Announcement

Collapse
No announcement yet.

Arch Linux, Clear Linux, Fedora Compete On The ASUS ROG Strix G15

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

  • #11
    Originally posted by david-nk View Post
    It's frustrating that after several years, Clear Linux is still ahead of the rest.
    Why isn't upstream incorporating Intel's changes?
    Because upstream can't be tuned towards Intel alone?

    Comment


    • #12
      Originally posted by bug77 View Post

      Because upstream can't be tuned towards Intel alone?
      Michael also showed that Clear Linux had benefits on AMD hardware. Besides ARM, Intel and AMD are the most popular, so if upstream included such changes, it would benefit the majority of desktops, laptops and non-ARM servers.

      Comment


      • #13
        Originally posted by MadCatX View Post

        As I recall Clear Linux used to be miles ahead of other distros. Now it pulls ahead only in a few specific tasks and even then the difference is usually not that great. My understanding is that not all of the performance tweaks used by Clear Linux are usable in a generic distro that is supposed to work everywhere. Clear may also be using some experimental patches that are not ready for upstream yet...
        Pretty much that. Intel tunes their compilers for Skylake/AVX-era and up feature sets which is why Clear works and gives boosts on AMD systems too and is why Clear won't work on older Intel Core, AMD Phenom and X64, and other older x86_64 architectures, why every distribution can't pickup Clear's performance benefits, and why we have such passionate discussions about x86_64_v# levels and repositories.

        Comment


        • #14
          Originally posted by andyprough View Post
          I'm surprised Clear didn't do a lot better. Those are three of the consistently worst performing distros that you put it up against. You used to get really strong results from Debian, but it seems like you've stopped testing it in these multi-distro competitions.
          I think it's more that you're seeing other distros catch up to the latest version of the latest and greatest. That's where all the performance gains are coming. That's where all the performance gains will always come. That and setting the tweaks out-the-box that you know 90% of people probably won't touch. ClearLinux goes that extra mile, and it shows.

          Happy to see my Arch bretherens (see I'm trying to be better) doing well. Tweak your kernels!! And do everything incrementally (and benchmark) and have fun while you're doing it.

          Comment


          • #15
            Originally posted by bug77 View Post
            In other words, besides Arch, no other distro worked out of the box.
            Fedora do release respins of fully updated images for cases like this. I usually use these on brand new hardware. Unfortunately most people don't know they exist, it's not a well publicized thing. Here you go: https://fedoraproject.org/wiki/Respins-SIG

            Comment


            • #16
              Originally posted by perpetually high View Post
              I think it's more that you're seeing other distros catch up to the latest version of the latest and greatest. That's where all the performance gains are coming. That's where all the performance gains will always come.
              Is this round of testing really showing performance gains? New hardware tends to give incremental gains, but newer versions of packages just as often introduce regressions. What we are seeing here is just as likely that newer kernel regressions are pulling Clear back to the pack, rather than the pack catching up with Clear. That's another good reason to continue putting Debian in these competitions. For quite awhile the 4.19 kernel was consistently out-performing newer kernel versions.

              Comment


              • #17
                Originally posted by jntesteves View Post

                Fedora do release respins of fully updated images for cases like this. I usually use these on brand new hardware. Unfortunately most people don't know they exist, it's not a well publicized thing. Here you go: https://fedoraproject.org/wiki/Respins-SIG
                Thank you very much. I didn't know the existence of those images and I'm a Fedora user.




                Comment


                • #18
                  Originally posted by andyprough View Post

                  Is this round of testing really showing performance gains? New hardware tends to give incremental gains, but newer versions of packages just as often introduce regressions. What we are seeing here is just as likely that newer kernel regressions are pulling Clear back to the pack, rather than the pack catching up with Clear. That's another good reason to continue putting Debian in these competitions. For quite awhile the 4.19 kernel was consistently out-performing newer kernel versions.
                  Yeah that's fair. I actually use the 5.4 kernel for that reason. But, you do lose out on some of the features you may want that came later down the road. I was able to patch in and backport a few things, but somethings I still couldn't get going.

                  And it makes sense when you think about it. Unless it's an obvious performance tweak, you're just adding more lines of code and wasting more execution time on stuff that doesn't apply to your setup (and all the new security stuff that came after that 4.19 kernel and 5.4 kernel). So I think there is a tradeoff and balance between using a really old kernel and a newish one that has new features. For me, 5.4 was that. Wasn't too impressed with 5.10.

                  One thing I did forget, is that with AMD gpu's and 5.4, it's using 28-30W for my multi-monitor setup, 5.13 and the newer ones are half of that at 12-14W, so I've been actually using the latest 5.13 kernel for that reason, plus all the other new featuers that are packed into the kernel. But benchmarks show that it is definitely slower than 5.4.

                  Comment


                  • #19
                    Originally posted by david-nk View Post
                    It's frustrating that after several years, Clear Linux is still ahead of the rest.
                    Why isn't upstream incorporating Intel's changes?
                    It's frustrating to believe Linux vendors are chasing performance over security. Fedora uses SeLinux and a good number of GCC flags which do slow down the resulting binary code. You're free to use Gentoo/LFS and disable all the built-in security mechanisms, including the Spectre mitigations in the kernel.

                    Best regards,
                    Sanity.

                    Comment


                    • #20
                      Originally posted by birdie View Post

                      It's frustrating to believe Linux vendors are chasing performance over security. Fedora uses SeLinux and a good number of GCC flags which do slow down the resulting binary code. You're free to use Gentoo/LFS and disable all the built-in security mechanisms, including the Spectre mitigations in the kernel.

                      Best regards,
                      Sanity.
                      I've wanted to do that for the past year -- Gentoo with and without the mitigations, mitigated toolchains, etc to see how much of an impact they actually have. Unfortunately, I'd rather play games with my free time than benchmark various Gentoo builds.

                      Comment

                      Working...
                      X