Announcement

Collapse
No announcement yet.

Linux 5.17 Released With AMD P-State Driver, Plenty Of New Hardware Support

Collapse
This topic has been answered.
X
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #11
    Is there some easy way to check that P-state driver is being used? I know how to check that UEFI supports it at least.

    Comment


    • #12
      In Fedora, I had to blacklist the acpi-cpufreq driver so amd-pstate would load. The AMD developers told me from a thread with them. Once that was done, then I could use it.

      Comment


      • #13
        Originally posted by spstarr View Post
        In Fedora, I had to blacklist the acpi-cpufreq driver so amd-pstate would load. The AMD developers told me from a thread with them. Once that was done, then I could use it.
        I guess acpi-cpufreq takes priority and is conflicting with amd-pstate?

        I can see it in my lsmod:

        Code:
        lsmod | rg freq
        acpi_cpufreq 28672 0

        Comment


        • #14
          I found this page: https://www.kernel.org/doc/html/late...md-pstate.html

          Which says:

          shared_mem Use a module param (shared_mem) to enable related processors manually with amd_pstate.shared_mem=1. Due to the performance issue on the processors with Shared Memory Support, so we disable it for the moment and will enable this by default once we address performance issue on this solution.
          Shared mem will be needed on regular CPUs, since "full MSR" is only available on APUs. Does it mean there are some performance problems with amd-pstate now with regular CPUs?

          Comment


          • #15
            Originally posted by shmerl View Post
            I found this page: https://www.kernel.org/doc/html/late...md-pstate.html

            Which says:



            Shared mem will be needed on regular CPUs, since "full MSR" is only available on APUs. Does it mean there are some performance problems with amd-pstate now with regular CPUs?
            Yes, especially in combination with the default "schedutil" governor, as Michael's benchmarks are going to reveal!
            However, as always, "performance" users will be fine.

            And since you wanted an easy way to check which CPU driver is in use:
            Code:
            sudo cpupower -c all frequency-info
            Look out for the first line of output.

            Comment


            • #16
              Thanks for the pointers!

              Originally posted by Linuxxx View Post
              Yes, especially in combination with the default "schedutil" governor, as Michael's benchmarks are going to reveal!
              So we need to wait for them fixing some issues before it becomes useful on regular CPUs?

              Being fine with performance rather than schedutil kinds of defeats the purpose this, since it just runs at max clocks.
              Last edited by shmerl; 20 March 2022, 11:01 PM.

              Comment


              • #17
                Originally posted by shmerl View Post
                Thanks for the pointers!



                So we need to wait for them fixing some issues before it becomes useful on regular CPUs?

                Being fine with performance rather than schedutil kinds of defeats the purpose this, since it just runs at max clocks.
                You're welcome!
                After all, sharing is caring, you know...

                And yes, the reason why the performance governor is so damn efficiently effective is because it will always go for the maximum clocks immediately.
                However, even on my i5-6500 with Intel's measly stock-cooler, this never causes any problems, because any modern CPU will always go into deep-sleep states intermittently, thus saving more power than working longer at reduced clockspeeds.

                A good way to test this and see it for yourself in action is to playback a software-decoded video while periodically looking at the output of yet another very useful command of CPUPOWER:

                Code:
                sudo cpupower monitor
                Here, you can read the percentages of the different C-states on the right, where the higher the number, the deeper the sleep.

                At least on my Intel system, even though my CPU is under constant load caused by the video playback, I can observe that it still manages to enter all sleep states successfully and only causing a minor increase in heat output, varying on how heavy the video in question is to decode.

                If You could test this out on Your Ryzen system, then I would certainly be interested to find out about Your observations.

                Take care!

                Comment


                • #18
                  Shouldn't running at performance use more power than necessary to achieve the same task? That's the whole point of schedutil after all, to scale frequency depending on the load. If it's not happening in your case - something is wrong with the driver I assume.

                  The whole point of amd-pstate is just to allow more granular control over frequency ranges. So surely you shouldn't be using performance for optimal operation if everything will be working as intended. But it sounds like their current implementation has some problem so far.
                  Last edited by shmerl; 21 March 2022, 02:02 AM.

                  Comment


                  • #19
                    Originally posted by Vorpal View Post
                    Huh? I'm not a native English speaker, so maybe it is some idiom I don't know, but I'm utterly confused.

                    I am a native Swedish speaker (Torvalds native language). Though I speak mainland Swedish, not Finland Swedish (there are subtle differences). That doesn't help decode this. And I don't speak Finnish at all.
                    Originally posted by NateHubbard View Post
                    I assume it's a reference to brown paper bags, but I don't get it either.
                    Originally posted by tildearrow View Post

                    Could also be that Torvalds has been in the US for a long time.
                    Originally posted by lumks View Post
                    it's something like this
                    Originally posted by cl333r View Post
                    And what does "stable fodder" mean?
                    Originally posted by numasan View Post
                    Probably the comments about how many bugs there are in the stable release?
                    Originally posted by moriel5 View Post
                    Vorpal cl333r I believe that "fodder" means food (or perhaps prey? English isn't my native language, however I did live in the US during part of my childhood), so perhaps food for the horses?

                    I would hazard a guess that the point is that something might not be as worth doing as it originally seems, or something similar to that.
                    Originally posted by er888kh View Post

                    I am not a native speaker, but this is what I think:
                    - Brown paper is just some cheap paper you wrap other stuff with
                    - Fodder is some low quality food that cannot be consumed by people, so it is fed to livestock

                    So my guess is:
                    "There were some easy to fix bugs that would be solved after rc7; if it wasn't for the rc8, they would have been fixed (some trivial and boring fixes) in the first stable release. And they were not important or interesting bugs at all."
                    Right all of you Linux kernel world has some unique terms.

                    http://catb.org/~esr/jargon/html/B/b...r-bag-bug.html
                    A bug in a public software release that is so embarrassing that the author notionally wears a brown paper bag over his head for a while so he won't be recognized on the net. Entered popular usage after the early-1999 release of the first Linux 2.2, which had one. The phrase was used in Linus Torvalds's apology posting.
                    The brown paper bag bug but is a term created by Linus Torvalds in 1999 for a bug in the Linux kernel 2.2 that everyone at the time who had anything to-do with it wanted to pretend it was not their fault because it was so bad.

                    Yes brown paper bags bugs (Linux kernel) and Blind Freddy(Australian legal usage) are basically terms that been "glaringly obvious mistakes" that have some how been missed without exactly calling a person out for "glaringly obvious mistakes".

                    Brown paper bag bug have nothing to-do with Swedish or Linus being the the USA for a long term. Yes Linux Torvalds coined the term brown paper bag bug when he had only been in the USA for 2 years.

                    There is a long list of cartons and anime and movies where a brown paper bag is used to hide a person identity. And some of those Linux Torvalds is known to have watched before leaving Finland. So this one is most likely nothing todo with Finland culture. Yes the brown paper bag mask we are not exactly sure where that first started. So is a quite common culture around the world of the brown paper bag to hide identity when brown paper bags were in heavy retail usage. Of course this has dropped in usage since plastic bags have taken over in retail. So you might say not getting the brown paper bag bug things is part age being too young to know the brown paper bag mask stuff for hiding identity.

                    Stable fodder
                    https://lore.kernel.org/all/20170509...inux.org.uk/T/
                    I don't have a good define for stable fodder. Stable Fodder is used for fixes to stable branches of the Linux kernel that fix stability problems like crashing. Yes it does have some relationship for fodder term for feed for live stock but those two words usage for Linux kernel developer is basically a class of bug fix. Yes bug fix for quite serous bugs.

                    Comment


                    • #20
                      To use the P-State driver on Ubuntu with a desktop CPU I had to do this:

                      Added "initcall_blacklist=acpi_cpufreq_init amd_pstate.shared_mem=1" to GRUB_CMDLINE_LINUX_DEFAULT in /etc/default/grub
                      Added "amd_pstate" to /etc/initramfs-tools/modules

                      Comment

                      Working...
                      X