Wednesday, 27 November 2019

linux-5.4-ck1, MuQSS version 0.196 for linux-5.4

Announcing a new -ck release, 5.4-ck1  with the latest version of the Multiple Queue Skiplist Scheduler, version 0.196. These are patches designed to improve system responsiveness and interactivity with specific emphasis on the desktop, but configurable for any workload.

linux-5.4-ck1:
-ck1 patches:
Git tree:
MuQSS only:
Download:
Git tree:


Web: http://kernel.kolivas.org


This is a resync from 5.3-ck1, done promptly to make up for the last release's delays.
Enjoy!
お楽しみ下さい
-ck

Posted by at
Labels: , ,

14 comments:

  1. Anonymous27 November 2019 at 15:50

    Thank you very much.

    ReplyDelete
  2. BurBurBur27 November 2019 at 21:58

    You are my hero. Thank you!

    ReplyDelete
  3. Anonymous28 November 2019 at 10:54

    Thank you, runs fine.

    ReplyDelete
  4. Anonymous28 November 2019 at 20:24

    That was fast! Thank you!

    ReplyDelete
  5. Anonymous29 November 2019 at 12:50

    I tested this on an ARM Chromebook (RK3288 CPU) and it seems to work fine, but I had to remove the references to cpu_llc_id and add `select GENERIC_CLOCKEVENTS_MIN_ADJUST` to arch/arm/Kconfig.

    ReplyDelete
    Replies
    1. Anonymous14 April 2020 at 05:52

      What did you do regarding 'cpu_llc_shared_map' in kernel/sched/MuQSS.c?
      It seems to be undeclared for anything but x86.

      Delete
  6. Yevhen20 January 2020 at 03:26

    Con, can you please merge this patch https://github.com/OpenMandrivaAssociation/kernel-release/blob/master/MuQSS-export-can_nice-for-binder.patch ?
    It's needed for kernel with Android Binder module (Ubuntu, Debian, OpenMandriva).

    ReplyDelete
    Replies
    1. Yevhen20 January 2020 at 17:46

      Nevermind, it's a Debian-specific patch ( https://salsa.debian.org/kernel-team/linux/blob/master/debian/patches/debian/export-symbols-needed-by-android-drivers.patch, https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=901492 )

      Delete
  7. Денис27 January 2020 at 07:52

    I experience strange "partial" freezes when low-priority (renice +20 ...) multithreaded process starts heavy calculations saturating all HT-cores.

    Some processes still work fine, but things like firefox freeze for minutes. After calculation finishes, firefox continues to work.

    I have "interactive" on, rr_internal equal to 4, yield_type all values checked, rqshare=mc.

    Any ideas?

    ReplyDelete
    Replies
    1. ck27 January 2020 at 07:54

      How much memory and I/O is that process using? Have you tried sched idleprio instead?

      Delete
    2. Денис27 January 2020 at 08:01

      That process is steady at ~20Gb on 96Gb system. No IO at all. Nope I haven't checked idleprio.

      Delete
    3. Денис27 January 2020 at 08:15

      I've tried using idle_prio, but so far process fails with error failing to create threads. I suspect it might be some library issue though.

      Delete
    4. Денис6 February 2020 at 18:45

      I switched to threadiqrs and rqshare=none (hanging issues on boot) and now everything smooth and nice! No partial freezes at all!
      Thanks, Con.

      Delete
  8. lealrodriguez2 February 2020 at 15:24

    Hi Kon,

    With this patch I get a Kernel Panic.

    See the picture here: https://imgur.com/a/iBWOiz4
    and my dmesg is here: https://pastebin.com/2mRubpti

    Any possibility that this gets fixed?

    I'm running rhel-8 it's a thinkpad T420.

    ReplyDelete

Subscribe to: Post Comments (Atom)