Hello;
On Wed, 2011-04-27 at 08:51 +0800, ext Zhao, Juan J wrote: > On Tue, 2011-04-26 at 10:19 -0400, Cory T. Tusar wrote: > > > The comparison between es2gears and glxgears in mesa-demos can be > > one in > > > my opinion. > > > > http://www.google.com/search?q=glxgears+is+not+a+benchmark > People always don't care the fps that glxgears show because it uses > display list and fixed function to draw gears and all these is discarded > by OGL in latest version. As we know the HW of graphcs is programmable > process named vertex processor and fragment processor. Fixed function > won't show it's full ability. > But now you want to compare gles and gl, the test case should show the > same contents. So I told you that you can compare these two > applications. And in fact, although glxgears uses fixed function, in the > driver it will change the protocol to HW protocols and use vertex and > fragment processor to draw and render. > So just for the comparison, these are two good examples. > Of course, you can also write some applications showing the same > contents and based on different graphics tool(I mean gl and gles) to > compare. As this thread is about configuring Clutter to use OpenGL ES 2.0, I think ideally you should use some applications/testcases written with Clutter and run those with desktop GL and GLES2.0 backends. This way you get more realistic benchmark of the whole pipeline you are intending to use, try to get more closer to your actual use case. Then as additional task one could investigate if Clutter utilizes these backends optimally or is there perhaps something to optimize. I consider these separate tasks with separate goals. For measurements and observations you could consider to use these recently released helpers from Nokia : Measuring fps : https://gitorious.org/swaplogger/ Tracing GL(ES) output of the application : https://gitorious.org/tracy // Tapani _______________________________________________ MeeGo-dev mailing list [email protected] http://lists.meego.com/listinfo/meego-dev http://wiki.meego.com/Mailing_list_guidelines
