[Shootout-list] Java benchmarks
Isaac Gouy
igouy2@yahoo.com
Wed, 30 Mar 2005 07:51:58 -0800 (PST)
> The central problem here is that the shootout benchmarks
> are very small and short.
If that's the problem then let's identify which benchmarks should be
run for a longer period.
Most of the benchmarks are new, and use an arbitrary data-range. Now
that we have a bunch of implementations in different languages we can
adjust the data-ranges for the benchmarks.
For example, takfp seems OK to me, fasta should probably be 2x 3x
longer, nsieve seems OK, etc.
Java -server doesn't necessarily take long to catch -client
http://www.lissett.com/ben/
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com