It's clearly not a significant improvement, but does seem to be consistently a tiny bit faster...sub blokhead (@) { my @a = (0 .. $#_); my $i = @_; my $n; map+( $n=rand($i--), $_[$a[$n]], $a[$n]=$a[$i] )[ 1 ], @_; }
On large data:
On small data:our @data = map { 'x' x 1000 } 1..1000; cmpthese -2, { map { $_ => "$_ \@data" } qw/naive listutil buk ikegami + blokhead/ }; Rate naive ikegami listutil buk blokhead naive 76.1/s -- -26% -47% -88% -88% ikegami 103/s 36% -- -28% -84% -84% listutil 143/s 89% 39% -- -78% -78% buk 638/s 739% 518% 345% -- -1% blokhead 643/s 745% 522% 348% 1% --
our @data = ("xxx") x 1000; cmpthese -2, { map { $_ => "$_ \@data" } qw/naive listutil buk ikegami + blokhead/ }; Rate naive listutil ikegami buk blokhead naive 328/s -- -26% -44% -49% -49% listutil 445/s 36% -- -24% -30% -31% ikegami 589/s 80% 32% -- -8% -9% buk 637/s 94% 43% 8% -- -1% blokhead 646/s 97% 45% 10% 1% --
blokhead
In reply to Re^2: About List::Util's pure Perl shuffle()
by blokhead
in thread About List::Util's pure Perl shuffle()
by blazar
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |