sailorgram': I was just realizing, that my power consumption would jump to a higher level, whenever I was moving outside the range of a known wifi connection. So now I found out, that the permanent 4G/3G connection of the 'sailorgram daemon' is causing this.

Originally Posted by mautz View Post

in my opinion SFOS isn't a very battery friendly OS. Everything causes battery drain...

My developer/computer skills are not good enough to be able to confirm your observation, but the Jolla1 phone is very good, when it comes to power consumption, even when running tons of apps in the background. At least I would say so...

Originally Posted by mautz View Post

At the moment i have to charge the Nexus every 2 days with my normal usage of the phone.

Very nice! Indeed very nice... Can't wait to get a v9 of your kernel...

The Following 2 Users Say Thank You to minimec For This Useful Post:
deprecated, juiceme