Announcement

Collapse
No announcement yet.

Visualizing Linux Performance Data In New Ways

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

  • Visualizing Linux Performance Data In New Ways

    Phoronix: Visualizing Linux Performance Data In New Ways

    One of the items I've been working on recently for Phoronix Test Suite 3.4-Lillesand is new ways to visualize performance result data generated by the many test profiles and suites available via OpenBenchmarking.org. Here's one of the new ways that was committed over the weekend to the Lillesand Git code-base...

    http://www.phoronix.com/vr.php?view=OTY1Mw

  • #2

    Srsly?

    I mean, please again improve the color choosing algorithm.

    Comment


    • #3
      Originally posted by ChrisXY View Post

      Srsly?

      I mean, please again improve the color choosing algorithm.
      Patches always welcome too...
      Michael Larabel
      http://www.michaellarabel.com/

      Comment


      • #4
        Originally posted by ChrisXY View Post

        Srsly?

        I mean, please again improve the color choosing algorithm.
        only 2 of them really problematic. the first and the second really optical the same.

        only the third are a little different.

        Comment


        • #5
          Yes, the colors are idiotic...

          if you want better color schemes use color brewer (http://colorbrewer2.org/). Also, some of the charts may look nice (the "half moonish" thing), but it makes comparing the first and the last bar very hard.

          Comment


          • #6
            Originally posted by ChrisXY View Post

            Srsly?

            I mean, please again improve the color choosing algorithm.
            That happens if you try to use 'corporate colors' for scientific(ish) graphs.

            There's no algorithm needed, just hardcode black, red, blue, yellow, green, brown, magenta and a few others (say the first 10 of gnuplot).

            Comment


            • #7
              Originally posted by not.sure View Post
              There's no algorithm needed, just hardcode black, red, blue, yellow, green, brown, magenta and a few others (say the first 10 of gnuplot).
              It's meant to be extensible... Not hard-coding any colors at all. The base colors and sizes and all other formatting is configurable by the user in an XML file.
              Michael Larabel
              http://www.michaellarabel.com/

              Comment


              • #8


                What is with the multiple highlighting of "best scores" on the same benchmark? There should be only one "best score" per benchmark test that is highlighted.

                Comment


                • #9
                  Originally posted by deanjo View Post

                  What is with the multiple highlighting of "best scores" on the same benchmark? There should be only one "best score" per benchmark test that is highlighted.
                  It's highlighting the worst and best, but due to an exhausted color pallete for that particular file, it ended up using the same color for the worst and best.
                  Michael Larabel
                  http://www.michaellarabel.com/

                  Comment


                  • #10
                    For that particular function on that table I would just hardcode the the colour selection to eliminate that happening since you should only have 1 best and 1 worst results per test. And maybe bold those two scores at the same time.

                    Comment


                    • #11
                      Are you familiar with Tufte? It would be well worth your time to read his book(s) if you haven't.
                      http://www.edwardtufte.com/tufte/books_vdqi

                      Comment


                      • #12
                        Originally posted by Michael View Post
                        It's highlighting the worst and best, but due to an exhausted color pallete for that particular file, it ended up using the same color for the worst and best.
                        Green or Blue for the best and Red for the worst?

                        Comment


                        • #13
                          Originally posted by FireBurn View Post
                          Green or Blue for the best and Red for the worst?
                          Right, it's usually green for best.
                          Michael Larabel
                          http://www.michaellarabel.com/

                          Comment


                          • #14
                            I don't actually understand what I'm looking at with the arc display. It's not immediately apparent what's being compared and which is better than which. Seems like a normal scatter plot would be better.

                            Comment


                            • #15
                              Originally posted by Qaridarium View Post
                              only 2 of them really problematic. the first and the second really optical the same.

                              only the third are a little different.
                              Depends on the screen...

                              Comment

                              Working...
                              X