Re: [Announce] kernel-power stable v50 in Extras-Testing
 

i think we should know what the meaning of 'do it at your own risk'

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

that is just with conky, and its not constant but alot, watching cpuinfo_cur_freq correctly shows 249600 with rare spikes to 900000 when needed

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

@majaczek
kernel-config show isn't proper way of measuring current frequency, as process of checking it via kernel-config always make device jump into highest possible frequency. It's not a bug etc - it just show a state tiny fraction of second after invoking kernel-config show, thus, creating overhead that is measured ;)

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

@kernel config show
I suppose it can be fixed if frequency is measured silently before the spike, remembered, and displayed at end. It can even include sleep for 100 msec. But yeah it displays most important info and the value can be measured another way if person know how.

@Overheat
my N900 is warm only while charging (temperature on the back case) so it seems overclocking was going happily. I however wonder why it heats on low cpu usage during charging, because before (KP49 default profile) it was cool (not enogh warm to feel) even after overnight charging. Nothing dangerous (it was warm not hot) but I feel device running while charging run low efficient (it was probably some mechanism fooled because voltage meter is integrated with charging metter, so mechanism got high voltage measured and not fixed input enough). It's rather just high charge usage while charger connected - only a bit decrased device life -mostly battery and charging chip.

@Overclock
after tweaking dosbox a bit (overlay, noscale, fullscreen, byhand cpu limits, big mixer buffer, small sound time buffer) and having good overlocking profile upto 900Mhz I can run Another World (VGA, HQ sound) on dosbox (sound limited to 16khz probing while HQ in AW is 10khz) really smoothly (including intro movie) so both given me important boost, so I perhaps may try win3.11/win95 on dosbox or some more recent dos game (although don't know which one because best dos games have remade engines to native compiled for arm so it's dumb to even install it on dosbox). It seems my N900 is capable of much more than I believed once having KP (overclock, VDD) and tweaking apps settings fair. Thanks pali and guys!

PS: some games lock at max despite were running smooth on 600. Does using such apps with overclock feature reduces my device life in a visble way? Should I write some launcher script to swap profiles before running them and swap back after closing to default?

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

Have a question for all those getting more than 24 hours on a single charge:

Are any of you "online" on your instant messaging accounts/services 24/7? Just want to get a baseline on where I have been as far as power consumption on KP50.

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

That may depend on other factors. Charging batteries get warm. They can get warmer if they're being charged at faster rates (quick charge vs slow charge). Also, if they're charging more, like now you charge at 10% vs before you were at 50%. Finally, batteries age, and as they do so they become less tolerant to charging, including a propensity to get warmer. This could just be a change in how you're using the device, charging it, or an effect of age.

I've had a couple games do that, even native ones. They were tested and launched expecting a default system, and as such didn't hit a race condition or other resource contention that they would at a faster speed. Happens more than you may think. :)

Using scripts or a QBW to launch them would probably be just as good as any other solution. I did that for TowerBloxx and it worked just fine. (At 500-900 it would get choppy, at 250-600 it was smooth... go figure...)

I haven't for about 6 months now, and my activity on them was low to start with. But in general I found it was more about how you were connected than what you had running. Running on 2G or wifi was fine; running on 3G drained. I also found that having bluetooth on in some areas (void of other devices) was fine, where in an office setting it was quite draining, even when not being accessed. (Apparently the bluetooth stuff wakes the CPU for most processing needs, which would cause far more battery use...)

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

Whayong, we just have running what we really need, as mayor rule of energy saving is just not wasting it. On my side I had shortcutd for making camera key software shortcut but it installs also proxymityd which is run all time if any function of it may be needed, so since I don't need additional shortcuts on proxymity sensor I just disabled a panel of options so I have shortcuts on camera button but proxymityd is always off and it would make 10 percent of cpu usage @500mhz if it have to be running.
Same on you disable things you really doesn't need mostly those you won't use anyway (stupid option of one app may suck energy wastefully). To check what sucks most energy just type powertop in terminal and wait for results - the info would be very detailed but most important is list of processes - if only xserver is substantially different from zero it's perfect because it's the one for display and system interface so even it would be almost zero if screen would be off - else you may want identify other programs which suck energy and consider disabling/removing them or just set power saving options and disable modules you won't use. The second important are hardware modules so you would know if switching between 2G/3G would make important changes and same for wifi etc.
Also be sure to have rational settings on each app - checking email every minute won't make you any happier but sucks battery and network usage. You may also consider limit of number of communicators - just using builtin skype is mostly enough or if you don't use skype disable it logging so your favourite one would work mostly solely. Also you won't use communicators while sleeping so temporary logout overnight.
I got about 1.5 days on not fully rational settings, with skype always on while online and medium usage including some usage of university's wifi while some power kernel energy saving options not worked yet, so you surely can get 48H on rational usage and rational settings without charging now, but also don't mind to charge it if it is trivial to (like the time it laying on your home desk).
Notice some poorly written apps or applets may suck lot of energy while they don't really need them (like refreshing screen of non active application or making some heavy request every second. Check if you have any offending apps or if you don't understand it much put output of powertop on forum and ask for help.
Kernel Power is generally only better for your battery since overclock would only trigger on demand (if set up) and there are some tricks to lower batterry usage on both OC and stock frequncies and they are so good you could have better battery life while having KP and overclock than having both stock kernel and stock frequency limit.

Long things short - just check it and if anything wrong ask for help as mostly other apps are problem mostly ones in early test version or needed to run always in background. If you really think there is some problem exactly in KP don't mind showing it here. People are mostly helpful even if it is false alarm but be polite and remember they can help but won't think instead of you.

EDIT: I were ninjad :P

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

Thanks for the reply guys. It is much appreciated. I don't think I have power consumption issues. I just wanted to get a base line from you guys that are seeing 1+ days on what your actual daily usage is.

I'm seeing about 18+ hours of battery life for my usage:

- "stock DSP" from KP50 with limits 500 805.
- no 3G (I have AT&T), so I'm always on 2G and/or wifi.
- don't use bluetooth.
- messenger (gtalk, ym, FB, and Skype) "online" the whole time.

Overall, I'm happy with the battery life I'm getting. It's just the "tinkerer" in me that wants to squeze out more every chance I get.

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

@Estel

so, Estel, again, any *good* reason to upgrade from v48 to v50?

pfff

grtz
Drax

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

Drax do not troll. The most important thing is both VDD working perfectly (including when overclocked) which result in lower battery usage.
For more info check changelogs and/or just all things pali, freemangordon, estel and vi_ said (do i miss anybody important to KP?) And try to understand.
It's generally rule of the thumb that if it isn't supposed to make new issues you should just upgrade to latest release version (the one with number only accesible via repository, the test/alfa versions are mostly for testers so while they may be better they are not guaranted stable).
So if you have older KP and you see "KP50 officially anounced" it's mostly time for upgrade and you can see on own eyes what is better. If you don't enable custom options upgrade from elder KP won't generally spoil anything.

Please leave to yourself if you install it or not - you read what is written and then decide or you just believe on "stable releases". If anything wrong you may report your claims here (but think a bit before posting), and if it is indeed something wrong you either may test a fix or just get back to nokia kernel and install elder version of KP you want on top of it.
They already stated what is new and shiny so you mostly should want reading some posts.

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

Will the kernel-power be further developed or is v50 the final last one?

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

@Mr. Incredible
Do you know what is "Constant Progress"?
There would be probably many future versions, but KP50 has "stable" status so is stable like final, but Kernel Power would probably be even more improved.

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

bugfixes! there cant be any better reason than this and no other reason is needed, even if you think that version you are on works fine

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

Lately, I've been setting up 2nd device for family member. For testing/debugging/practical purposes, I've decided to set it up from scratch (well, many times using config files from my 1st device, but never copying things as "batch"), instead of using backupmenu backups.

Going straight to merit, I've run into very strange issue with kernel-power-settings. The problem is, that after every reboot, device is using default (stock Nokia) settings, instead of one that I've defined. I'm absolutely sure, that /etc/default/kernel-power is my custom settings. Of course, I've also saved my settings to /home/user/.kernel/estel, and tried - few times already - to re-save it as defaults, via kernel-config default estel.

I'm absolutely sure, that it doesn't have anything to do with unexpected reboots - there is *no* /etc/kernel-power/.notloaded file, and there *is* /etc/kernel-power/.loaded file after every reboot - yet, device is set to stock settings - unless I do kernel-config unlock (despite fact, that it wasn't locked - it just loads /etc/default/kernel-power) or kernel-config load default, or kernel-config load estel. Then, everything works as it should - until next reboot.

I've even went so far, that I've replaced /usr/share/kernel-power-settings/default with my settings (saved as default) - all to no avail.

Anyone got a clue WTF? Any ideas how can I debug it? freemangordon, Pali, I'm looking at You... But of course, any ideas welcome.

/Estel

// Edit

On my 2nd device, I'm *not* using multiboot, but straight away kernel-power and backupmenu (not that it matters). What actually *can* matter, is that first thing I did to configure kernel settings, was copying my /etc/default/kernel-power from 1st device to 2nd device. I wanted to easily, as fast as possible, make this default settings, loaded on every boot.

Maybe, just maybe, it confused kernel-power-settings, somehow, and screwed it for good.

// Edit 2

Other question, unrelated to previous issue:

Using file /etc/default/kernel-boot, we're able to set options for FSCK of /HOME? and SSH. But, FSCK probably uses options for ext3? If someone use other filesystem (ext4, ext2, ReiserFS...), the result is FCKD'ing system, instead of repairing it. Any way to customize it? Where exactly fsck command controlled by this setting resides?

/Estel

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

@Estel,

About FSCK_HOME: /etc/event.d/rcS-late sources /etc/default/mount-opts, which kernel-power patches so as to source /etc/kernel-power/pre-mount.

On pre-mount FSCK_HOME is checked and if true, the first partition on mmcblk0 with type 83 will be fsck'ed setting the type explicitly to ext3.

So you'd need to patch that script either to suite your partitioning scheme or to make it more flexible..

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

@Estel,

About the kernel-config load. I'm sure the /usr/sbin/kernel-config script has a (number of) bugs, but I can't tell for sure now.

In any case, you might want to manually kernel-config load your settings and then kernel-config default (i.e. without giving it a name) and see if that does it.

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

Thanks a lot for Your answers, reinob. As for FSCK, it's what I mean. But, as for loading kernel-config manually and kernel-config default it, I've written, that I tried it already, numerous times - amongst other things, to no avail. It just seems like it "stuck" on loading stock on boot. I don't even know from where it gets it - as I've written, I've gone so far, that I've replaced template "default" with my scheme.

/Estel

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

Hey Estel.

reinob said
kernel-config default
and you said
kernel-config default estel

There is a difference. First sets the currently used scheme as default and second, independent of current scheme, sets the named scheme (estel) as default.
I am not sure, but after updating to kp50 (yes, I finally did it ;)), I think I had similar problems:
kernel-config default schema
did not set the default one, but kept old settings (from kp47) after reboot.
But after using
kernel-config load custom
kernel-config default
everything was okay again.
Maybe any other experienced that also? Then it could be bug (just assumption) or just inconvenience (when you know it!).

--edit
Just tested it once more. And no problems at all!
But I am sure, that there were some similar irritations after switching :confused:

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

Purge kernel-settings and re-install. Perhaps the event.d script is barfed. It is not 'loading' stock settings, it is loading NO settings, stock is what is set at boot and NOT being changed.

(probably)

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

You're right, I haven't written about it - yet, I've also tried kernel-config default ;)

vi_, what You've suggested is what I would probably try, but, sudden of nothing, it started to work properly... I've rebooted it countless times, cycled (turn off, remove battery for minute, put battery again, turn on, set date and time and other sh|t ;) ), to no avail.

Now, left overnight with battery inside (just turned off), and after booting, it's as it should be. WTF, really?

BTW, little info about placebo effect...

I discovered that it's OK now, when my Son was playing Frogatto, and I helped him with boss. I got irritated on "little stutters" - obviously, happening due to lower frequency, so I've minimized it and went into terminal to load proper settings. Yet, I've checked currently loaded ones, just in case - of course, I've discovered that it's actually running on 500-900 as it should. Well, from this moment, I haven't felt any "laginess" in-game ;)

/Estel

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

So about "same" experience as I had. Strange behaviour with default settings after installation of kernel-power-settings, but self curing one.
I could not reproduce it.

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

Hi Pali and others developers,maybe this is off topic but i want to ask something,is there no way to make rootfs more than 50 or 80mb(available for applications) for example 400mb because although i have 6.58gb for applications,always take into consideration how to left from rootfs after installed every application.It is true that through this forum there are many info how to free some space from rootfs but if is possible to created 400 or 300mb(available for applications) rootfs this will be great,by the way thanks for kp-50
regards

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

AFAIK rootfs is part of other chip (cortex a8 chip?) than n900 flash memory chip so I don't think it's possible to repartition it (sounds like making RAID in mobile phone :> ) apps are installed on "home" partition which has 2gb, and it CAN be repartotioned if you want more.

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

If KP devs would be able of doing such miracles, I would ask them to "create" more RAM first. And then, bring them a sheep as a sacrifice ;) Although, thinking about it, ramzswap is something like that, yet a little flawed.

Maybe we need another sheep?

https://d1ij7zv8zivhs3.cloudfront.ne...jpg?1318331261

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

Thumb2 coming!!

Start the sacrifice :D

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

Enough votes, KP50 can be pushed to extras :D

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

i agree all working fine with me on KP50 :p even fcam drivers

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

Thanks for for explanation
estel i want to tell you something,i think although we live in XXI century the miracles continue to occur.About sheeps and sacrifice,maybe the only difference between estel and sheeps is a more legs on the sheeps,nothing else.Nothig personal just my opinion
regards

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

@disappear
don't you think this is a wrong thing to say ? for Estel ??!! man you need to show some respect at least

./sifo

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

Relax guys, keep the discussion on the technical issues.

This is the third time I say this today... :(

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

No worries, he just made me laugh a little - if it was insult, it was the strangest I've ever heard, with, apparently, insulter struggling to even express it properly ;)

Stay on target, stay on target, Luke...

/Estel

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

hi guys i have a problem.. i edited the DSP profile and changed the min and max frequencies to 500 and 900 respectively.. but the problem is when i attend a call the phone and end it,it turns very unstable.. when i type kernel-config show it says min freq is 125.. i had to repeatedly load my profile again.. can u suggest a solution??

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

same here. min freq shows 125 in kernel-config show but in the profile it shows 500 min.

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

As I've been talking with visN900 via PM, I would like to add that his problem is that - at least, as it seems to Him - that device *stays* on 125 mhz after call, until he reload profile.

I never encountered such weirdness - AFAIK, this bug was hammered light years away and centuries ago - so it was me who suggested him to write here.

/Estel

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

@vis900 hmmm... back to the old 125mhz after making/receiving a call, no need to reload the profile just ignore it you got the answer already on page 33 post 325 & 327

edit: the device will NOT use 125mhz it shows 125 but if you check with conky you will se the min stays at 250, if your phone is unstable is because probably can not handle 500 to 900 mhz profile, how is on the default profile or dsp my devise is unstable some times when I overclock over 850

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

my phone is very stable before i answer a call.. only afther call the min freq changes to 125mhz and in the avoid frequencies it says 125 and 250..??!! i tired the dsp profile too.. even in that profile i get the same problem that 250 changes to 125mhz

i ll check those posts.. tks:)

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

Do the following things
this may help you

kernel-config load default
kernel-config save default
kernel-config default default

leafpad /home/user/.kernel/default

delete frequency below 500
(that are 125 250 along with volts)
reboot

note:in place of default you can use your own profile

edit/ you can use this link for more information
http://talk.maemo.org/showpost.php?p...postcount=4917

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

i did that too.. i think this should illustrate u my problem.. though 125 and 250 are removed from the config file the problem persists..

this one is before call

https://lh5.googleusercontent.com/-T...520-111925.png

this one is after i attended the call


https://lh3.googleusercontent.com/-m...520-111935.png

even i can not understand from where does it load that 125mhz :D

for now batterypatch is ok for me.. it stays in the same profile(ie.,500-805mhz).. phone's stable and working fine..;)

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

Do you see 125mhz on active frequencies? NO!
Sorry it says min = 125 mhz but I won't actually use it.
Can you describe what "unstable" is? My bet is, it's just swap fragmentation or I/O collisions...

Re: [Announce] kernel-power stable v50 in Extras-Testing
 

when i surf the net via my wifi connection it lags like hell..on loading my profile again it performances as before.. i can hear like some disturbances when listening to songs too,(while multi-tasking) it lags there too.... :(