> > Maybe, this is just a guess, but if there is a memory leak associated
> > with video bitmaps, that they are being reallocated somehow or some way.
> > Perhaps when the leaks occur, it occupies some memory, and the video
> > card runs out so Windows or the AGP card starts placing new regions
> > where the old ones used to be and you lose some old surfaces, but you
> > don't see corruption since the leaked surfaces aren't being written to?
>
> I don't think so, the blanking stops after some time.
Sorry, you were talking about the _leak_... that's possible. 300k is exactly
the size of the video buffer... hmm... very handy.
Neil, can you send me a recompiled version of your EXE *without*
stretch_blit()? I'll see if it leaks for me.
|