Test machine: Dual PowerMac G5, GeForce FX 5200, java build 1.5.0.16-
b06-284
Below is a chunk of code which demonstrates a major graphics
performance regression on this machine between Java 1.4.2 and the
latest Java 1.5. It appears that the choice of Sun renderer is
ignored. I have reason to believe this is due to recent changes in
1.5; on my (Intel MacBook), without the latest upgrade to 1.5, I've
seen no regression yet. I'd compare to the newest upgrade on the
Macbook, but it's my primary machine and I'm scared to update to
this _huge_ drop in performance.
The code draws a bunch of fillrects, then prints out the time to do
so. In 1.5.0_16, it takes about 8 seconds. In 1.4.2, it takes
about 4 seconds. I believe other primitives (circles, etc.) are
even worse. Stretching an image may be _much_ worse. No
modifications to rendering hints, properties, etc., appears to have
any effect
Can anyone else verify this on another kind of machine? I'm hoping
for some data points before tossing the bug report into Apple's bug-
reporter black hole.
<snip/>
/* /// None of this makes any difference
System.setProperty("com.apple.hwaccel","true");
System.setProperty("com.apple.graphics.EnableQ2DX","true");
System.setProperty("com.apple.graphics.EnableDeferredUpdates",
"false");
System.setProperty("com.apple.graphics.UseQuartz","true");
*/
Of course that won't make a difference, the real property is:
System.setProperty("apple.awt.graphics.UseQuartz","true");
<snip/>
I'm under the general impression that the Sun 2D graphics renderer
only because practical in Java SE 6, and the J2SE 5.0 implementation
should probably be set back to Quartz by default (since Sun isn't
going to be improving it anytime...uh...ever).
Hope this helps,
Mike Swingler
Java Runtime Engineer
Apple Inc.
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Java-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/java-dev/email@hidden
