Re: GCJ Native Proposal

[Tom Tromey <tromey@redhat.com>]

>>>>> "Barry" == Barry Hawkins <barry@alltc.com> writes:

Barry> It must be underscored that this is a _potential_ speed gain.  We
Barry> actually are not certain what speed gains, if any, we will have.

There's a fair amount of benchmarking data out there:

http://klomp.org/mark/free-vm-benchmarks/
http://www.shudo.net/jit/perf/
http://www.spindazzle.org/benchmarks/

Of course whether these benefits materialize in something like eclipse
can't be determined from those pages.

Barry>A third option I would also posit is:
Barry> c) Take a deep, hard look at what is broken with JIT in the free
Barry> runtimes right now and how we could significantly contribute to that
Barry> part of free java that really needs help.

The way I think about gcj these days is that it is a caching JIT with
a somewhat odd procedure for deciding what to compile :-)

Tom