On Fri, Sep 08, 2006 at 23:38 -0400, Uri Guttman wrote:
> >>>>> "CD" == Chris Dolan <[EMAIL PROTECTED]> writes:
>
> CD> Changing the topic a little bit... I've always suspected there was a
> CD> speed difference between local and my. So, I just benchmarked the
> CD> two versions, and "my" wins by a wide margin:
[...]
> makes perfect sense. my vars are not in the symbol table and just need
> an initialization at run time and they exist in the pad of a sub. localized
> vars need to save the previous value in some form of a stack (real work)
> and also get initialized. so local should be slower.
And using no variable is again quite a bit faster (A thing I found very
disappointing for the sake of readable code).
| ice:~/some_random_test>./perl_local_my_speed_test.pl
| Rate local my global novar
| local 549356/s -- -16% -22% -39%
| my 656410/s 19% -- -7% -27%
| global 703297/s 28% 7% -- -21%
| novar 895105/s 63% 36% 27% --
| ice:~/some_random_test>perl -v
|
| This is perl, v5.8.8 built for i386-freebsd-64int
| (with 1 registered patch, see perl -V for more detail)
Here is the modified script.
#!/usr/bin/perl -w
use strict;
use Benchmark qw(cmpthese);
cmpthese(1_000_000, {
'local' => sub {
flatten_copy_local('bar');
},
'my' => sub {
flatten_copy_my('bar');
},
'global' => sub {
flatten_copy_global('bar');
},
'novar' => sub {
flatten_copy_novar('bar');
},
});
sub flatten_copy_local {
local $_ = shift;
ref $_ eq 'SCALAR' ? "$$_" : "$_";
}
sub flatten_copy_my {
my $s = shift;
ref $s eq 'SCALAR' ? "$$s" : "$s";
}
sub flatten_copy_novar {
ref $_[0] eq 'SCALAR' ? "${$_[0]}" : "$_[0]";
}
no strict 'vars';
sub flatten_copy_global {
$s = shift;
ref $s eq 'SCALAR' ? "$$s" : "$s";
}
==============
CU,
Sec
--
The Feynman problem solving Algorithm
1) Write down the problem
2) Think real hard
3) Write down the answer
