Jump to content


21/Dec 1.91 - Teknologist kernel with tun.ko, ext4, cifs, and compache ramzswap for GSM Hero

* * * * * 1 votes

578 replies to this topic

#521
kendon

kendon

    Addict

  • Members
  • PipPipPipPipPip
  • 923 posts
  • Devices:HTC Hero, ZTE Blade

View Postteknologist, on Jan 5 2010, 21:43, said:

Maybe we'll be releasing a geek pack with the setcpu app and a no perflocks kernel
good idea, combine it with a poll so people can vote whether it works or not, pretty sure it will run dead pretty fast when people voting not working...


#522
teknologist

teknologist

    Diehard

  • Members
  • PipPipPipPip
  • 460 posts
  • Location:Paris, France
  • Devices:HTC Hero, NexusOne
  • Twitter:@teknologism

View Postkendon, on Jan 5 2010, 22:02, said:

i have switched to your solution with the performance governor too. i was running the ondemand governor with a modified threshold, but the phone got really sluggish after a day or so... i'll see what this can do, tbh i am off of the overclocking idea, don't see the advantages weigh out the disadvantages / effort...


Exactly !


#523
teknologist

teknologist

    Diehard

  • Members
  • PipPipPipPip
  • 460 posts
  • Location:Paris, France
  • Devices:HTC Hero, NexusOne
  • Twitter:@teknologism

View Postkendon, on Jan 5 2010, 22:03, said:

good idea, combine it with a poll so people can vote whether it works or not, pretty sure it will run dead pretty fast when people voting not working...


Exactly again...plus only 3 people have been asking for it over a +10000 TCK installed... ;-)


#524
teknologist

teknologist

    Diehard

  • Members
  • PipPipPipPip
  • 460 posts
  • Location:Paris, France
  • Devices:HTC Hero, NexusOne
  • Twitter:@teknologism
Been using MCR 3.1Beta for a day and I can tell it's very very fast and responsive and no sluggishness/slowdowns at all with compcache enabled !
Is it me , or others are experiencing the same ? Are people who reported sluggishness with compcache, using MCR 3.1beta, how does it feel with compcache ?

Thanks Paul, this is definitely a huge improvement for the MCR ROM on Hero !

Edited by teknologist, 06 January 2010 - 07:58 AM.


#525
freeasabeer

freeasabeer

    Regular

  • Members
  • PipPip
  • 88 posts
  • Location:Paris, France
  • Devices:HTC Hero
I've been playing with 3.1 beta/rc since beta1 and I have to say the slowdowns look to have vanished. However the beta release cycle is so fast that I maybe never had enough time to reach the point where slowdowns start to appear.... So need to wait a bit to have a real feedback on this point!


#526
jpalo

jpalo

    Newbie

  • Members
  • Pip
  • 8 posts
  • Devices:HTC Hero
1.91 Teknologist kernel on Generic 2.73.405.66 ROM makes the phone perfect (also after few days), thank you! :)


#527
wvn

wvn

    Newbie

  • Members
  • Pip
  • 20 posts
  • Devices:HTC Desire / Blackberry 9700

View Postjpalo, on Jan 9 2010, 14:08, said:

1.91 Teknologist kernel on Generic 2.73.405.66 ROM makes the phone perfect (also after few days), thank you! B)


I second that. The phone now flies. Almost got a speeding ticket.  :)


#528
DanH48

DanH48

    Newbie

  • Members
  • Pip
  • 45 posts
  • Gender:Male
  • Location:Sussex
  • Devices:HTC Hero
Just noticed in MCR 3.2 that it contains a 1.92 Tek kernel. Care to share what has changed please.


#529
teknologist

teknologist

    Diehard

  • Members
  • PipPipPipPip
  • 460 posts
  • Location:Paris, France
  • Devices:HTC Hero, NexusOne
  • Twitter:@teknologism

View PostDanH48, on Jan 12 2010, 23:02, said:

Just noticed in MCR 3.2 that it contains a 1.92 Tek kernel. Care to share what has changed please.


Hi

1.92 is just a minor update to the ramdisk to prepare it for the MCR Settings app we are developing.
This app will let you toggle compache on the fly and many more...

Stay tuned !


#530
teknologist

teknologist

    Diehard

  • Members
  • PipPipPipPip
  • 460 posts
  • Location:Paris, France
  • Devices:HTC Hero, NexusOne
  • Twitter:@teknologism
To all those still experiencing sluggishness after a while I have completely solved these on my Hero and thought I'd share...

I removed 2 things:

The HTC clock widget from Home screen (replaced it by android standard which has no seconds display)

Last but not least, removed the Taskiller widget. Seems it goes wild after a while, using ~25% CPU for extended amount of time and periodically !

All sluggishness has disappeared and this is after 48H uptime !

Running with a 128MB compcache on MCR 3.2beta - TCK 1.92


#531
maxld

maxld

    Newbie

  • Members
  • Pip
  • 31 posts
  • Location:Łódź, Poland
  • Devices:HTC Hero
Yeah, removing the taskiller seems to help me too


#532
masterpfa

masterpfa

    Diehard

  • MoDaCo Ad Free
  • PipPipPipPip
  • 473 posts
  • Gender:Male
  • Location:London UK
  • Interests:Music, art, history, photography and traveling
    Not done much of any of these lately!
  • Devices:Galaxy Nexus/HTC Desire HD
  • Twitter:@masterpfa

View Postteknologist, on Jan 13 2010, 11:56, said:

Hi

1.92 is just a minor update to the ramdisk to prepare it for the MCR Settings app we are developing.
This app will let you toggle compache on the fly and many more...

Stay tuned !

Hmmmm sounds interesting
OK guess we'll have to wait the next kitchen update?
:)


#533
Trekster2000

Trekster2000

    Newbie

  • MoDaCo Ad Free
  • Pip
  • 22 posts
  • Devices:White HTC Hero
Joining the choir...removing Task Killer widget did wonders!


#534
RatSon

RatSon

    Regular

  • MoDaCo Ad Free
  • PipPip
  • 135 posts

View Postteknologist, on Jan 15 2010, 13:27, said:

To all those still experiencing sluggishness after a while I have completely solved these on my Hero and thought I'd share...

I removed 2 things:

The HTC clock widget from Home screen (replaced it by android standard which has no seconds display)

Last but not least, removed the Taskiller widget. Seems it goes wild after a while, using ~25% CPU for extended amount of time and periodically !

All sluggishness has disappeared and this is after 48H uptime !

Running with a 128MB compcache on MCR 3.2beta - TCK 1.92

Hi all,

my experience is that it's not the taskiller widget which kills your CPU and causes laggyness, but the TasKiller process running in background. you can have taskiller widget on you screen, but always remember to kill taskiller process after you use it. (you can easily check cpu use with 'top | grep TasKiller' command). anyways i have already contacted Taskiller's developer about the topic and he answered 'taskiller must run in background to use  autokill. This feature consume cpu.' (sic.) my autokill is turned off and taskiller still consumes the same amount of cpu.
anyways i still keep taskiller to see how mutch free mem i have and to kill those non-wanted processes. if you have a good recommendation to a better task killer and/or system monitorin app (with widgets) i would make the switch.

sry. for my bad english.

Edit: Process manager seems to be a good alternative. eats no cpu, has a nice widget.

Edited by RatSon, 18 January 2010 - 11:33 AM.


#535
crashd

crashd

    Newbie

  • Members
  • Pip
  • 33 posts
teknologist: first up, great work dude :)

I have a question though, regarding the 'usb/sync' issue that some people (including myself) have been having. When you've changed some of the elements of the kernel, I don't suppose you've come across anything that changes nvram or other flashram settings to do with power saving, syncing or usb behavior?

links regarding issue :

http://android.modac...4/no-adb-no-sd/
http://android.modac...nothing-at-all/


#536
teknologist

teknologist

    Diehard

  • Members
  • PipPipPipPip
  • 460 posts
  • Location:Paris, France
  • Devices:HTC Hero, NexusOne
  • Twitter:@teknologism

View Postcrashd, on Jan 18 2010, 18:28, said:

teknologist: first up, great work dude :)

I have a question though, regarding the 'usb/sync' issue that some people (including myself) have been having. When you've changed some of the elements of the kernel, I don't suppose you've come across anything that changes nvram or other flashram settings to do with power saving, syncing or usb behavior?

links regarding issue :

http://android.modac...4/no-adb-no-sd/
http://android.modac...nothing-at-all/

Not at all... In fact I haven't changed the source itself since the first releases and my changes don't have anything to do with USB. Just tweaked the setup.

SD mounting works and has always worked like a charm here. I don't use HTCSync.
Maybe it's got something to do with new ROMs.

Does it work for you on earlier MCR releases ?


BTW in the posts you linked people are also having the issue in RA recovery which uses its own kernel ! So you see there is no relation to TCK....maybe it's the hardware usb ports failing....

Edited by teknologist, 19 January 2010 - 09:57 AM.


#537
teknologist

teknologist

    Diehard

  • Members
  • PipPipPipPip
  • 460 posts
  • Location:Paris, France
  • Devices:HTC Hero, NexusOne
  • Twitter:@teknologism

View PostRatSon, on Jan 18 2010, 10:13, said:

Hi all,

my experience is that it's not the taskiller widget which kills your CPU and causes laggyness, but the TasKiller process running in background. you can have taskiller widget on you screen, but always remember to kill taskiller process after you use it. (you can easily check cpu use with 'top | grep TasKiller' command). anyways i have already contacted Taskiller's developer about the topic and he answered 'taskiller must run in background to use  autokill. This feature consume cpu.' (sic.) my autokill is turned off and taskiller still consumes the same amount of cpu.
anyways i still keep taskiller to see how mutch free mem i have and to kill those non-wanted processes. if you have a good recommendation to a better task killer and/or system monitorin app (with widgets) i would make the switch.

sry. for my bad english.

Edit: Process manager seems to be a good alternative. eats no cpu, has a nice widget.

Yep it's exactly what I meant. In fact the widget makes Taskiller stay in the background, hence the sluggishness....but of course it also applies to using taskiller manually and not killing it after use...

[EDIT] Tried ProcessManager and I am very happy so far ! widget is great too and as it doesn't auto refresh it doesn't kill CPU/battery ! :-)

Also my battery now lasts as long as 72Hours !!!! amazing !

Edited by teknologist, 19 January 2010 - 10:54 AM.


#538
crashd

crashd

    Newbie

  • Members
  • Pip
  • 33 posts

View Postteknologist, on Jan 19 2010, 09:54, said:

Not at all... In fact I haven't changed the source itself since the first releases and my changes don't have anything to do with USB. Just tweaked the setup.

SD mounting works and has always worked like a charm here. I don't use HTCSync.
Maybe it's got something to do with new ROMs.

Does it work for you on earlier MCR releases ?
BTW in the posts you linked people are also having the issue in RA recovery which uses its own kernel ! So you see there is no relation to TCK....maybe it's the hardware usb ports failing....

I tried reverting to stock MCR, I thought it might be something that is being set at a low level but that you may have seen in the source is all. Worth a shot :) It could be the USB but if you go into FASTBOOT it still comes up as 'usb' if you connect it to a computer. Bit weird.


#539
teknologist

teknologist

    Diehard

  • Members
  • PipPipPipPip
  • 460 posts
  • Location:Paris, France
  • Devices:HTC Hero, NexusOne
  • Twitter:@teknologism

View Postcrashd, on Jan 19 2010, 11:29, said:

I tried reverting to stock MCR, I thought it might be something that is being set at a low level but that you may have seen in the source is all. Worth a shot :) It could be the USB but if you go into FASTBOOT it still comes up as 'usb' if you connect it to a computer. Bit weird.


So does it work with stock kernel or not ?


#540
crashd

crashd

    Newbie

  • Members
  • Pip
  • 33 posts

View Postteknologist, on Jan 19 2010, 12:22, said:

So does it work with stock kernel or not ?

No, as per a lot of other users, even going back to stock (or downgrading) doesn't fix the problem. pulse dug something out about stuff being set somewhere in nvram but we don't know where atm.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users