Announcement

Collapse
No announcement yet.

Ryzen 9 7950X Performance With The New AMD P-State Default Of Linux 6.5

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

  • Ryzen 9 7950X Performance With The New AMD P-State Default Of Linux 6.5

    Phoronix: Ryzen 9 7950X Performance With The New AMD P-State Default Of Linux 6.5

    The Linux 6.5 kernel is expected to be released as stable this weekend barring any last minute issues from being raised. One of the notable changes with this new kernel version is Linux now defaulting to the AMD P-State "EPP" active driver configuration for modern Ryzen systems rather than the long-used generic ACPI CPUFreq driver default. In some cases this can mean better performance but particularly should yield a nice improvement to the power efficiency of Ryzen Zen 2 and newer platforms, especially laptops and other portable Linux systems like the Steam Deck and ROG Ally. I am working on some fresh AMD Ryzen Linux laptop comparison benchmarks but for this article is a look at Linux 6.5 on the desktop side with the AMD Ryzen 9 7950X.

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    There is always going to be a “I wish they(not just Michael)” item(s). Would have been nice to have a 6.5 control using the same driver to see how much is the driver and how much is the newer kernel in general.

    Overall those are some nice improvements, especially in the short burst benchmarks like the browsers. Not surprised by the compile benchmark.

    Comment


    • #3
      So likely wouldn't help threadripper 1950x?

      Comment


      • #4
        Originally posted by jeisom View Post
        There is always going to be a “I wish they(not just Michael)” item(s). Would have been nice to have a 6.5 control using the same driver to see how much is the driver and how much is the newer kernel in general.

        Overall those are some nice improvements, especially in the short burst benchmarks like the browsers. Not surprised by the compile benchmark.
        FTA: "Again I'll have some laptop benchmarks soon and a broader comparison looking at the different CPU frequency scaling driver/governor combinations as well: such detailed benchmark combinations have been looked at with prior kernel versions in Ryzen Mobile Power/Performance With Linux 6.3's New AMD P-State EPP Driver and AMD P-State EPP Performance With EPYC On Linux 6.3.
        "

        Wanted to get some tests out ahead of Linux 6.5 stable release and this is what I had time for to do this week.
        Michael Larabel
        https://www.michaellarabel.com/

        Comment


        • #5
          Originally posted by Michael View Post

          FTA: "Again I'll have some laptop benchmarks soon and a broader comparison looking at the different CPU frequency scaling driver/governor combinations as well: such detailed benchmark combinations have been looked at with prior kernel versions in Ryzen Mobile Power/Performance With Linux 6.3's New AMD P-State EPP Driver and AMD P-State EPP Performance With EPYC On Linux 6.3.
          "

          Wanted to get some tests out ahead of Linux 6.5 stable release and this is what I had time for to do this week.
          No worries. No matter what you do, someone is going to be disappointed. 😉

          Comment


          • #6
            Am I understanding correctly in that this change improves the thread pinning on processors with multiple CCXs ? Even previous ZEN generations ? Would be nice to see if this helps on the other ZENs with multiple CCXs.

            Comment


            • #7
              Originally posted by Goddard View Post
              So likely wouldn't help threadripper 1950x?
              i don't think it would work on your 1950x because it doesn't have the proper cppc support. i think its zen 2 and later. it be interesting to see results with zen 2 and zen 3 since their desktop parts don't have proper hardware? support for cppc, its software? based support iirc.

              Comment


              • #8
                Once again, FIrefox has been thoroughly beaten by Chrome.

                Comment


                • #9
                  Nice. Beats winblows a$$ even more. fell88 crying in his basement,

                  Comment


                  • #10
                    Originally posted by edxposed View Post
                    Once again, FIrefox has been thoroughly beaten by Chrome.
                    I wonder how much browser benchmarks like Selenium are even relevant anymore? My experience has been Firefox is much better behaved on Linux for the sites _I frequent_. I've unintentionally reused the same window for opening random Youtube videos for weeks and literally had a Firefox session with THOUSANDS of open tabs, and my Zen+ laptop with 16gigs of RAM handles it just fine. Then occasionally I open a random B2B site, car enthusiast forum, etc. and whatever broken javascript they use blows up CPU cycles, memory, or both. Sometimes Firefox chokes on these bad sites (almost never a crash, just bad perf), but sometimes Chromium has its issues as well.

                    Since Firefox has the extensions and privacy tools I want available, then it's still my default.

                    Comment

                    Working...
                    X