Announcement

Collapse
No announcement yet.

A Fresh Look At The Asahi Linux Performance On Apple's M2

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

  • A Fresh Look At The Asahi Linux Performance On Apple's M2

    Phoronix: A Fresh Look At The Asahi Linux Performance On Apple's M2

    Back during the summer once Asahi Linux introduced initial Apple M2 SoC support I ran many Apple M2 Linux benchmarks including a look at how the M2 competes with AMD and Intel laptop processors. With months having passed since then and the Apple M1/M2 Linux support has continued to advance upstream as well as more work hitting the Asahi Linux tree, here is a fresh look at where the performance of the M2 is currently at compared to that initial at-launch support.

    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
    Interesting results, Michael have you tried running using the
    Code:
    linux-asahi-edge
    repo, according to their latest blog post

    Comment


    • #3
      I wish I could say this was surprising but wasn't there just an article comparing a whole slew of older kernels against 6.1-rc and 5.15 basically came out ahead of them all? Plus, there's that info regarding that 3-liner regression which would figure into the 6.1rc that was probably used for testing. I'd think these numbers would look different after regressions get reverted or fixed and 6.1 gets minted? 6.1rc probably has more M1/M2 HW enablement but at the expense of the existing perf regression. Just needs some time to shake out the regressions. At this rate maybe we'll see 6.1-rc9 next week.

      Comment


      • #4
        Scaling Governor: apple-cpufreq schedutil
        Oh no. This POS governor is poisoning everything.

        Comment


        • #5
          They should rewrite everything in Rust.

          Comment


          • #6
            seems - linux has the same problems on apple hardware like apple has - with every software-update it gets slower! - this is no joke for apple users - but for linux there is hope - the asahi - linux developers will find the glitch which lead to this regression soon - i think - if they not allready had - when they saw your performance analyse. so i am looking forward on this cool project on apple - arm - hardware!

            Comment


            • #7
              Honestly, this is to be expected. You first get things to run, then make it work as expected, then you optimize. When you get things to run, you are usually doing this so wrong that high variable performance is to be expected. Once you start fixing those things, the likelihood your first pass is going to be faster than broken is very low. It does happen, but rarely. Then you can start optimizing for performance.

              Comment


              • #8
                Would be interesting to see the benchmarks vs MacOS; I'm assuming the Test Suite would (mostly) work given the Posix compliance, but even a subset would be very interesting.

                Comment


                • #9
                  Originally posted by spiral_23 View Post
                  seems - linux has the same problems on apple hardware like apple has - with every software-update it gets slower! - this is no joke for apple users - but for linux there is hope - the asahi - linux developers will find the glitch which lead to this regression soon - i think - if they not allready had - when they saw your performance analyse. so i am looking forward on this cool project on apple - arm - hardware!
                  Not in the slightest, the dude just mentioned earlier that there's a generic regression trend in current Linux kernel releases, so it has nothing to do with Apple. I'm sorry it won't fit your narrative this time

                  Comment


                  • #10
                    Originally posted by ferry View Post
                    They should rewrite everything in Rust.
                    You forgot to add /s

                    Comment

                    Working...
                    X