Page 2 of 15 FirstFirst 123412 ... LastLast
Results 11 to 20 of 150

Thread: RIFS-ES Linux Kernel Scheduler Released

  1. #11
    Join Date
    Oct 2011
    Posts
    56

    Default

    Another idea regarding your statistic:
    It is not meant to show the latency of the overall system, but the performance of a single function ordering the loads compared?
    (the logN efficiency you mentioned erlier)

  2. #12
    Join Date
    May 2012
    Location
    GuangDong,China
    Posts
    141

    Default

    Quote Originally Posted by ulenrich View Post
    Another idea regarding your statistic:
    It is not meant to show the latency of the overall system, but the performance of a single function ordering the loads compared?
    (the logN efficiency you mentioned erlier)
    It shows the latency of 'sleep 10'

  3. #13
    Join Date
    Dec 2011
    Posts
    2,102

    Default

    Quote Originally Posted by ulenrich View Post
    going up from
    60 to 200 clients
    you will have latency going from
    50000 down to 5000

    This is funny!
    Is this a quantum machine running faster on high loads?
    Yes, this is very weird.
    I don't think this is possible.
    Why the graph look like this?

  4. #14
    Join Date
    May 2012
    Location
    GuangDong,China
    Posts
    141

    Default

    Quote Originally Posted by uid313 View Post
    Yes, this is very weird.
    I don't think this is possible.
    Why the graph look like this?
    Hi

    Here is a newer benchmark around latt -c255 sleep 10 between these 3 cpu scheduler.
    http://code.google.com/p/rifs-schedu...Result-003.PNG

    RIFS-ES-Low-Spec has been posted, if you want a link please click the follow site.
    http://rifs-scheduler.googlecode.com...ec-kernel3.4.x

    After using it if you think it is useful please invite the others to use it.
    Thanks

    Chen

    EDIT 2

    All right I am going to make a conclusion for the benchmark
    Although BFS get the worst average latency but it also get the best max_latency value.
    Last edited by 3766691; 06-21-2012 at 06:10 AM.

  5. #15
    Join Date
    Dec 2011
    Posts
    2,102

    Default

    Quote Originally Posted by 3766691 View Post
    Hi

    Here is a newer benchmark around latt -c255 sleep 10 between these 3 cpu scheduler.
    http://code.google.com/p/rifs-schedu...Result-003.PNG
    Looks very weird.
    The latency jumps up and down unpredictably instead of scaling smoothly.

  6. #16
    Join Date
    May 2012
    Location
    GuangDong,China
    Posts
    141

    Default

    Quote Originally Posted by uid313 View Post
    Looks very weird.
    The latency jumps up and down unpredictably instead of scaling smoothly.
    Even it jumps up and down it is still lower than the other 2 scheduler.
    During the test I will move the mouse(I just move the mouse and I don't click anything).
    With BFS the mouse stalls.
    With CFS or RIFS-ES-Low-Spec the mouse is very smooth.

    With average latency RIFS wins, with maximum latency BFS wins.
    Chen

  7. #17
    Join Date
    Oct 2011
    Posts
    56

    Default smooth! well done Chen!

    I just run a linux-3.4.4rc using the additional patch
    RIFS.ES-v1-low-spec-kernel3.4.x

    No issues with audio as was before!
    Very smooth experience with high load
    compile + flash + video
    all at the same time!

    Thank you very mush Chen!

    The BFS alternative will have issues with the newest linux-3.4.4
    regarding a rpc patch as I pointed to at Kon Colivas blog.

  8. #18
    Join Date
    May 2012
    Location
    GuangDong,China
    Posts
    141

    Default

    Quote Originally Posted by ulenrich View Post
    I just run a linux-3.4.4rc using the additional patch
    RIFS.ES-v1-low-spec-kernel3.4.x

    No issues with audio as was before!
    Very smooth experience with high load
    compile + flash + video
    all at the same time!

    Thank you very mush Chen!

    The BFS alternative will have issues with the newest linux-3.4.4
    regarding a rpc patch as I pointed to at Kon Colivas blog.
    Thanks for trying.
    If the mainline can make CFS smooth then RIFS will stop publishing.
    The existance of RIFS is to make the kernel smooth.
    If you want you can invite the other to us it, or you can hack the scheduler and post your hacking.

    We have to let the maintainer to know that they should improve the desktop user experience now.

    Chen

  9. #19
    Join Date
    May 2012
    Location
    GuangDong,China
    Posts
    141

    Default

    Quote Originally Posted by 3766691 View Post
    Thanks for trying.
    If the mainline can make CFS smooth then RIFS will stop publishing.
    The existance of RIFS is to make the kernel smooth.
    If you want you can invite the other to us it, or you can hack the scheduler and post your hacking.

    We have to let the maintainer to know that they should improve the desktop user experience now.

    Chen
    I have made an decision that I will still continue maintaining and developing it even after the mainline has improved the interactive

  10. #20
    Join Date
    Oct 2011
    Posts
    56

    Default

    I am publishing my esperience here:
    http://siduction.org/index.php?name=...93b39deb#21450

    Siduction is an effective an little community around Debian unstable. Towo, the kernel guy there, had applied BFS patches for Linux-3.2 back then. But you dont have to use it to discuss in the forums. My main distribution yet is Gentoo, which is not only a "rollin' release" but a rolling MYrelease.

Posting Permissions

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