[kaffe] initial performance comparisons between kaffe-1.1.5 and jamvm 1.3.0 on ARM

[gian paolo ciceri]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello all,
I've tested with a simple "hello world" program my arm ports
of both latest kaffe and jamvm: here what it happens

- ----
bash# time
/usr/local/kaffe-1.1.5-JIT-arm-qt-crosstool-glibc213/bin/kaffe Hello
Hello World!
24.09user 4.11system 0:28.20elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (1634major+5559minor)pagefaults 0swaps

bash# time /usr/local/jamvm-1.3.0-arm-glibc213/bin/jamvm -Xmx16M
- -classpath /ProveJava Hello
Hello World!
0.91user 0.60system 0:01.51elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (170major+411minor)pagefaults 0swaps
- ----

On my Gnu/Intel FC3 here's the same test results
- ----
[gpciceri at sandra ~]$ time /usr/local/kaffe-1.1.5/bin/kaffe Hello
Hello World!

real    0m0.281s
user    0m0.239s
sys     0m0.028s
[gpciceri at sandra ~]$ time /usr/local/jamvm-1.3.0/bin/jamvm Hello
Hello World!

real    0m0.122s
user    0m0.087s
sys     0m0.026s
- ----

Can this "speeddown" between kaffe and JamVM on Arm be justified from
the fact that (on Arm) I've had to build kaffe with --with-engine=jit
(intrp does not work, jit3 is disabled on Arm) while on inlet I've used
defaults, so jit3 ?

Any test to suggest in order to try to reduce this speeddown ?
TIA
/gp

- --
gian paolo ciceri - japanman
suddenthinks, excellence through refactoring
++39 340 7218101
gp.ciceri at acm.org
gp.ciceri at suddenthinks.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org

iD8DBQFCU6eX5CXmEF0gaHgRAkxBAKCbMaxkiseAmkd8YiOfjoCsPBiyGgCfbARR
h1SW/gqMN9HZuJsCEZ368q0=
=goaF
-----END PGP SIGNATURE-----

-- 
gian paolo ciceri - japanman
suddenthinks, excellence through refactoring
++39 340 7218101
gp.ciceri at acm.org
gp.ciceri at suddenthinks.com