gc_boehm_2 branch stats

Jonathan Leto jaleto at gmail.com
Sun Feb 14 20:36:21 UTC 2010


Howdy,

This looks like great progress, but individually run benchmarks are
often grossly inaccurate. You may want to use euler_bench [0] to do
microbenchmarks. You can find a description of how to use it here [1].
And I will of course help if there issues getting it setup. I would
really like to write a web-based system for asking about benchmarks
against branches/releases and even patches, but I have not had the
time to set that up.

Duke

[0] http://github.com/notbenh/euler_bench
[1] http://www.nntp.perl.org/group/perl.perl5.porters/2009/10/msg152921.html

On Fri, Feb 12, 2010 at 11:48 PM, Vasily Chekalkin <bacek at bacek.com> wrote:
> Hello.
>
> Here I declare vict^W good start.
>
> bacek at icering:~/src/parrot$ time ./parrot --gc ms ext/nqp-rx/nqp-rx.pbc
> fib2.nqp
> Got grabage 1000000
> fib(28) = 317811
>
> real    0m12.887s
> user    0m12.333s
> sys     0m0.528s
>
> bacek at icering:~/src/parrot$ time ./parrot --gc boehm ext/nqp-rx/nqp-rx.pbc
> fib2.nqp
> Got grabage 1000000
> fib(28) = 317811
>
> real    0m9.509s
> user    0m9.445s
> sys     0m0.048s
>
>
> fib2.nqp is
>
> sub fib($n) {
>    ($n < 2) ?? $n !! fib($n-1) + fib($n-2);
> }
>
> my $N := 28;
>
> # Generate some long-lived garbage
> my @garbage;
> my $i := 1000000;
> repeat {
>    @garbage.push($i);
>    $i--;
> } while $i > 0;
>
> pir::say("Got grabage " ~ + at garbage);
>
> pir::say("fib($N) = " ~ fib($N));
>
>
> Basically it's original fib.nqp from pmichaud's mail + a lot of garbage.
>
> --
> Bacek.
> _______________________________________________
> http://lists.parrot.org/mailman/listinfo/parrot-dev
>



-- 
Jonathan "Duke" Leto
jonathan at leto.net
http://leto.net


More information about the parrot-dev mailing list