Page faults and libgcj.so startup time

[Jonathan P. Olson]

I'm really glad that this topic has now surfaced as a performance problem
in Linux applications as well.  From personal experience with GCJ since
the compiler was first released back in Sep-1998, I've found that Java 
metadata
accounts for more than half of a binary size.  Worse yet, all the 
relocations
in this metadata force a trememdous number of run-time fixups.  In 
embedded
applications this was a real concern because of the memory footprint of
Java applications vs. their C++ counterparts.

Once GCJ 3.0 has stabilized I really do plan to revisit this whole 
serialized
compressed metadata issue if nobody else does in the meantime.

On Friday, April 13, 2001, at 01:03  AM, Jeff Sturm wrote:

>
> I think it was Jon Olson who first suggested compressed metadata.  I
> really like that idea both for shrinking the image footprint and
> eliminating most load-time fixups.  I'd prefer however deferring the
> "uncompress" action to class initialization rather than the first use of
> reflection.
>
> http://gcc.gnu.org/ml/java/2000-06/msg00086.html
>