This is the mail archive of the
java@gcc.gnu.org
mailing list for the Java project.
much smaller bytecode with gcj?
- From: Adam Megacz <adam at xwt dot org>
- To: java at gcc dot gnu dot org
- Cc: brian at xwt dot org
- Date: 28 Nov 2003 18:21:52 -0800
- Subject: much smaller bytecode with gcj?
- Organization: XWT
I've mentioned mips2java here before. Strangely, gcj -O3 emits *much*
smaller bytecode than either javac or jikes when compiling the output
from our tool. Any ideas why this might be? The difference is striking:
Perhaps I shouldn't look a gift horse in the mouth ;)
- a
______________________________________________________________________________
From: Brian Alliet <brian@brianweb.net>
Subject: Re: [core] GCJ vs. javac tests
Date: Fri, 28 Nov 2003 21:15:55 -0500
To: core@xwt.org
> Øyvind Harboe <oyvind.harboe@zylin.com> writes:
>> 2. Transmission time. Wow. DJpeg.class zipped.
>> GCJ 202k
>> javac 360k
>
> Is this for real?
Sure is. Whatever -O3 does makes a huge difference.
(uncompressed/zipped)
javac - 909k/356k
gcj -C - 977k/411k
gcj -O3 -C - 741k/198k
I'll look into it some more.
-Brian
_______________________________________________
core mailing list
core@lists.xwt.org
http://lists.xwt.org/listinfo/core
--
"Education is not filling a bucket but lighting a fire." -- WB Yeats