tusharmax is offline tusharmax
2010-12-10 , 03:49
Posts: 126 | Thanked: 56 times | Joined on Apr 2010
#301
just removed N900 boost script, and noticed after a reboot that although the device feels more responsive , the game NFSU feels more jerky.. havent tested sound quality tho !
 

The Following User Says Thank You to tusharmax For This Useful Post:
etuoyo is offline etuoyo
2010-12-10 , 08:43
Posts: 1,179 | Thanked: 770 times | Joined on Nov 2009
#302
Originally Posted by SavageD View Post
Was anything else working with sound after the call like the media player?
Yes media player sound was working,
 

The Following User Says Thank You to etuoyo For This Useful Post:
retsaw is offline retsaw
2010-12-11 , 00:00
Posts: 701 | Thanked: 585 times | Joined on Sep 2010 @ London, England
#303
Originally Posted by SavageD View Post
Actually most of the patch is simply personal hack I made after looking into pulse audio and observing the file systems and the various functions in the device.
I know, I've been following this thread with interest since the beginning and have read all of it.

About page cache....lowering the value makes the system slower, this I know for certain. The n900 needs swap for most things. Swap uses page-cache to perform more efficiently.
Page cache is intended to improve efficiency on a traditional hard drive. It does not necessarily follow that it will be beneficial on a SSD with significantly different performance characteristics (though it may do anyway).

The default value is 5, I turned it up to 6 just to make things relating to swap a little more efficient.
That is Nokia's default value, not the linux kernel devs, and we do not know how much consideration and testing went into deciding that value (if any).

Things were notably slower with page-cache set to 0 in all my tests.
I can believe that, I thought that reducing the page cache to a single page at a time was too aggressive. Though here I would like to know what tests you used to determine this. The swappolube tweaks were focused on improving responsiveness above pure performance which is quite subjective and difficult to benchmark. From my testing of it I felt setting it to 0 was sub-optimal but I couldn't really be sure.

The best way to go is to actually leave it at "5".
Which raises the question, why did you change it to 6?

Sorry for dragging you off the main focus of you patches, but you did decide to include this tweak in with them and I was curious about why you made that particular decision.
 
SavageD is offline SavageD
2010-12-11 , 01:55
Posts: 436 | Thanked: 406 times | Joined on Jan 2010
#304
Originally Posted by retsaw View Post
I know, I've been following this thread with interest since the beginning and have read all of it.

Page cache is intended to improve efficiency on a traditional hard drive. It does not necessarily follow that it will be beneficial on a SSD with significantly different performance characteristics (though it may do anyway).

That is Nokia's default value, not the linux kernel devs, and we do not know how much consideration and testing went into deciding that value (if any).

I can believe that, I thought that reducing the page cache to a single page at a time was too aggressive. Though here I would like to know what tests you used to determine this. The swappolube tweaks were focused on improving responsiveness above pure performance which is quite subjective and difficult to benchmark. From my testing of it I felt setting it to 0 was sub-optimal but I couldn't really be sure.

Which raises the question, why did you change it to 6?

Sorry for dragging you off the main focus of you patches, but you did decide to include this tweak in with them and I was curious about why you made that particular decision.

You don't have to be sorry about anything, It's only normal to be curious about things.


Taken from swappolube wiki:
page-cluster controls the number of pages which are written to swap in a single attempt. The swap I/O size.It is a logarithmic value - setting it to zero means "1 page", setting it to 1 means "2 pages"
Setting page cluster to a higher value means that the more pages has to be written out to swap for every attempt at swap thereby making it more aggressive. My next patch ignores page-cluster altogether and leaves at the default value "5"....

I changed it to 6 after observing that setting page-cluster to 0 made the device slightly slower.

The tests done to determine the best page-cluster values were simple. Return everything else to their default values, and play with the page cluster digits, then note the devices behavior/performance.

To see the differences I used things that would normally perform slowly on the device such as large webpages, emulators (vgba mainly) etc...things that would make it easier to see performance increases...

I decided to incorporate the tweaks after some experimentations in trying to find ways to make the n900 faster. But honestly I don't remember exactly how I came across the tweaks :/, I must've installed swappolube at one point....but I can't remember.

I do however remember the device performing alot faster when combining the tweaks with my original pulse audio package.

The next patch would contain documentation as to why I chose the values as I did, along with my findings and observations.
 

The Following User Says Thank You to SavageD For This Useful Post:
SavageD is offline SavageD
2010-12-11 , 02:04
Posts: 436 | Thanked: 406 times | Joined on Jan 2010
#305
Originally Posted by tusharmax View Post
just removed N900 boost script, and noticed after a reboot that although the device feels more responsive , the game NFSU feels more jerky.. havent tested sound quality tho !
Hmm I too notice this...seems that the device remains fast after removing the patch but certain things, return to normal like the NFS game...Emulators however retain their speed, very confusing. I would have to look into this more, sound quality returns to normal, this I know for certain.

Anyone else notice the same thing after removing the patch?
 
devdsp is offline devdsp
2010-12-12 , 12:37
Posts: 44 | Thanked: 11 times | Joined on Nov 2010 @ kz
#306
Removed.
1. NFS lags, with no background apps running. (When installed no lags even when 2 or more heavy apps running)
2. Media player lags playing FLAC audio, when using device for something else like surfing etc..
3. System lags and slowdowns when app manage updating repos.
4. Battery accord to battery-eye app becomes empty quicker.
Installed again. Waiting for updates.
__________________
Got Milk?
 

The Following User Says Thank You to devdsp For This Useful Post:
SavageD is offline SavageD
2010-12-12 , 14:27
Posts: 436 | Thanked: 406 times | Joined on Jan 2010
#307
Originally Posted by devdsp View Post
Removed.
1. NFS lags, with no background apps running. (When installed no lags even when 2 or more heavy apps running)
2. Media player lags playing FLAC audio, when using device for something else like surfing etc..
3. System lags and slowdowns when app manage updating repos.
4. Battery accord to battery-eye app becomes empty quicker.
Installed again. Waiting for updates.
O lol guess everything returns to normal for some people ;-). Thx man .

Yea got an even faster patch that seems to save the battery a tad bit better than the others, but I want to test epitaphs script first, so might take a while, maybe another week to upload, just making certain everythings safe with this one...
 

The Following User Says Thank You to SavageD For This Useful Post:
tusharmax is offline tusharmax
2010-12-12 , 14:46
Posts: 126 | Thanked: 56 times | Joined on Apr 2010
#308
i installed back the patch again, made a substantial diff in some games and apps
 
tusharmax is offline tusharmax
2010-12-12 , 14:51
Posts: 126 | Thanked: 56 times | Joined on Apr 2010
#309
dude we want the new patch asap, pls pls. btw can something be done to enhance bass effect or better stereo widening effects
 
PartyboyXP is offline PartyboyXP
2010-12-12 , 16:31
Posts: 159 | Thanked: 116 times | Joined on Oct 2010
#310
This patch increase the volume of the speakers or only increase the volume of the earphones? Thanks
 
Reply
Page 31 of 45 « First 212930 31 323341 Last »

« Previous Thread | Next Thread »
Thread Tools

 
Forum Jump


All times are GMT. The time now is 13:55.

Contact Us - Home - Archive - Top