I am using the samples in the recently released GPU SDK and am having trouble getting the hardware acceleration working under X11.
Just to be clear, I am not talking about using hardware to accelerate the X11 windowing system itself, rather I am trying to get accelerated 3D to work in a window. I am using Open GL ES 2.0 (although using ES 1.1 seems to have the same issue).
The samples come with two makefiles, one for X11 and one for frame buffer device. I have built one of the samples (Samples/GLES2.0/06_Texturing) for frame buffer device, and performance is terrific (over 500 fps full screen 1024x768). If I build the same sample for X11, it does work and it does render in a window, but the performance is very poor (about 15 fps in a large window which is almost full screen 1024x768). The X11 sample also uses considerable CPU so I am pretty sure it is using software rendering / rasterisation.
So my question is this - how can I get decent frame rates using hardware acceleration for an OpenGL application running in an X11 window?
Obviously I have to change over the symbolic links from libGAL-fb.so and libEGL-fb.so to libGAL-x11.so and libEGL-x11.so
The sample won't even build properly if I don't do that, so it must be something else that I'm not doing (or doing wrong).
Some background info which may be useful:
- i.MX6Q (rev 1.2 silicon)
- Linux kernel 3.0.35 from the 12.09.01 BSP
- Ubuntu 11.10 rootfs (armel)
- GPU SDK v1.00
- gpu-viv-bin-mx6q-12.09.01 (untarred on top of the rootfs)
Have any of you got this working? I'd be most grateful for any comments / suggestions.
I've just worked out that this is addressed in the new Linux BSP release L3.0.35 1.1.0 - see https://community.freescale.com/message/312735#312735
Message was edited by: davecbluechip