Announcement

Collapse
No announcement yet.

Benchmarking Ubuntu 18.04 On Windows Subsystem for Linux: WSL Leading Bare Linux In More Tests

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

  • Benchmarking Ubuntu 18.04 On Windows Subsystem for Linux: WSL Leading Bare Linux In More Tests

    Phoronix: Benchmarking Ubuntu 18.04 On Windows Subsystem for Linux: WSL Leading Bare Linux In More Tests

    Canonical and Microsoft have rolled out Ubuntu 18.04 LTS to the Microsoft Store for running with Windows Subsystem for Linux (WSL). Here are some initial benchmarks of Ubuntu 18.04 LTS on Windows 10 WSL with the April 2018 Update compared to the Bionic Beaver installed bare-metal and then also the older Ubuntu 16.04 WSL setup.

    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
    Give me your time machine for a ride , anyway, small mistake on page 4:
    In the Fhourstones test, WSL was actually faster than the bare metal Ubuntu 18.10 installation

    Comment


    • #3
      The headline may be true, but it's still very misleading as to the performance differences. More accurate might be "WSL not being slaughtered on as many tests by pulling ahead a tiny bit in them--gets killed on all the others by a huge margin".

      Comment


      • #4
        the headline is heresy

        Comment


        • #5
          Seriously now.. How in the @#%$ could you publish this article without comparing it against the previous version of windows in the graphs? Useless. Must have had too many beers before you put it together.

          Comment


          • #6
            on that virus plagued os...

            Comment


            • #7
              In the Fhourstones test, WSL was actually faster than the bare metal Ubuntu 18.04 installation, due to better CPU frequency scaling on Windows or other optimizations
              Or perhaps Windows isn't even close to Spectre and Meltdown mitigation protections in comparison to Linux.

              Comment


              • #8
                Originally posted by darkfires View Post
                Seriously now.. How in the @#%$ could you publish this article without comparing it against the previous version of windows in the graphs? Useless. Must have had too many beers before you put it together.
                He probably couldn't do that. Depending on which edition of Windows he owns he may not be able to prevent automatic updates.
                Only enterprise and education editions allow to postpone feature updates for a limited time.

                Comment


                • #9
                  Michael

                  Two observations:

                  1. Performance cannot be slow or fast, just like price cannot be cheap or expensive. I keep pointing that out and you keep making this error.

                  2. IO performance is low in WSL quite expectedly because unlike Linux Windows runs all created->closed/modified->closed files via Windows Defender or whatever the current AV software in Windows is. Linux on the other hand doesn't have any AV.

                  Comment


                  • #10
                    Originally posted by elvenbone View Post

                    He probably couldn't do that. Depending on which edition of Windows he owns he may not be able to prevent automatic updates.
                    Only enterprise and education editions allow to postpone feature updates for a limited time.
                    Completely and utterly untrue, you have several options that work on even the home edition. It's as simple as turning on metered connection, adding a firewall rule to block svchost.exe from downloading anything or disabling windows update service in services.msc and disabling the task in task scheduler that turns update back on. You can also use winaero tweaker to disable windows updates. Regardless, he previously had done the benchmarks already so it shouldn't have been too difficult to add the data into the graphs since he already has the data.

                    Comment

                    Working...
                    X