> At 04:10 PM 11/5/2003 -0700, Luke Palmer wrote:
> >Actually that was pretty good for an early version. You could help IMCC
> >out by not creating those PerlUndefs until you're going to assign to
> >them.
>
> Also, anytime you use a temporary to assing a constant literal, you
> should be able to use a I/S/N reg. Parrot's ops are polymorphic
> so you should have a lot of flexibility even with the basic types.
>
> >Vtable jump isn't really that much overhead in a language like C.
> >Parrot's already running pretty fast. We're making things fast by
> >keeping it in mind while implementing it, including a rich opcode
> >and vtable so the number of said slow dispatches are small, among other
> >various things that I don't know. Maybe Dan will chime in here :-)
>
> Thies, have you actually benchmarked your sample
> with JIT against PHP?
>
A bench with a mandelbrot set of depth 500 (script attached)::
php mandel2.php 1.93s user 0.01s system 76% cpu 2.552 total
../../parrot/parrot mandel.imc 1.08s user 0.02s system 41% cpu 2.643 total
../../parrot/parrot -j mandel.imc 0.77s user 0.01s system 57% cpu 1.368 total
Is the result.
> >It's great that you're doing PHP for Parrot.
> >
> >Luke
>
> Darn right!
>
Btw, incase people on the list are interested there is a presentation
about PHP and Parrot we did at http://www.edwardbear.org/pap.pdf
-Sterling
<?php
$b = " .:,;!/>)|&IH%*#";
$y=0; $C=0; $x=0; $c=0; $z=0; $Z=0; $t=0; $k=0; $i=0; $r=0;
//float r, i, z, Z, t, c, C;
for ($y=30; $C = $y*0.1 - 1.5, $y--;){
echo "\n";
for ($x=0; $c = $x*0.04 - 2, $z=0, $Z=0, $x++ < 75;){
for ($r=$c, $i=$C, $k=0; $t = $z*$z - $Z*$Z + $r, $Z = 2*$z*$Z + $i, $z=$t,
$k<500; $k++)
if ($z*$z + $Z*$Z > 10) break;
echo $b[$k%16];
}
}
echo "\n";
?>
