Announcement

Collapse
No announcement yet.

Linux 5.4-rc1 Kernel Steps Forward With Next-Gen GPU Bits, Arm Laptop Support & exFAT

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

  • Linux 5.4-rc1 Kernel Steps Forward With Next-Gen GPU Bits, Arm Laptop Support & exFAT

    Phoronix: Linux 5.4-rc1 Kernel Steps Forward With Next-Gen GPU Bits, Arm Laptop Support & exFAT

    It's coming one day late due to the last minute entropy/RNG patches to improve the random behavior during boot time (among other late patches), but Linus Torvalds has just tagged Linux 5.4-rc1 as what will be the last major stable kernel release of 2019...

    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
    I'm really liking the progress involved on ARM laptops. It'll sure be nice when I can finally ditch my current x86 laptop.
    Last edited by schmidtbag; 30 September 2019, 02:47 PM.

    Comment


    • #3
      I don't know why but every time I see "ARM laptops" I always think "these have more limitations than x86...".

      Comment


      • #4
        Originally posted by tildearrow View Post
        I don't know why but every time I see "ARM laptops" I always think "these have more limitations than x86...".
        I agree, but doesn't open source software have an advantage here in comparison to Windows 10 on ARM? At least when the software wasn't coded full of x86-isms.

        Comment


        • #5
          Originally posted by ms178 View Post

          I agree, but doesn't open source software have an advantage here in comparison to Windows 10 on ARM? At least when the software wasn't coded full of x86-isms.
          i'd love to see steam remote play for linux arm (and maybe power? risc-v in the future?). for me that would be a game changer

          Comment


          • #6
            Building the kernel right now. These amdgpu patches are included in v5.4-rc1 for anyone wondering:

            multi-monitor: support mclk switching when monitors are in sync
            screen corruption/flickering fix on 75 Hz displays
            allow cursor async updates for framebuffer swaps

            Not included are the bulk moves patch and FreeSync low framerate compensation (LFC) patch so I'm still adding those in. (along with Valve's fsync patches).

            Also, it looks like the Makefile finally supports compiling the kernel with -O3 without additional patches.

            diff from 5.3:

            Code:
            < ifdef CONFIG_CC_OPTIMIZE_FOR_PERFORMANCE
            < KBUILD_CFLAGS += -O2
            < else ifdef CONFIG_CC_OPTIMIZE_FOR_PERFORMANCE_O3
            < KBUILD_CFLAGS += -O3
            < else ifdef CONFIG_CC_OPTIMIZE_FOR_SIZE
            < KBUILD_CFLAGS += -Os
            ---
            > ifdef CONFIG_CC_OPTIMIZE_FOR_SIZE
            > KBUILD_CFLAGS    += -Os
            > else
            > KBUILD_CFLAGS   += -O2
            Code:
             config CC_OPTIMIZE_FOR_PERFORMANCE_O3
                 bool "Optimize more for performance (-O3)"
                 depends on ARC
                 imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED  # avoid false positives
                 help
                   Choosing this option will pass "-O3" to your compiler to optimize
                   the kernel yet more for performance.
            (while we're on the subject: to compile with "march=native" along with -O3, you'll want the appropriate graysky gcc patch here)

            EDIT: for x86_64, to enable -O3 you'll have to drop the "depends on ARC" in the code above. By default, it will only show -Os and -O2. So once you add CONFIG_CC_OPTIMIZE_FOR_PERFORMANCE_O3=y to your config, you'll need the following patch:

            Code:
            diff --git a/init/Kconfig b/init/Kconfig
            index b4daad2ba..983a951b3 100644
            --- a/init/Kconfig
            +++ b/init/Kconfig
            @@ -1244,7 +1256,6 @@ config CC_OPTIMIZE_FOR_PERFORMANCE
             
             config CC_OPTIMIZE_FOR_PERFORMANCE_O3
                 bool "Optimize more for performance (-O3)"
            -    depends on ARC
                 imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED  # avoid false positives
                 help
                   Choosing this option will pass "-O3" to your compiler to optimize
            then you should be able to see it when editing the config in General Setup -> Compiler Optimization Level.
            Last edited by perpetually high; 30 September 2019, 11:10 PM.

            Comment


            • #7
              Did I miss the KUnit merge? I see that Shuah said he was going to send a separate pull request but I never saw it posted to the mailing list.

              Comment


              • #8
                They do have many limitations but they also have the one thing that makes them potentially attractive: battery autonomy that a x86-based laptop can never give you.

                Comment


                • #9
                  Originally posted by jacob View Post
                  They do have many limitations but they also have the one thing that makes them potentially attractive: battery autonomy that a x86-based laptop can never give you.
                  Apollo Lake and Gemini Lake SoCs with 6w TDP can already last close to an entire work day on a single charge.

                  Comment


                  • #10
                    Originally posted by Sonadow View Post

                    Apollo Lake and Gemini Lake SoCs with 6w TDP can already last close to an entire work day on a single charge.
                    But what can you actually do with them if they are to last entire day? My dream is a laptop I could charge once and use for a ~14 hours flight to watch movies, not just stare at a still desktop.

                    Comment

                    Working...
                    X