Announcement

Collapse
No announcement yet.

Image Quality: Intel Ivy Bridge vs. Radeon Gallium3D

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

  • Image Quality: Intel Ivy Bridge vs. Radeon Gallium3D

    Phoronix: Image Quality: Intel Ivy Bridge vs. Radeon Gallium3D

    Aside from all the Linux gaming news this week, the past few days were also particularly exciting due to Intel's launch of the much-anticipated Ivy Bridge processor line-up. There were launch-day Linux benchmarks of the Intel Core i7 3770K on Phoronix plus many more articles are currently in the publishing pipeline...

    http://www.phoronix.com/vr.php?view=MTA5MzU

  • #2
    Seriously what is the point?

    Comment


    • #3
      Originally posted by FireBurn View Post
      Seriously what is the point?
      I agree comparing image quality is something that's not very interesting. But I'd love to see opengl performance comparisons for windows 7 and linux. I know that previous Intel drivers were significantly better on windows than on linux, but now times have changed, so a new comparison is something that we should definitely see.

      And I'm sure other people will also be interested in other articles.

      Comment


      • #4
        Originally posted by Anarchy View Post
        I agree comparing image quality is something that's not very interesting. But I'd love to see opengl performance comparisons for windows 7 and linux. I know that previous Intel drivers were significantly better on windows than on linux, but now times have changed, so a new comparison is something that we should definitely see.

        And I'm sure other people will also be interested in other articles.
        The Win comparison is already mentioned in the articles...

        - A comparison of the Sandy Bridge and Ivy Bridge OpenGL graphics performance under Microsoft Windows 7 and Ubuntu 12.04 LTS Linux.
        Michael Larabel
        http://www.michaellarabel.com/

        Comment


        • #5
          1) Render
          2) Save the rendering in a lossless format
          3) Use a image diff'ing tool

          Only if you really have a point to make about rendering differences product of something like a bug or overall testing for regressions for example . Of course, the comparison is legit as long as you take care that the renderings are THE SAME (same shaders, textures, and such) ... maybe apitrace could help with that?
          Otherwise than that, this entire article is simply a waste of time for readers (if no point is made)

          Regards.
          Last edited by vertexSymphony; 04-28-2012, 09:22 PM.

          Comment


          • #6
            Originally posted by vertexSymphony View Post
            1) Render
            2) Save the rendering in a lossless format
            3) Use a image diff'ing tool

            Only if you really have a point to make about rendering differences product of something like a bug or overall testing for regressions for example . Of course, the comparison is legit as long as you take care that the renderings are THE SAME (same shaders, textures, and such) ... maybe apitrace could help with that?
            Otherwise than that, this entire article is simply a waste of time for readers (if no point is made)

            Regards.
            It's already doing that and you can look at a image diff by running the result through phoronix-test-suite analyze-image-delta, etc.
            Michael Larabel
            http://www.michaellarabel.com/

            Comment


            • #7
              Originally posted by Michael View Post
              It's already doing that and you can look at a image diff by running the result through phoronix-test-suite analyze-image-delta, etc.
              Sorry if I may sound harsh, but I am really confused.
              What do you mean by that?
              We have to have that hardware to generate an image diff? Or do we have do download the images you generated and generate a diff?
              Wouldn't it make more sense to have a diff percentage between the two images and some kind of mechanism (e.g. on-mouse-over) to switch between the two images on the same canvas to actually be able to see the differences?

              Comment


              • #8
                Originally posted by Michael View Post
                It's already doing that and you can look at a image diff by running the result through phoronix-test-suite analyze-image-delta, etc.
                http://openbenchmarking.org/result/1...SU-NEXUIZGRA95

                I only see the renderings, no image diffs ... and if I have to buy the cards **and/or** run the diff by myself, what's the point of the article then? sorry if it's kinda harsh, but the article could be better
                --

                Firefox allows to download base64 encoded images, so ... I just tried different ways to diff images using compare and I have to say that the renderings are *very* similar, dunno what's the point of the article, I'm confused
                In any case, it's a good idea to run a rendering diff from time to time just to check if there are any regressions and where it appeared by running a bisect. But post an article if there's something relevant to say about it ( if diff'ing really shows a perceptible difference)

                Regards.

                P.S → here's how to make a diff of images (if you are curious, there are *tons* of ways to make diffs) : http://www.imagemagick.org/Usage/compare/

                Comment


                • #9
                  What I mean is that you can already do image diffs with PTS, etc by just running one command, etc. It's very streamlined from what's in the article and anything else can be easily done from there with small tweaks or other commands from there for those interested. This was just a quick weekend article pushing out the data for anyone interested in exploring or playing further with it.
                  Michael Larabel
                  http://www.michaellarabel.com/

                  Comment


                  • #10
                    You should really do the comparison yourself. Maybe even something like Tom's do, here's a sample if you're not aware:

                    http://www.tomshardware.com/reviews/...cs,3173-3.html

                    It's really kind of silly to just see 6 images like that, and be like "oh, so what is the point"?

                    Comment

                    Working...
                    X