OpenGL – OS X Mountain Lion vs OS X Mavericks

During Apple’s WWDC 2013 keynote, the graphically-inclined may have noted “OpenGL 4″ among OS X Mavericks’ feature list. I was curious and decided to study OpenGL support between Mountain Lion and the developer preview of Mavericks. The  comparison was done on my non-production workstation using OS X Mountain Lion (10.8.4) and OS X Mavericks (10.9 DP1). I utilized OpenGL Extensions Viewer 4.0.8 (App Store version) for the following assessment.

First, I looked at the OpenGL core feature support (Core OpenGL profile, SGI database).

OS X Mountain Lion (10.8.4)

OS X Mavericks (10.9 DP1)

One can see that while Mountain Lion didn’t even completely support OpenGL 3.3, Mavericks has support through OpenGL 4.1. :-)

Next, I did a compatibility test for OpenGL 1.1 – 2.1 using the default settings (cube, default frame buffer, etc.) The results speak for themselves.

OS X Mountain Lion (10.8.4)

OS X Mavericks (10.9 DP1)

While I completely understand that this is a crude assessment and the aforementioned results could easily change (for better or worse) in subsequent OSX 10.9 builds, it’s nice to see more modern graphics libraries being incorporated into OS X.

Legal

My posts are not to serve as a replacement for recommendations provided by licensed physicians nor do they represent the opinions of Baylor College of Medicine or its affiliated institutions. Please read this site's Disclaimer and Terms of Use for more information.

Comments

  • staraffinity

    Cool. Still wonder why they didn’t go all the way to 4.3.

  • JohnAlan

    Cool comparison – fire up a game and do a comparison?

    • didito
    • Fadzlan

      Well, this is not about performance, just compatibility.

      Without compatibility, there is no way one could start performance testing on those areas.

      Of course, if you want to test speed, we might get a hint right now by running it, but since this is not release, certain things might be better or worse during the final release then.

      I’d prefer to just wait on that one.

      • http://rk.md/ Rishi

        Absolutely – this was intended purely to show compatibility as performance will vary based on GPU, build number, etc.

  • Sven

    Aren’t you violating the NDA with this article?

    • http://rk.md/ Rishi

      Thanks for the comment, Sven! I’d be more than happy to take down the article if others feel this is a violation of the NDA. I reviewed my developer account’s terms, and felt that since OpenGL 4 was announced at the WWDC 2013 keynote, end users should be aware of just to what extent OpenGL has been updated. I tried to stray from any confidential “surprises” by limiting the scope of my snapshots to the benchmarking utliity. If anything, this just proves that Mavericks *is* a great update which general users should anticipate this fall.

      Please upvote Sven’s comment if you feel this article should be removed.

      • desi daru

        Very poor logic. par for the course for being a developer.

        The Seed is in NDA. You took the benchmarking information
        from the OS not from the Keynote.
        I am pretty sure benchmarking and publishing is prohibited.
        No Grey area.

        Don’t worry Apple is less strict these day.
        In the past you would have gotten an Cease and Desist Letter

        from Apple Lawyers but these days they are busy with Patent stuff.

  • Sherwin Noorian

    Imagine my surprise as I casually follow a front-page HN link and realize halfway through the post that it’s my old friend Rishi’s blog… haha! Glad to see your interest in tech is still strong even during the hard work of residency. And thanks for the relevant post, too, as I just switched to Mac last week after 20+ years solely on Windows (2013 MBA). Hope you’ve been well, man!

  • tipoo2

    I really want to see testing done in native mac games vs Windows 8. I saw some tests long ago on Anandtech with I think Snow Leopard, and it made GPUs perform like they were a generation older than they are compared to Windows, I wonder if that’s fixed.

    Also, despite the huge numbers in absolute FPS remember that this, scaled, is like going from 60 to 62FPS.

  • jean

    Hi, what GPU did you test?

    • http://rk.md/ Rishi

      Geforce 660Ti

      • jean

        Interesting. On the geforce 9400M and 9600M GT, Mavericks is quite a bit slower at all GLview tests.

        • jean

          and to continue on this, I get better results than yours on my Mac Pro (2.66 xeon 5150, radeon 4870), which should be much slower than your hackintosh: 640 fps on average in the default windowed mode, 3700 fps on average in fullscreen mode (1680*1050). BTW, the huge performance increase in fullscreen despite the higher resolution makes me wonder about the usefulness of this tool as a benchmark.

      • Jean

        The GLView cube test really doesn’t mean much as the Unigine Heaven benchmark gives totally different results (where 10.9 is actually faster than 10.8 on my system). You should try that benchmark. It is much closer to a real-world situation than a rotating cube.

  • breiti

    Are the values the same with DP2 of Mavericks?