Re: penalty for double buffering

[Russell Shaw <rjshaw iprimus com au>]

Steven Jenkins wrote:
> Michael Torrie wrote:
>
> > On Sat, 2003-12-13 at 13:43, Steven Jenkins wrote:
> > It seems to me that since you're drawing to the screen 50 times a second
> > that double buffering would buy you nothing anyway.  Double buffering is
> > mainly to eliminate flickering when the widget redraws or changes 
> > size. Since you're drawing yourself, with double buffering on, it 
> > seems to me
> > that you'd be drawing to the buffer and then to the screen.  Since
> > you're taking it upon yourself to handle all the invalidate events, such
> > a middle buffer isn't necessary (and won't do anything for you).  So
> > turning off the buffering returns you to the speeds you were used to
> > under gtk 2.0, and I don't think there would be much flickering.
>
>
> The rate at which the display is updated is determined by the user. The 
> application is a real-time data acquisition package with an interpreter 
> for a small programming language. The graphics primitives are there if 
> the user wants to draw plots, etc. 50 Hz is the likely uppper limit of 
> user requirements, but higher and lower rates are allowed. Sometimes the 
> users only update every few seconds.
>
> But you may be right. I don't notice any flicker in the testing I've 
> done so far. The widget can't be resized, so maybe I don't need the 
> double buffering anyway. I'd like to be sure before I remove it.
>
> Even if I don't need it, I'm still curious why it's so costly.

Make a timer with timer=g_timer_new(), then put things like t0=g_timer_elapsed(timer)
everywhere in the code, then have: g_printf("Time: %f %f %f\n",t3-t2,t2-t1,t1-t0)
in the 50Hz loop somewhere.