This is the mail archive of the
java-discuss@sourceware.cygnus.com
mailing list for the Java project.
Re: gcj performance: benchmarks
- To: java-discuss@sourceware.cygnus.com
- Subject: Re: gcj performance: benchmarks
- From: Per Bothner <per@bothner.com>
- Date: 16 Aug 1999 22:07:30 -0700
- References: <37B8E6F0.4403E42D@albatross.co.nz>
Bryce McKinlay <bryce@albatross.co.nz> writes:
> Some interesting but slightly disappointing numbers.
> Conclusions: ...
(3) There is a lot of room for profiling and tuning in Gcj/Gij.
We have mostly concentrated on functionality, but we know many
things we could improve. Some profiling might help point out
the hot spots.
Btw - note also that Gcj can do noticably better when compiling
from source than when compiling from bytecode. For example you
get much better loop optimizations from source. The Gcc back-end
people think it should not be very difficult to "recover" the
loops from the bytecode, but for now I recommend compiling
from source any benchmarks (using a de-compiler if need be) - at
least any that we want public!
--
--Per Bothner
bothner@pacbell.net per@bothner.com http://home.pacbell.net/bothner/