On Wed, Jul 22, 2009 at 12:11, Bryan R
Harris<[email protected]> wrote:
snip
> If your arrays could be very large, it's a waste of cycles to sort the list
> then throw most of that effort away. You could do:
snip
Due to the fact that sort is written in C, it can be faster to sort
than to loop over the array in Perl. I remember the value as being
around 250, but there is no need to count on memory, just benchmark:
sort => z
maxstr => z
loop => z
for a list of 10 items:
Rate loop sort maxstr
loop 378300/s -- -37% -71%
sort 601510/s 59% -- -54%
maxstr 1310719/s 246% 118% --
for a list of 100 items:
Rate sort loop maxstr
sort 34909/s -- -44% -81%
loop 61837/s 77% -- -66%
maxstr 183991/s 427% 198% --
for a list of 200 items:
Rate sort loop maxstr
sort 14628/s -- -56% -85%
loop 33185/s 127% -- -65%
maxstr 94575/s 547% 185% --
It looks like my memory is wrong, (or there is a difference between
5.6 and 5.10). For this you can see that it is always better to use
the List::Util's versions (about an order of magnitude better). This
is the code that generated the benchmark:
#!/usr/bin/perl
use strict;
use warnings;
use List::Util qw/maxstr/;
use Benchmark;
sub random_strs {
my $n = shift;
return map { ("a" .. "z")[rand 26] } 1 .. $n;
}
my @a = random_strs 10;
my %subs = (
loop => sub {
my $max = $a[0];
for my $val (@a) {
$max = $val if $max lt $val;
}
return $max;
},
maxstr => sub {
return maxstr(@a);
},
sort => sub {
return +(sort @a)[-1];
},
);
for my $sub (keys %subs) {
print "$sub => ", $subs{$sub}->(), "\n";
}
for my $n (10, 100, 200) {
@a = random_strs $n;
print "\nfor a list of $n items:\n";
Benchmark::cmpthese -1, \%subs;
}
--
Chas. Owens
wonkden.net
The most important skill a programmer can have is the ability to read.
--
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
http://learn.perl.org/