On Wed, 12 Jun 2002, Denis Oliver Kropp wrote: <snip>
> It's not a bug, it's just a stupid heuristic. Imagine the following sequence: > > Action | S | V | > --------------------------------------------- > Surface created | + | . | > Software writes to surface | + | . | writes to system memory > Software stretch blits from surface | + | . | reads from system memory > Hardware blits from surface | + | + | copied to video memory > Software writes to surface | - | + | writes to video memory > Software stretch blits from surface | - | + | reads from video memory > > S = system memory instance > V = video memory instance > . = invalid (not allocated) > + = stored (allocated, contains recent data) > - = restore (allocated, but data is outdated) i understand now, is there anyway to fix this ? is it possible to copy the data back to system memory when the software stretchblit detects that the surface is in video memory ? thank you very much for your help and detailed explainations. cheers, alex -- Info: To unsubscribe send a mail to [EMAIL PROTECTED] with "unsubscribe directfb-dev" as subject.
