Announcement

Collapse
No announcement yet.

2009 Linux Graphics Survey Results

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

  • 2009 Linux Graphics Survey Results

    Phoronix: 2009 Linux Graphics Survey Results

    For the month of November we ran the 2009 Linux Graphics Survey, which is a survey in regards to X.Org and the Linux graphics stack that we have been hosting annually for the past three years. This year there was 13,836 results submitted and we have now had the time to go over these results and are publishing all of the numbers today.

    http://www.phoronix.com/vr.php?view=14429

  • #2
    The first thing that I notice is the percentage of VESA users almost equals the percentage of VIA users.

    Comment


    • #3
      Looking at the distributions of various things, I was wondering: next year, why not run the questions by the community before posting them. Do some brainstorming and get some idea of what questions may be problematic, or hard to contextualise.

      For example, brand is pretty easy but driver is not so much. What hardware are they using? R100-R200? R300-R500? That sort of thing. How many of those nVidia users are on the legacy drivers? Which legacy driver?

      Also, there need to be more "none of the above" choices. How many people aren't using any of the X.org display features? Because everyone can select multiple things, there's a loss of context.

      The seventh question could probably use an option for "autodetection". That's more or less what I use on any machine I set up these days, and the closest option is "manually editing xorg.conf"-- not quite the same. The option conflates two types of users. (This isn't to say that I won't bust out vim on the thing myself on occasion; I'm well familiar with that file and its simple syntax. X.org is just getting better and better about not needing to do so.)

      The desktop resolution question seems prone to misleading data. There is a proponderance of annoyingly low-resolution 1280x800 screens on many laptops, thus skewing the low-range and not allowing us a clear picture of who is running older hardware versus newer devices like nettops. And it doesn't account for common, but weird-ass things like 1680x1050: where does that go?

      The key-interests question would probably be best served with a "choose the top three items"-type question and get numbers that way.

      The "Active Linux Tasks" could use "other" or "none of the above" options, so as to get a better idea of how many people are actually working. The last question has similar issues, in my opinion. What's a "mainstream user"? Are developers really something you could call "mainstream"? I don't feel it is; not many people know what programming entails at all.

      Still, the numbers we've got are the numbers we've got, and they're still decently interesting. Thanks again, Michael.

      Comment


      • #4
        What I found interesting was if you add up the users of the various ATI drivers, and add up the users of the various nvidia drivers, the numbers are very very close.

        Given the difference in uses of ATI cards versus nvidia, it would seem to suggest that most of the VESA driver users would have nvidia cards.

        Comment


        • #5
          Well some use both (or even all 3) drivers for ATI depending on purpose. So summing up is not always the correct way.

          Comment


          • #6
            Originally posted by Wyatt View Post
            Looking at the distributions of various things, I was wondering: next year, why not run the questions by the community before posting them. Do some brainstorming and get some idea of what questions may be problematic, or hard to contextualise.
            There was in late October on the forums and an announcement on the main site too I believe...
            Michael Larabel
            http://www.michaellarabel.com/

            Comment


            • #7
              Originally posted by Michael View Post
              There was in late October on the forums and an announcement on the main site too I believe...
              Then my apologies; I missed it completely. (Was also in the middle of a move, so things were rather hectic).

              Comment


              • #8
                I'd love to use TV-Out... but don't have time to program it.

                Comment


                • #9
                  Suggestion for next year

                  Michael,

                  I have two suggestions for next year's survey. First, make it clear at the top of the forum that you should fill out the form once per computer you have (I filled it out taking into account my three computers, and when I got to the end it said "now fill it out again if you have another computer").

                  The second suggestion is similar to what Wyatt proposed. You should ask not only what (or how old) hardware they have (which makes a big difference for ATI users, I'm among the users who are stuck with the open source ATI driver after ATI dropped support), but a question getting at whether the user bought new hardware during the past year (or what hardware they would buy in the future), or if they recently switched drivers, etc. These types of questions would help determine whether people are switching in droves to (or from) any one vendor due to a change in their policies (again, I would have a hard time buying another ATI card after they decided to drop support for my less than 3 year old, still under warranty laptop).

                  - Ben

                  Comment


                  • #10
                    Wish they would fix the All-In-Wonder parts so the 9800's could operate independently of the driver.

                    Comment


                    • #11
                      It's rather sad to see that in 2009/2010 that so many users continue to modify their configuration files by hand that there aren't graphical or CLI utilities available (or that they are aware of) that can suit their configuration needs. If Linux is to become a mainstream operating system with consumers, they must not need to be introduced to editing the xorg.conf themselves.
                      I couldn't disagree more. The number of people configuring xorg.conf manually doesn't indicate either lack of a good graphical utility or a poor user experience. This survey was for the most part answered by enthusiasts who are comfortable editing /etc and do it that way because they want to, not because they have to. I for one would continue to maintain an xorg.conf manually no matter what graphical utilities are available, even though X normally auto-detects my hardware and starts up just fine without even any xorg.conf present.

                      I use krandrtray for simple resolution adjustments, the CLI xrandr utility to adjust multi-display settings, and xorg.conf for advanced driver and hardware options. I generally use Xorg -configure to build a base config file and then tweak a few things from there.

                      It really doesn't have anything to do with bad front-ends and certainly doesn't degrade the user experience. A new user coming to Linux doesn't need to know anything about xorg.conf.
                      Last edited by Smorg; 12-09-2009, 06:13 AM.

                      Comment


                      • #12
                        Originally posted by Smorg View Post
                        It really doesn't have anything to do with bad front-ends and certainly doesn't degrade the user experience. A new user coming to Linux doesn't need to know anything about xorg.conf.
                        I can agree that the survey question didn't help clarify if the GUI tools were adequate, but if people find it easier to go to the conf files, then that is disturbing.

                        When you need help with something in Linux - the knowledgeable users like yourself send the inexperienced user to the CLI, have them edit conf files, or worse ripping out whole parts of the system and installing different packages. It's more likely to scare a user away than "educate them", and certainly (beyond reading tonnes of man pages), makes the system relatively 'un-discoverable' (easy to learn once shown once)

                        And one's comfort in using multiple tools/user interfaces to modify a system has created an inconsistency in the user experience. I find most GUI tools come up shy of complete due to the acceptance that one can always go to the CLI.

                        If things work perfectly, it's a nice little system; otherwise things get unfriendly pretty quickly. (I've been inclined to leave things broken out of lack of desire of getting into "linux" style fixes)

                        Comment


                        • #13
                          I configure my resolution and refresh rate from the KDE system settings. Anything else should be left to the distro makers in having working drivers by default.

                          Nobody should have to touch the inner workings of a Linux based OS, but they should be able to.

                          Comment


                          • #14
                            Originally posted by Craig73 View Post
                            I can agree that the survey question didn't help clarify if the GUI tools were adequate, but if people find it easier to go to the conf files, then that is disturbing.

                            When you need help with something in Linux - the knowledgeable users like yourself send the inexperienced user to the CLI, have them edit conf files, or worse ripping out whole parts of the system and installing different packages. It's more likely to scare a user away than "educate them", and certainly (beyond reading tonnes of man pages), makes the system relatively 'un-discoverable' (easy to learn once shown once)

                            And one's comfort in using multiple tools/user interfaces to modify a system has created an inconsistency in the user experience. I find most GUI tools come up shy of complete due to the acceptance that one can always go to the CLI.

                            If things work perfectly, it's a nice little system; otherwise things get unfriendly pretty quickly. (I've been inclined to leave things broken out of lack of desire of getting into "linux" style fixes)
                            I guess I don't think config files are intrinsically a bad thing. Maybe its more like graphical tools for administration are generally harder because they aren't universal and you have to learn/create a new one with every system you want to configure. They aren't expandable or scriptable either - which violates the unix principle of everything-as-a-file, and the interoperability that comes along with programs working with stdio.

                            Unfortunately I don't see a good way to remedy this. Linux graphics aren't one size fits all. Most GPU manufacturers have multiple choices when it comes to drivers and there's so many more considerations that go into Xorg, window managers, desktop environments, different distros, etc etc. Having lots of choice is a good thing which comes at the cost of having homogeneous interfaces; and the only real homogeneous interface that makes sense is config files, man pages, and the CLI.

                            I'm all for making Linux more digestible for beginners so long as we don't have to sacrifice choice or power. Then, I use Xmonad and Gentoo on my primary workstation. If you think xorg.conf is bad... full Haskell configuration is some serious power, yet I suppose a bit inaccessible for many.
                            Last edited by Smorg; 12-09-2009, 03:50 PM.

                            Comment


                            • #15
                              Originally posted by V!NCENT View Post
                              The first thing that I notice is the percentage of VESA users almost equals the percentage of VIA users.
                              I suggest that for some people, VESA is the only driver that they know about so that's what they wrote on the survey. My knowledge of graphics is somewhere between the "Vesa" users and yours. I would guess that some of my responses were wrong. (X-server?)

                              Next year, you might add a little note below each question: (To see which X-server you are using, do this ...) You will get better data and you will help educate some users

                              Comment

                              Working...
                              X