Gcj binaries performance (arm-linux)

Fri May 9 09:11:00 GMT 2008

2008年5月9日 Andrew Haley <aph@redhat.com>:
> Andrew John Hughes wrote:
> > 2008年4月29日 ffileppo <ffileppo@libero.it>:
> >> >
> >> > aph@debian:~$ /home/aph/gcc-4.3.0-RC-20080301-install/bin/gcj MyPanel.class -O --main=MyPanel
> >> > aph@debian:~$ ./a.out
> >> > Test "Bubble Sort" (Num loop: 100)
> >> > First loop time: 22
> >> > Total time elapsed: 2197
> >> >
> >> > aph@debian:~$ /home/aph/gcc-4.3.0-RC-20080301-install/bin/gcj MyPanel.class -O2 --main=MyPanel
> >> > aph@debian:~$ ./a.out
> >> > Test "Bubble Sort" (Num loop: 100)
> >> > First loop time: 20
> >> > Total time elapsed: 1943
>> >>
>> >> Thanks for your help Andrew,
> >> I'm getting good performance on target using -O / -O2
>> >
>> > So does gcj default to no optimisation? This is what these results
> > suggest, as the -O version is about 18 times quicker.
>> Yes, of course. That's standard for UNIX, as far as I'm aware: I don't
> think I've ever come across a compiler that optimized unless it was
> asked to.
>> Andrew.
>
That makes sense, especially as something like -O2 will make debugging harder
and probably has effect on some of the reflection properties of Java.
I guess this is what you get from Java people who tend to expect
everything to work
best with no options... ;) I always pass plenty when hacking C or C++ code.
-- 
Andrew :-)
Support Free Java!
Contribute to GNU Classpath and the OpenJDK
http://www.gnu.org/software/classpath
http://openjdk.java.net
PGP Key: 94EFD9D8 (http://subkeys.pgp.net)
Fingerprint: F8EF F1EA 401E 2E60 15FA 7927 142C 2591 94EF D9D8