On May 8, 2008, at 8:44 AM, Dario Accornero wrote:
Hello,
In my AGL windowed application I am enabling the kCGLCEMPEngine flag
when running on machines with more than one CPU. This effectively
improves performance, but... (live-)resizing the window results in a
few visual glitches. It looks as though VRAM is being paged or
shuffled around for some milliseconds: I can't manage to obtain a
screenshot with the garbled output, but overall the impression is
that the paging is not in sync with the context resizing (which
calls aglUpdateContext as per the docs).
The MP Engine doesn't change how VRAM pages are managed but it does
buffer instructions differently (more to the point it simply does more
instruction buffering) than the serial engine.
Try flushing the context glFlush/glFinish in your reshape/resize
method. This will empty the additional instruction buffer the MP
Engine maintains.
If this doesn't work, please file a bug.
If you're using multiple or shared contexts, this is an important
detail.
If I turn off the kCGLCEMPEngine flag, (live-)resizing is always
smooth, no matter how large the window is or what it displays.
Any ideas about this? I'm in 10.5.2 on a MacBook Pro 2.4GHz w/8600M
GT 256MB.
Thank you,
Dario
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Mac-opengl mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/mac-opengl/email@hidden
This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Mac-opengl mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/mac-opengl/email@hidden
