Announcement

Collapse
No announcement yet.

MuQSS Updates For The Linux 4.10 Kernel

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • MuQSS Updates For The Linux 4.10 Kernel

    Phoronix: MuQSS Updates For The Linux 4.10 Kernel

    Con Kolivas has announced the MuQSS CPU scheduler v0.152 release with support for the Linux 4.10 kernel...

    http://www.phoronix.com/scan.php?pag...QSS-Linux-4.10

  • #2
    What are the drawback of MuQSS?
    Why isn't this in the mainline kernel?
    Why isn't this used by default if it so good?

    Comment


    • #3
      He, he, he might be popularising this now as i see he started providing Ubuntu packages... 4.9 for 16.04 and 4.10 for 16.10

      http://ck.kolivas.org/patches/4.0/4....1/Ubuntu16.04/
      http://ck.kolivas.org/patches/4.0/4....1/Ubuntu16.10/

      And i guees Michael will do some bench using these

      Comment


      • #4
        I feel like somebody should find a more specific number than "100hz". It seems entirely arbitrary. Why not 60hz (typical monitor refresh rate) or 120hz (double that)?

        Comment


        • #5
          Some years ago i think Con had some patches with 864Hz or something like that as ideal for Desktop

          eidt: on first sight, these Ubuntu kernels seems full but firmwares missing
          Last edited by dungeon; 20 February 2017, 09:18 AM.

          Comment


          • #6
            Originally posted by uid313 View Post
            What are the drawback of MuQSS?
            Why isn't this in the mainline kernel?
            Why isn't this used by default if it so good?
            Because he's been disgusted with how the kernel's hierarchy works. He actually proposed several schedulers to improve desktop responsiveness in the mid-2000's that were refused, then Ingo Molnar up and wrote the Completely Fair Scheduler that was pretty much a re-hash of Con's ideas; when the latter protested, Molnar acknowledged the influence; Con insisted that CFQ was still made for 16-core workstations, not Average Joe's single or dual core.

            He then decided to implement the Brain Fuck Scheduler as a low-overhead, responsive scheduler for the everyday Joe ; acknowledging that his BFQ was geared towards low core-count CPUs, he rewrote it as MuQSS following the same design principles but with better scaling on multi-CPU architectures. Still, he has no desire to upstream it himself. If you feel like it...

            The main drawback of MuQSS is that it might not be as versatile as CFQ; barring people ready to test it on a very wide range of hardware, it won't be mainlined. As a matter of fact, CFQ only came about because a thousand core distributed system was seeing a very high overhead from 0(1) and was ready to test a new one.
            mitch074
            Senior Member
            Last edited by mitch074; 20 February 2017, 09:25 AM.

            Comment


            • #7
              Originally posted by debianxfce View Post
              The mainline kernel is barely tested with 1000Hz kernel timer...
              As Fedora uses 1000Hz timer does that mean it is barely tested.

              , try that and you notice some glitches when using a desktop.
              Please describe these glitches thoroughly, it looks like you just have unreported bug or you just messed up your config
              Last edited by dungeon; 20 February 2017, 09:57 AM.

              Comment


              • #8
                Originally posted by mitch074 View Post
                Con insisted that CFQ was still made for 16-core workstations, not Average Joe's single or dual core.
                So with Ryzen offering 8 core 16 thread and later a 16 core 32 thread variant, are you better off with CFQ with such a CPU or would BFQ or MuQSS have something to offer? I'm planning on a high core count CPU later in the year for a desktop/workstation machine where the extra cores will help a bunch with running VMs and other workloads. Even just as a regular desktop use case with one of those CPU would BFQ/MuQSS have a negative impact on the system with that hardware?

                Comment


                • #9
                  Since someone liked this picture in a different thread, I thought it could be nice to share it here too...

                  Comment


                  • #10
                    Originally posted by polarathene View Post
                    So with Ryzen offering 8 core 16 thread and later a 16 core 32 thread variant, are you better off with CFQ with such a CPU or would BFQ or MuQSS have something to offer?
                    You'll probably have to test this. From experience, BFS offered somewhat better responsiveness when the system was under full load, but it had *massive* core hopping issues, cutting down the single-threaded performace in half when using the ondemand governor - I'm currently compiling a -ck kernel to test whether this has been resolved with MuQSS.

                    Edit: No, it hasn't, single-threaded tasks still perform poorly, hopping around between Cores 0-2 while cores 3-5 are doing absolutely nothing. With CFS there's hardly any hopping, but CFS doesn't even allow me to watch youtube videos while compressing something with pbzip2 in the background. Meh.
                    VikingGe
                    Senior Member
                    Last edited by VikingGe; 20 February 2017, 12:02 PM.

                    Comment

                    Working...
                    X