200 line miracle kernel patch?
 

Has anyone else seen this:

http://www.phoronix.com/scan.php?pag...37_video&num=1


If this kernel patch really does such wonders for desktop usage, perhaps it could be applied to the kernel we use for Maemo?

Re: 200 line miracle kernel patch?
 

I don't think this will matter much on n900, unless you're compiling kernels on it all the time ;)

I am really looking forward to titan integrating BFS, that should help much more....

Re: 200 line miracle kernel patch?
 

Looking at the videos, the difference is substantial. This piece of news should reach Titan and MohammadAG as soon as possible... ;) I wish I were able to compile a kernel on my own

Re: 200 line miracle kernel patch?
 

Yes, and now stop looking at the videos and read the text that is above and below them, and more importantly, the text on page 1.

They are comparing responsiveness while overloading the system with a kernel compile using 64 threads.

As I wrote, I am sure BFS will help *much* more than this patch *on n900*

Re: 200 line miracle kernel patch?
 

Isn't the n900 overloaded at times? Think app manager when invokes apt-worker in pr1.2 - is it only i/o or there is also a strenuous cpu effort? And tracker, maybe.
Could this patch be useful in such situations?
I'm curious to know the effects of the patch. Aren't you? ;)

Re: 200 line miracle kernel patch?
 

it would have to be overloaded by several/many threads all wanting full access to the cpu. that really.. never happens on the n900.
apt-worker is just one thread, now if we had 20 of the running at the same time, and you wanted to make a phonecall at the same time, there MIGHT be some benefits with this patch.

Re: 200 line miracle kernel patch?
 

Linux kernel patch offers massive performance boost
 

http://www.phoronix.com/scan.php?pag...37_video&num=1

The power kernel could definitely use this :cool:

If there is going to be a PR1.4 this should be in it.

Re: Linux kernel patch offers massive performance boost
 

there's a lot of choices for which cpu & i/o scheduler to use, and it's not easy to switch between them.
come back when you can tell me what makes this patch better than the CFS or BFS, or why deadline / no-op is supposed to be good for flash memory. confused yet?

-edit-
deadline / noop is i/o scheduler, not cpu...
but my point is, you can swap them all around, it's not necessarily going to improve things.
CFQ is the default i/o scheduler for n900, palm pre, some versions of android...
BFS is used by some custom android roms, most distros use CFS though

Re: Linux kernel patch offers massive performance boost
 

Ooh, that looks cool. I can't wait to see someone take the latest power-kernel and add this too it - who cares if it's "officially" adopted?