What sort of thread are we talking about below? A cooperative (thread
manager) thread, an MP services (preemptive) thread or a raw pthread?
Or does it not matter?
cheers,
os.
On 16 Sep 2004, at 02:08, Shaun Wexler wrote:
On Sep 15, 2004, at 1:34 PM, Os wrote:
I'm writing a Carbon app, using agl for OpenGL interfacing. I'm using
AGL_SWAP_INTERVAL to lock the buffer swaps to the screen refresh
rate.
My app will do OpenGL drawing in response to Carbon events. If I
receive events more frequently than the refresh rate, how I can I
know this and avoid redundant screen redraws (that will just stall
until the next vblank)? Is there a way to know that e.g.
aglSwapBuffers() is still pending?
If you're calling AGK to perform your drawing on the main thread, the
whole app will block while a VBL-sync'ed swap is pending, so it's all
automatic... your app won't run any faster than the graphics allows.
Similarly, is there a mechanism for knowing if a vblank is still a
long way off? E.g. if one received an event, one might not want to
call the drawing code immediately if you knew that you were likely to
get another event in before the next refresh.
Perform all of your drawing in another thread, and use (atomic) flags
as semaphores to coordinate them. When new data arrives, test & set a
"newData" flag; if FALSE, signal the drawing thread. In your drawing
thread (which can block), when it returns from the flushSwap, test for
newData; sleep on a semaphore or condition if FALSE, otherwise redraw
again. Simple? ;)
--
Shaun Wexler
MacFOH http://www.macfoh.com
_______________________________________________
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
