maemo.org - Talk - Lagging Solutions for N9

maemo.org - Talk (https://talk.maemo.org/index.php)

- MeeGo / Harmattan (https://talk.maemo.org/forumdisplay.php?f=45)

- - Lagging Solutions for N9 (https://talk.maemo.org/showthread.php?t=90281)

Yisroel491

2013-05-31 09:19

Lagging Solutions for N9

I've installed fasterN9 and still suffering tons of lag especially playing urban fat burner. are there any other easy methods of speeding up the phone?

Lirion

2013-05-31 10:26

Maybe you can overclock your phone - frew month ago i read about a software called opptimizer. But when you do this your phone gets faster battery drain ...

F2thaK

2013-05-31 10:28

Re: Lagging Solutions for N9

not true, overclock is awesome!

Yisroel491

2013-05-31 11:14

apparently to use that app inception needs to be installed. I'm not sure if i wanna do this and even if i did I'm not sure that i can install inception as is. i might need to undo some of the tweaks or even reflash the phone. this ceases to be an EASY solution.

low	2013-05-31 11:54

Re: Lagging Solutions for N9

maybe you want to check what is causing the lag first... ie are other processes eating up the cpu cycles, are you running low on memory... and focus on those issues before you start shotgun-style installing foo speed tweaks or overclocking the device.

Just my EUR0.02, conditions may apply.

Yisroel491

2013-05-31 12:32

sound advice, but i don't know how to do all that.

Yisroel491

2013-06-01 21:48

Re: Lagging Solutions for N9

3 Attachment(s)

i got resource monitor from the ovi store although some reviews claimed it's inaccurate. unless there's alternatives suggested this is what i got and here's the screen shots. I don't know what to make of it. also i remember reading something a while back on everything N9 about altering the usage and allocation of swap but it looked mad complicated.

low	2013-06-01 23:06

Re: Lagging Solutions for N9

Well thanks for taking the long shot.

To me it looks like your RAM is full and whatever more needs shuffling goes to the swap space (which is much slower). I just don't know if this comes from lazy un-caching or a serious memory hog.
Could you get a similar listing as for CPU% with the top RAM allocators?

Yisroel491

2013-06-01 23:36

Re: Lagging Solutions for N9

1 Attachment(s)

i guess this is what you're asking to see?

Yisroel491

2013-06-02 00:00

Re: Lagging Solutions for N9

i realized that the pic is hard to read. this is more legible.

Code:

Mem: 965924K used, 42724K free, 0K shrd, 34084

CPU: 60.9% usr 15.6% sys  0.0% nic 22.6% idle

Load average: 0.94 0.96 0.80 1/357 7897

  PID  PPID USER     STAT   VSZ %VSZ %CPU COMMA

 6709     1 user     R <   213m 21.5 39.8 {meeg

 7088     1 user     S <   104m 10.4 18.7 {rscm

  558     1 root     S <  84040  8.2  8.5 {Xorg

  568     1 pulse    S <  49756  4.9  3.1 {puls

 6956     1 user     R <  89980  8.8  1.5 {meeg

 6965  6959 user     R <   2164  0.2  1.5 top

  531     2 root     SW       0  0.0  1.5 [sgx_

  641     1 root     S    20676  2.0  0.7 {sens

  565     1 root     S <   7164  0.7  0.7 {mce}

  260     2 root     SW       0  0.0  0.7 [vibr

 6250   876 user     S     196m 19.7  0.0 {feni

 6826   876 user     S <   191m 19.2  0.0 {foru

 6690   876 user     S     188m 19.0  0.0 {grob

 2272   876 user     S     180m 18.2  0.0 {came

 1599   876 user     S     171m 17.2  0.0 {call

 1356   876 user     S     161m 16.2  0.0 {call

 1124     1 user     S     150m 15.2  0.0 {sysu

~ $

ladoga

2013-06-02 07:27

Re: Lagging Solutions for N9

Looks like the resource monitor from ovi store doesn't show buffers/cache, which makes it useless for following RAM usage.

RAM should be in use so that's normal, question is how much of it is used as cache and thus reallocated when some application needs it. Running "free" from the command line would be more informative.

Code:

~ $ free -m

             total         used         free       shared      buffers

Mem:           985          760          224            0           17

-/+ buffers:                743          241

Swap:          412           94          317

Above shows that here we have 241 MB free (free memory + buffered memory) for applications to claim. Even when there are no apps running that are started by you the most of stock applications are prestarted by start up scripts (see all those -prestart flags in top output) and thus using the RAM. I suppose it's made that way so that quick response of basic functions (phone calls, web browser, etc.) would be guaranteed. Above output is from N9 with current uptime of 29 days.

@Yisroel.
Your top output seems to be cropped from the right side so names of the processes aren't shown. Try to print out something like this:

Code:

Mem: 778016K used, 230632K free, 0K shrd, 18144K buff, 296660K cached

CPU:  0.3% usr  0.5% sys  0.0% nic 99.0% idle  0.0% io  0.0% irq  0.0% sirq

Load average: 0.00 0.01 0.00 1/353 12708

  PID  PPID USER     STAT   VSZ %VSZ %CPU COMMAND

12708 12671 user     R     3104  0.3  0.4 top -n 2

  364     1 root     S <   1776  0.1  0.2 /sbin/dsme -p /lib/dsme/libstartup.so

  945     1 user     S <   202m 20.4  0.0 /usr/bin/meego-im-uiserver -remote-theme -use-self-composition

10804   897 user     S     193m 19.5  0.0 /usr/bin/fenix -prestart

11148   897 user     S     178m 17.9  0.0 /usr/bin/call-history -prestart

 2832   897 user     S     177m 17.9  0.0 /usr/bin/camera-ui -prestart

 1871   897 user     S     172m 17.3  0.0 /usr/bin/organiser -prestart

 6235   897 user     S     169m 17.0  0.0 /usr/bin/messaging-ui -prestart

  952     1 user     S <   159m 16.1  0.0 /usr/bin/meegotouchhome -remote-theme -upstart -graphicssystem meego -prestart

  918     1 user     S <   159m 16.1  0.0 /usr/bin/sysuid -graphicssystem meego

 1826   897 user     S     156m 15.8  0.0 /usr/bin/contacts -prestart

11047   897 user     S <   155m 15.6  0.0 /usr/bin/grob -prestart

11166   897 user     S     136m 13.7  0.0 /usr/bin/call-ui -prestart

  993   897 user     S <   125m 12.6  0.0 /usr/bin/conndlgs

 5540   897 user     S     115m 11.6  0.0 /usr/bin/duicontrolpanel.launch -appletloader -outprocess

  995   897 user     S N   113m 11.4  0.0 /usr/bin/applifed.x

12557   897 user     S <   113m 11.4  0.0 booster-m

10784   897 user     S <   110m 11.1  0.0 booster-d

 1438  1326 user     S N   109m 11.0  0.0 msgindexer

 1773     1 user     S <   108m 10.9  0.0 /usr/bin/commhistoryd

 1636     1 user     S N   105m 10.6  0.0 /usr/bin/contactsd

 2229  2223 user     S <   104m 10.5  0.0 {n9-button-monit} /usr/bin/python /opt/n9-button-monitor/n9-button-monitor.py

  897     1 user     S <   103m 10.4  0.0 /usr/bin/applauncherd.bin --re-exec                                                   

 5455   897 user     S <   103m 10.4  0.0 booster-q

It would be also nice if TMO had [code] tags that didn't wrap the lines. :/

tortoisedoc

2013-06-02 08:11

Re: Lagging Solutions for N9

Quote:

Originally Posted by Yisroel491 (Post 1348802)

i guess this is what you're asking to see?

Hah.
Funny, I posted somehow related thread on memory consumption a way back (don't have the time to look it up).
The reply I got was "linux uses all the available memory by default".

I also noticed phone lagging VERY much once RAM is 100% full (and swap is necessary). This makes sense as flash is slow(seems to be especially slow on N9 but that is my own opinion).

ladoga

2013-06-02 10:35

Re: Lagging Solutions for N9

Quote:

Originally Posted by tortoisedoc (Post 1348853)

N9 reserves a portion of RAM for swap (ramzswap) so no flash storage is used by swap as default. If I understand correctly the system compresses pages and then stores them into ramdisk. I don't know what is the performance hit of (de)compressing pages, but it's probably much faster than using flash based swap.

Bit off-topic, but the performance of N9's flash storage seems to vary quite wildly from N9 to another, Maybe Nokia aquired chips from several sources.

For example I have two N9's which give very different results using hdparm.:

Rev. 1603 Serial: 5093798

Code:

~ # for i in {1 2 3}; do hdparm -tT /dev/mmcblk0 2>/dev/null; done



/dev/mmcblk0:

Timing buffer-cache reads:   280 MB in 0.51 seconds = 561679 kB/s

Timing buffered disk reads:   94 MB in 3.00 seconds = 32071 kB/s



/dev/mmcblk0:

Timing buffer-cache reads:   280 MB in 0.50 seconds = 562285 kB/s

Timing buffered disk reads:   95 MB in 3.02 seconds = 32176 kB/s



/dev/mmcblk0:

Timing buffer-cache reads:   302 MB in 0.51 seconds = 604549 kB/s

Timing buffered disk reads:   95 MB in 3.02 seconds = 32137 kB/s

Rev. 1603 Serial: 2354235

Code:

~ # for i in {1 2 3}; do hdparm -tT /dev/mmcblk0 2>/dev/null; done



/dev/mmcblk0:

Timing buffer-cache reads:   138 MB in 0.51 seconds = 276993 kB/s

Timing buffered disk reads:   80 MB in 3.00 seconds = 27297 kB/s



/dev/mmcblk0:

Timing buffer-cache reads:   144 MB in 0.51 seconds = 288364 kB/s

Timing buffered disk reads:   79 MB in 3.02 seconds = 26757 kB/s



/dev/mmcblk0:

Timing buffer-cache reads:   152 MB in 0.51 seconds = 305113 kB/s

Timing buffered disk reads:   81 MB in 3.01 seconds = 27466 kB/s

Even with those differences both devices are equally smooth in normal use. Maybe loading of uncached apps takes slightly longer on serial 2354235, but I wouldn't notice really. Both devices are pretty much lag free unless they are downloading feeds and I'm navigating between views or scrolling the apps/events view at the same time.

Yisroel491

2013-06-02 11:40

Re: Lagging Solutions for N9

okay it was difficult to maneuver but i managed to copy top's output from landscape mode.

Code:

Mem: 980084K used, 28564K free, 0K shrd, 37772K buff, 443284K cached

CPU: 27.5% usr  5.1% sys  0.0% nic 66.9% idle  0.0% io  0.3% irq  0.0% sirq

Load average: 0.40 0.35 0.19 1/362 10180

  PID  PPID USER     STAT   VSZ %VSZ %CPU COMMAND

10078     1 user     S <  86728  8.5 24.0 /usr/bin/meego-terminal

  558     1 root     S <  82504  8.1  3.6 /usr/bin/Xorg -logfile /tmp/Xorg.0.log -cor

 1124     1 user     S     149m 15.0  0.9 /usr/bin/sysuid -graphicssystem meego

  641     1 root     S    21700  2.1  0.9 /usr/sbin/sensord

10180 10081 user     R <   2164  0.2  0.5 top

 2553     1 user     S <  60540  5.9  0.3 /opt/theshaketorch/bin/theshaketorchd

  565     1 root     S <   7288  0.7  0.3 /sbin/mce --force-syslog

  355     1 messageb S <   6144  0.6  0.3 /usr/bin/dbus-daemon --system --nofork

  531     2 root     SW       0  0.0  0.3 [sgx_misr]

   56     2 root     SW       0  0.0  0.3 [irq/221-atmel_m]

  630     1 csd      S    11604  1.1  0.1 /usr/sbin/csd -z -p call autorej=0,autoans=

  624   513 root     S     4184  0.4  0.1 {hald-addon-inpu} hald-addon-input: Listeni

  870     2 root     SW       0  0.0  0.1 [ipolldevd]

  287     2 root     SW       0  0.0  0.1 [sgx_perf]

 6709     1 user     S <   219m 22.1  0.0 /usr/bin/meego-im-uiserver -remote-theme -u

 6250   876 user     S     196m 19.7  0.0 /usr/bin/fenix -prestart

 9687   876 user     S <   192m 19.3  0.0 /opt/forum-qml/bin/forum-qml --tmo

 2272   876 user     S     180m 18.2  0.0 /usr/bin/camera-ui -prestart

 1599   876 user     S     176m 17.8  0.0 /usr/bin/call-history -prestart

 1824   876 user     S     173m 17.4  0.0 /usr/bin/messaging-ui -prestart

 9681   876 user     S     172m 17.3  0.0 /usr/bin/grob -prestart

edit
and here's the output for free.

Code:

~ $ free -m

             total         used         free       shared      buffers

Mem:           985          956           28            0           34

-/+ buffers:                921           63

Swap:          255           10          245

p.s. i'm not sure if y'all appreciate the severity and extent of my computer illiteracy. i don't understand more than 5% of what anyone is saying. on the other hand you can see that i am capable of following instructions. still i understand if i have exceeded the supply of patience and tolerance for newbie ignorance available within the relatively advanced community of TMO.

tortoisedoc

2013-06-02 11:48

Re: Lagging Solutions for N9

Quote:

Originally Posted by ladoga (Post 1348903)

Bit off-topic, but the performance of N9's flash storage seems to vary quite wildly from N9 to another, Maybe Nokia aquired chips from several sources.

Might also depend on the battery level of each device - and / or battery qualiy?

Yisroel491

2013-06-02 12:34

Quote:

Originally Posted by ladoga (Post 1348846)

It would be also nice if TMO had [code] tags that didn't wrap the lines. :/

apparently the TMO app doesn't wrap the lines. :)
it just cuts them off. lol. but in landscape you can get pretty much the whole line. probably not the best solution especially if you want to be on your desktop.

Jeffrey04

2013-06-03 13:43

Re: Lagging Solutions for N9

Quote:

Originally Posted by Yisroel491 (Post 1348257)

I've installed fasterN9 and still suffering tons of lag especially playing urban fat burner. are there any other easy methods of speeding up the phone?

ever tried playing without sound effect and bgm? not really a solution, but gameplay should be smoother

AndyNokia232

2013-06-03 14:38

Re: Lagging Solutions for N9

You could try using DropCache from the ovi store. It seems to help me and speed my N9 up if it's been busy for a while.

Worth checking out:
http://everythingn9.com/drop-cache-f...an-app-review/

Yisroel491

2013-06-03 15:42

Re: Lagging Solutions for N9

@Jeffery04
hey that actually worked pretty well. i can actually play urban fat burners now. but like you said it isn't a real remedy for whatever is lagging out the memory. but at least now I'll be able to achieve my anti junk food trophy :)

@AndyNokia232
picked up and used the app and here are the results:

Code:

Mem: 673224K used, 335424K free, 0K shrd, 12860K buff, 181308K cached

CPU: 76.9% usr 23.0% sys  0.0% nic  0.0% idle  0.0% io  0.0% irq  0.0% sirq

Load average: 0.11 0.32 0.42 3/359 15387

  PID  PPID USER     STAT   VSZ %VSZ %CPU COMMAND

15196     1 user     R <  84336  8.3 38.3 /usr/bin/meego-terminal

14360     1 user     R <   221m 22.3 23.0 /usr/bin/meego-im-uiserver -remote-theme -u

  558     1 root     S <  86540  8.5  7.6 /usr/bin/Xorg -logfile /tmp/Xorg.0.log -cor

  568     1 pulse    S <  49756  4.9  7.6 /usr/bin/pulseaudio --system --high-priorit

10379   876 user     S     203m 20.5  0.0 /usr/bin/fenix -prestart

~ $ free -m

             total         used         free       shared      buffers

Mem:           985          657          327            0           12

-/+ buffers:                644          340

Swap:          255           59          196

definitely freed up some memory. edit: it would be nice to be able to increase my total amount of swap available.

tortoisedoc

2013-06-03 20:03

Re: Lagging Solutions for N9

Quote:

Originally Posted by Yisroel491 (Post 1349250)

definitely freed up some memory. edit: it would be nice to be able to increase my total amount of swap available.

There was an article on everythingn9 at some point last month I think (about swap)

ladoga

2013-06-04 09:27

Re: Lagging Solutions for N9

Quote:

Originally Posted by tortoisedoc (Post 1348923)

Might also depend on the battery level of each device - and / or battery qualiy?

Battery in slower testing one is in better shape and was at full charge. Faster results are from my main phone, so battery might have degraded a bit (not much though as it still charges nearly to the max.)

@Yisroel

Good. Nothing out of ordinary in CPU usage, but something seems to be hogging your RAM. Try to hit "m" key while running top and it will sort processes by highest memory usage. Maybe some app has a memory leak or has crashed and is still reserving RAM for itself.

You are also running some daemon called theshaketorchd in the background, which is using some CPU and RAM. Though if you find the app useful and haven't experienced too much of battery drain t's probably not a problem.

PS. While looking at this I also ran top on my own phone ane noticed that cutetube had crashed (I didn't have it open and I have disabled the cutetubed daemon) and hogging over 200MB of RAM. :)

To kill processes you have two ways. Using PID (process ID), which you can see from top or ps output. Or by using the name of the process.

Former is done with command "kill [PID]" and latter by command killall [name]. Both take flag -9 to force kill if the process doesn't respond. So for example:

Code:

$ killall -9 cutetube

don_falcone

2013-06-04 09:58

Re: Lagging Solutions for N9

...or use 'kill -9 $(pidof process_name)' -- SIGHUP may be a better first try than forcekill.

Yisroel491

2013-06-04 10:32

here's what i got. is there anything that shouldn't be there? i don't wanna start killing processes without knowing what they are.

Code:

Mem: 712988K used, 295660K free, 0K shrd, 9740K buff, 197776K cached

CPU:  5.7% usr  0.9% sys  0.0% nic 92.4% idle  0.1% io  0.5% irq  0.0% sirq

Load average: 0.30 0.19 0.27 1/355 18672

  PID  PPID USER     STAT   VSZ %VSZ %CPU COMMAND

14360     1 user     S <   228m 22.9  0.0 /usr/bin/meego-im-uiserver -remote-theme -u

10379   876 user     S     203m 20.5  0.0 /usr/bin/fenix -prestart

18471   876 user     S <   191m 19.3  0.0 /opt/forum-qml/bin/forum-qml --tmo

13206   876 user     S     189m 19.1  0.0 /usr/bin/call-history -prestart

16938   876 user     S     180m 18.2  0.0 /usr/bin/camera-ui -prestart

 1824   876 user     S     180m 18.2  0.0 /usr/bin/messaging-ui -prestart

16505   876 user     S     165m 16.6  0.0 /usr/bin/call-ui -prestart

  957     1 user     S <   163m 16.5  0.0 /usr/bin/meegotouchhome -remote-theme -upst

 2080   876 user     S     161m 16.2  0.0 /usr/bin/contacts -prestart

18058   876 user     S     155m 15.7  0.0 /usr/bin/grob -prestart

 1124     1 user     S     152m 15.3  0.6 /usr/bin/sysuid -graphicssystem meego

 2207   876 user     S     146m 14.7  0.0 /usr/bin/organiser -prestart

  931   876 user     S <   123m 12.4  0.0 /usr/bin/conndlgs

18561   876 user     S <   115m 11.6  0.0 booster-m

  933   876 user     S N   113m 11.3  0.0 /usr/bin/applifed.x

 1347     1 user     S N   112m 11.3  0.0 /usr/bin/msyncd

18626   876 user     S <   111m 11.2  0.0 booster-d

 1528  1385 user     S N   109m 11.0  0.0 msgindexer

 1629     1 user     S N   107m 10.8  0.0 /usr/bin/contactsd

 1762     1 user     S <   107m 10.8  0.0 /usr/bin/commhistoryd

  876     1 user     S <   104m 10.5  0.0 /usr/bin/applauncherd.bin --boot-mode

~ $

as for the shake torch it is very useful. it lets you turn on/off your torch flashlight just by shaking the phone even from the lock screen.

ladoga

2013-06-04 13:01

Re: Lagging Solutions for N9

Everything there looks normal. Now you have also more free memory (ca. 305MB) compared to your previous posting.

There also exists app called n9-button monitor, which lets you to use long press of volume keys to light the flash LED when the phone is idle. I haven't used that shaketorch app yet and I don't really know if polling accelometer is more resource intensive (just have a gut feeling it might be), but maybe someone with more knowledge can comment on that.

Other than that your N9 is probably as fast as it can be without overclocking. Even if this device is quite tweakable the hardware has its limits.

Also if swapping becomes a problem, then reducing systems tendency to swap *might help a little. I expect that you won't even notice the difference, but if you wish to try you can write some smaller value to /proc/sys/vm/swappiness (default if I recall correct is 25). Smaller value means that kernel will be more conservative in moving pages to swap. Values can range from 0 to 100.

Example:

Code:

~ # cat /proc/sys/vm/swappiness

10

~ # echo 1 > /proc/sys/vm/swappiness

~ # cat /proc/sys/vm/swappiness

1

~ #

Setting swappiness like above is not permanent and is in effect only until next reboot (then sysctl sets it to default value, which is defined in /etc/default/sysctl). As to what value works best for N9, no idea. I use 10 right now.

MSameer

2013-06-04 19:10

Re: Lagging Solutions for N9

Quote:

Originally Posted by ladoga (Post 1348903)

Bit off-topic, but the performance of N9's flash storage seems to vary quite wildly from N9 to another, Maybe Nokia aquired chips from several sources.

For example I have two N9's which give very different results using hdparm.:

You can check the output of dmesg and you will realize that you are correct. There are 2 chipsets being used in N9 and 3 camera modules too IIRC ;-)

All times are GMT. The time now is 06:37.

vBulletin® Version 3.8.8