Announcement

Collapse
No announcement yet.

Ubuntu 18.10 Aims To Lower Power Use, Default To New Desktop Theme

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

  • Ubuntu 18.10 Aims To Lower Power Use, Default To New Desktop Theme

    Phoronix: Ubuntu 18.10 Aims To Lower Power Use, Default To New Desktop Theme

    Will Cooke, the Director of the Ubuntu Desktop at Canonical, has outlined the major desktop plans for the Ubuntu 18.10 "Cosmic Cuttlefish" cycle...

    http://www.phoronix.com/scan.php?pag...-Desktop-Plans

  • #2
    - Following the work done by Red Hat on improving power consumption with Fedora 28, Canonical just so happens to now be planning power consumption improvements for Ubuntu 18.10 too. They will try to turn on more power-saving settings without hurting stability.
    Yeah good luck with that. ACPI is notorious and not just in Linux or the rest of the open source world. It can be horribly broken even with OEM Windows drivers trying to cover up non-functional states - laptops forgetting function and pad settings, broken suspend/resume, desktops subtly corrupting RAM contents after resume, etc. Just for example, my almost-brand-new (year old) MSI AMD B350 chipset main board will hang on a shutdown/reboot if it's been sent into suspend mode for that session when running Linux (works fine in Windows). I've experienced similar brokenness from other motherboard providers over the years both Intel and AMD, so yeah, that "without hurting stability" is pretty much a pipe dream as far as power management is concerned. Were I to guess, they failed to communicate the part where it doesn't affect stability only applies to the core team's computers.
    Last edited by stormcrow; 05-18-2018, 01:00 PM. Reason: grammar

    Comment


    • #3
      Originally posted by stormcrow View Post

      Yeah good luck with that. ACPI is notorious and not just in Linux or the rest of the open source world. It can be horribly broken even with OEM Windows drivers trying to cover up non-functional states - laptops forgetting function and pad settings, broken suspend/resume, desktops subtly corrupting RAM contents after resume, etc. Just for example, my almost-brand-new (year old) MSI AMD B350 chipset main board will hang on a shutdown/reboot if it's been sent into suspend mode for that session when running Linux (works fine in Windows). I've experienced similar brokenness from other motherboard providers over the years both Intel and AMD, so yeah, that "without hurting stability" is pretty much a pipe dream as far as power management is concerned. Were I to guess, they failed to communicate the part where it doesn't affect stability only applies to the core team's computers.
      I think Linux partially suffers from the unspoken assumption that it should work perfectly on Windows hardware. More often than not it actually does, which is remarkable, but ultimately to avoid compatibility and support problems, the only solution is to buy machines certified for Linux. Fortunately there are many of those now. If you want to run MacOS, normally you wouldn't think of buying anything other than an Apple, if you want to use Windows you look for that omnipresent Windows compatible logo, and it's basically the same with Linux. You want stable suspend&resume in Linux? Go on Dell, HP or System 76 and get yourself a Linux computer, it's as simple as that.

      Comment


      • #4
        Snaps are a terrible idea. Huge waste of space, just so some apps can use outdated libraries who don't know how to do versioned/backwards compatible interfaces correctly. What a shitshow. In my eyes it completely goes against one of the main advantages of using Linux: the package manager.

        I think I'll pass on Ubuntu and its future releases. Snap philosophy is poop.

        Comment


        • #5
          Originally posted by jacob View Post

          I think Linux partially suffers from the unspoken assumption that it should work perfectly on Windows hardware. More often than not it actually does, which is remarkable, but ultimately to avoid compatibility and support problems, the only solution is to buy machines certified for Linux. Fortunately there are many of those now. If you want to run MacOS, normally you wouldn't think of buying anything other than an Apple, if you want to use Windows you look for that omnipresent Windows compatible logo, and it's basically the same with Linux. You want stable suspend&resume in Linux? Go on Dell, HP or System 76 and get yourself a Linux computer, it's as simple as that.
          No, sorry, but it's not "simple as that". You entirely ignored my point that even hardware 'certified' for MS Windows is sometimes fundamentally broken and yes, from these very same vendors that you're listing (well Dell and HP anyway... I know little about System 76). It's immaterial what operating system a PC is certified to run on if the firmware, chip sets, or processors contain bugs that hamper operation on any OS, including Microsoft Windows. That's the point I'm getting at. ACPI as a standard is incomplete, and when hardware companies can't even get implementations right which are not ambiguous or incomplete like USB or PCI-E (see the crap with broken USB charging implementations from various vendors like Nintendo's Switch , that Google engineer that tests cable certifications finding many manufacturers that can't even put in the right resistors, non-compliant but supposedly compliant mass storage, AMD's RX480 out-of-spec PCI-E bus power draw, there are many others) what do you think is going to happen when you have a purposely incomplete standard where a certain part is written and the rest is up to the manufacturer? That's right, they're going to fuck it up. Often.

          I don't expect (any) closed hardware to work "perfectly" on Linux (or BSD) whether its certified or not. I've been around the block enough times to know that's not always going to be the case. But in the case of power management, the PC world is in general a crapshoot. What I mentioned up there in my OP was not just my experience with Linux, but my experience with Windows PCs as well. The RAM corruption issue on resume? Windows XP resume on an HP Pavilion. Laptops forgetting their user settings on resume from sleep? Windows. Sometimes this isn't even a firmware problem, it can actually be the CPU itself which has nothing to do with which OS it runs - see the C-state problems with some Intel CPUs over the years, Nehalem was one of those and no software fix was going to fix it.

          Comment


          • #6
            Originally posted by stormcrow View Post

            No, sorry, but it's not "simple as that". You entirely ignored my point that even hardware 'certified' for MS Windows is sometimes fundamentally broken and yes, from these very same vendors that you're listing (well Dell and HP anyway... I know little about System 76). It's immaterial what operating system a PC is certified to run on if the firmware, chip sets, or processors contain bugs that hamper operation on any OS, including Microsoft Windows. That's the point I'm getting at. ACPI as a standard is incomplete, and when hardware companies can't even get implementations right which are not ambiguous or incomplete like USB or PCI-E (see the crap with broken USB charging implementations from various vendors like Nintendo's Switch , that Google engineer that tests cable certifications finding many manufacturers that can't even put in the right resistors, non-compliant but supposedly compliant mass storage, AMD's RX480 out-of-spec PCI-E bus power draw, there are many others) what do you think is going to happen when you have a purposely incomplete standard where a certain part is written and the rest is up to the manufacturer? That's right, they're going to fuck it up. Often.

            I don't expect (any) closed hardware to work "perfectly" on Linux (or BSD) whether its certified or not. I've been around the block enough times to know that's not always going to be the case. But in the case of power management, the PC world is in general a crapshoot. What I mentioned up there in my OP was not just my experience with Linux, but my experience with Windows PCs as well. The RAM corruption issue on resume? Windows XP resume on an HP Pavilion. Laptops forgetting their user settings on resume from sleep? Windows. Sometimes this isn't even a firmware problem, it can actually be the CPU itself which has nothing to do with which OS it runs - see the C-state problems with some Intel CPUs over the years, Nehalem was one of those and no software fix was going to fix it.
            Actually it doesn't work that way. It's not about whether a piece of hardware strictly adheres to a spec or if that spec is perfect, that's a pipe dream. What it means is that if you buy a Windows-certified laptop from, say, HP, they install whichever settings and crappy proprietary blob drivers it takes for Windows to work as expected on that particular machine, however non-compliant its hardware may be, and that's all. It's the same with Linux, you get a probably broken ACPI implementation and a Linux distro with drivers, firmware blobs, patches and kludges pre-installed to make it work on that machine (and, likely, crash on another one). Nothing more, but nothing less.

            Comment


            • #7
              Solus, here I come!

              Comment


              • #8
                Originally posted by JoshuaAshton View Post
                Snaps are a terrible idea. Huge waste of space, just so some apps can use outdated libraries who don't know how to do versioned/backwards compatible interfaces correctly. What a shitshow. In my eyes it completely goes against one of the main advantages of using Linux: the package manager.

                I think I'll pass on Ubuntu and its future releases. Snap philosophy is poop.
                While snaps/flatpaks are awful there isn’t much of an alternative for Chromium, for LTS, as Google refuses to support building with older libraries. RHEL doesn’t have Chromium due to that reason.

                Comment


                • #9
                  Originally posted by JoshuaAshton View Post
                  Snaps are a terrible idea. Huge waste of space, just so some apps can use outdated libraries who don't know how to do versioned/backwards compatible interfaces correctly. What a shitshow. In my eyes it completely goes against one of the main advantages of using Linux: the package manager.

                  I think I'll pass on Ubuntu and its future releases. Snap philosophy is poop.
                  Can you expain? If I understand correctly, you're saying that because snaps can choose whether to use shared dependencies or bundle them, like rpm and deb, snap is a waste of space, but rpm and deb is not? How does this work logically?

                  By the way; are you familiar with the concept of deduplication?
                  Last edited by jo-erlend; 05-20-2018, 09:55 AM.

                  Comment


                  • #10
                    Originally posted by jo-erlend View Post

                    Can you expain? If I understand correctly, you're saying that because snaps can choose whether to use shared dependencies or bundle them, like rpm and deb, snap is a waste of space, but rpm and deb is not? How does this work logically?

                    By the way; are you familiar with the concept of deduplication?
                    I'm not sure about rpm's since there are so many different sources of them but debs generally comply with Debian Policy which requires you not randomly bundle libraries into a package. This greatly reduces any potential for duplication, snaps and flatpaks are pretty much the opposite of this, wanting to allow developers to build against any version of a library they choose. There are other non-technical issues with them, such as lack of thorough vetting, and required source only uploads, as was shown with the recent snap malware. I suspect the snap/flatpak stores will eventually end up like the Google Play store inundated with malware.

                    https://www.debian.org/doc/debian-po...-ch-sharedlibs

                    Comment

                    Working...
                    X