On Sat, 2006-06-10 at 10:11 +0200, Roman Kennke wrote: > Hi again, > > Am Samstag, den 10.06.2006, 10:02 +0200 schrieb Roman Kennke: > > This enables the use of VolatileImages as backbuffer for Swing. This > > effectively solves the performance problems brought by the Java2D > > rewrite. > > Well, actually this seemed to be wishful thinking. Comparison in the > FillRect demo shows that painting is comparable for the 'standard' > benchmark, and even slower on VolatileImage with both checkboxes checked > (on my machine ~1000ms for VI vs ~500ms for BI). I guess we need to fix > the VolatileImage impl.
Forget about the fillRect demo. It's a terrible benchmark. The values it gives do not actually provide any meaningful measure of drawing speed. /Sven