Can you give us some details about the performance tests?
Apple's Java VM for Power PC processors has different performance
characteristics than Suns Java VM for Intel processors on Windows.
For example,
a) If you are testing graphics performance with Java 6, you may want
to run your performance tests with Quartz turned off.
java -Dapple.awt.graphics.UseQuartz=false
b) If you are testing file I/O performance you may want to verify if
your performance tests use buffered input and output streams. Windows
does a lot of buffering "for free" whereas on Mac OS X, if you are
asking for unbuffered unbuffered I/O, you are getting it.
c) For computational tasks, in average you may expect a correlation
between clock speed and performance. Meaning that a P4 @ 2.8 GHz will
probably perform 2.8 / 1.2 = 2.33 times faster than a G4 @ 1.2 Ghz.
But "in average" is not true for individual operations. There are
operations at which the G4 performs better than the P4 and vice
versa. That is, if your performance test contains optimizations for a
JVM on a Pentium processor, you may want to remove these
optimizations, and start over again with profiling for the G4 processor.
To give you more useful advice, we need details about your
performance tests.
With kind regards,
Werenr
On 17.07.2006, at 18:52, Carlos Santiago wrote:
I am using Java 5 and 6 in my MAC (iBook 1.2Ghz with 768MB of
memory) with MAC OS X 10.4.7, but the performance tests that I have
made is very slow whether compared with a P4 with Windows XP
Profesional 2.8Ghz with 1GB of memory.
Somebody already perceived this?
Thank´s
Carlos
_______________________________________________
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
