Page 1 of 14 12311 ... LastLast
Results 1 to 10 of 138

Thread: CK patch set

  1. #1
    Join Date
    Oct 2006
    Beans
    124
    Distro
    The Feisty Fawn Testing

    CK patch set

    Hello forum folks. I'm not sure where to put this. Can you please move this to an appropriate spot?

    I am creating this forum topic to support this bug report which involves the CK patchset for the kernel: https://bugs.launchpad.net/ubuntu/+s...ux/+bug/424927
    Please post any and all comments and bugs and crashes and help requests to this forum topic. If you find something wrong with, or have a request regarding, the packaging or confguration of the ck patched kernel, please post on the bug report. Thanks!

    The CK patch set is intended to improve desktop interactivity performance. The biggest change is the inclusion of BFS: http://en.wikipedia.org/wiki/Brain_****_Scheduler (The link is censored because it includes a common swear word)

    https://launchpad.net/~chogydan/+archive/ppa
    Last edited by danbh; April 1st, 2011 at 07:41 PM. Reason: updating with new info

  2. #2
    Join Date
    Jan 2008
    Beans
    4,311

    Re: CK patch set

    Hi! I plan on using and testing your packages. Everything seems to go well. Need any assistance?

  3. #3
    Join Date
    Oct 2006
    Beans
    124
    Distro
    The Feisty Fawn Testing

    Re: CK patch set

    I'm finally able to push an update to a released kernel. I hadn't been able to get the source code till last week.

    NightwishFan, I don't really know about the bigger picture, ie getting this into the main archives, but I think it would be useful if people could figure out whether this patch is worth it. I know that it was _way_ better in karmic. Please, report back what kind of performance you are getting. better? worse? the same? I haven't had a chance to test myself. I had been waiting for a new release, just because I was having trouble with the older releases.

  4. #4
    Join Date
    Feb 2007
    Beans
    62

    Re: CK patch set

    Hey, I have been using your build with the BFS patch...First of let me say thank you for taking the time to do it, and second of all, one major difference I have noted in lucid using your ck-preempt compared to the mainline preempt is a way lower load level, plus when it comes to compiling anything, the BFS scheduler really shows its worth, it has been cutting down most of my compile times by around half!!...of course nothing's perfect, I have been having issues with waking up from suspend...but I noticed that today Con Kolivas updated his patch and the changelog hints that it may address this, and now it seems that you are updating the ppa, so we will see..........thanks again

  5. #5
    Join Date
    Oct 2006
    Beans
    124
    Distro
    The Feisty Fawn Testing

    Re: CK patch set

    I'm glad you are getting such good results chronniff. At some point we will have to tabulate all of our experiences.

    I realized that I was pretty vague about what I meant with "new release." Since 2.6.32-20 through 2.6.32-22, I had always been getting -21 when I tried to download the source. I didn't get it, it was weird, but it looks like it is over. Thusly, I was able to send an update to the ppa.

    Also of note, I reconfigured. I think I got it right. AMD64 server builds should be configured to 100HZ, and everything else should be 1000HZ. Hopefully that fixes the AMD64 being at 250Hz that was reported earlier.

  6. #6
    Join Date
    Jan 2008
    Beans
    4,311

    Re: CK patch set

    I thank your hard effort. After I solve my bout of personal tragedy I will be in touch. I really like easy access to the desktop oriented kernel patches and I pledge my support in keeping it going. I will probably be settled in a week and be able to help out, even if only with testing.

    Feel free to contact me here or otherwise, my email is:
    uraharakisuke153@gmail.com

  7. #7
    Join Date
    Nov 2008
    Beans
    21

    Re: CK patch set

    Dan, thanks for making the ppa available. I've tried the ck kernels on several laptops and desktops, and the results (seem to) range from little different to apparently smoother user experience. Great work!

    It would be useful if you indicate that metrics are meaningful for you in comparing performance of bfs v cfs scheduling. At the moment it seems like keyboard and interactive graphics (eg. OpenOffice) are the areas there is noticeable improvement, but even then, it is purely qualitative.

  8. #8
    Join Date
    Feb 2007
    Beans
    62

    Re: CK patch set

    just to specify my testing of this, I have used it on both my new custom core i7 960 desktop, and also on an old dell inspiron 1420 with core 2 duo chip and integrated (yup those first ones dell sold with ubuntu preinstalled lol!).....results on the laptop are minimal performance enhancement, although I haven't tested it much
    On the core i7 however, there is an extremely noticeable performance boost..it's a quad core using hyperthreading remember, so 4 x2 equals 8 logical cpu cores. BFS does a much better job taking advantage of the multiple cores...If we are trusting what conky shows for each core's % of use, when running BFS the use of each core is relatively equally distributed...as opposed to the mainline CFS which makes use of only a few cores at a time for the most part resulting in some cores not even being used while others are over 50% usage....I haven't done any real benchmarking, but am happy to if someone has a particular benchmark in mind let me know.
    I can say that everything is much faster on the i7, and I know for sure that to compile a kernel takes somewhere between 15-20 minutes on the mainline CFS preempt kernel, and using the BFS-preempt it takes 8 to 10 minutes

  9. #9
    Join Date
    Feb 2007
    Beans
    62

    Re: CK patch set

    also, Dan feel free contact me if you need my help testing anyting as well.

    I'm at:
    daveconniff@gmail.com

  10. #10
    Join Date
    Nov 2008
    Beans
    21

    Re: CK patch set - possible problem

    Heads up - there is a possible problem.

    I was using -22ck kernel (AMD64) on my Dell Studio 15 with Radeon RV620 graphics. What I noticed a couple of times, fairly consistently, was that after 2-3 suspend/resume cycles, X performance would slow down. Response to mouseclicks, cycling through windows, window refreshes all became jerky.

    I had previously never seen this on my system. I have reverted to the non BFS kernel, and will look out for this slowdown problem.

    On my MSI Wind netbook, repeated suspend/resume cycles does not show this problem with -22ck kernel.

Page 1 of 14 12311 ... LastLast

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •