This is the mail archive of the
java@gcc.gnu.org
mailing list for the Java project.
Re: .class code optimization
- From: Adam Megacz <gcj at lists dot megacz dot com>
- To: Lapo Luchini <lapo at lapo dot it>
- Cc: java at gcc dot gnu dot org
- Date: 11 Oct 2002 07:51:06 -0700
- Subject: Re: .class code optimization
- Organization: Myself
- References: <3DA6DCF4.6030507@lapo.it>
Lapo Luchini <lapo@lapo.it> writes:
> Is there any specific reason .class produced by GCJ are no better
> (well, only slightly better in fact) that javac ones?
I don't think so.
> Shuoldn't dead code elimination and other optimization come
> automatical using GCC?
No, I was told that the bytecode backend isn't a "real" backend like
all the others; you also can't compile C++ code using the bytecode
backend as a target.
There is a program floating around the 'net called 'javago' that does
a class->class inlining transformation. It causes *huge* speedups
for code running in the MSJVM (which does not do runtime inlining),
but doesn't noticably improve HotSpot performance.
- a
--
"Through your rags I see your vanity" -- Socrates