On Sun, Nov 15, 2009 at 04:50:42PM +0100, Uriel wrote:
As kris said, that doesn't compare languages, but implementations, and
not just implementations of the language/compilers, but
implementations of the benchmark itself.
Rog (of Vitanuova fame) posted an update to the Go implementation of
the benchmark that improves performance by 20%, and he said he had
barely spent half an hour to get that far.
Also when running the benchmark they don't seem to have set
GOMAXPROCS, so they failed to get any parallelism.
It's also worth noting that several of the GCC benchmarks make
use of pthreads. The Go implementations are (amazingly) not
parallelized, though they certainly should be given the design
of the language.
--
Kris Maglione
The world is a tragedy to those who feel, but a comedy to those who
think.
--Horace Walpole
