use strict;
use Benchmark qw/ cmpthese timethese /;
use List::Util qw/ reduce /;

my @list = 1..10;

cmpthese(
  timethese(1000000, {
    'map_void'    => '&map_void',
    'for_loop'    => '&for_loop',
    'list_reduce' => '&list_reduce',
  }));
  
# printf "%d, %d, %d\n", map_void(), for_loop(), list_reduce();

sub map_void {
    my $result = 1;
    map {$result *= $_} @list;
    return $result;
}

sub for_loop {
    my $result = 1;
    $result *= $_ foreach @list;
    return $result;
}

sub list_reduce {
    return reduce { $a * $b } @list;
}

__END__

Benchmark: timing 1000000 iterations of for_loop, list_reduce, map_voi
+d...
  for_loop:  4 wallclock secs ( 5.43 usr +  0.00 sys =  5.43 CPU) @ 18
+4229.92/s
(n=1000000)
list_reduce:  9 wallclock secs ( 8.46 usr +  0.00 sys =  8.46 CPU) @ 1
+18175.37/s
 (n=1000000)
  map_void:  5 wallclock secs ( 4.52 usr +  0.00 sys =  4.52 CPU) @ 22
+1385.88/s
(n=1000000)
                Rate list_reduce    for_loop    map_void
list_reduce 118175/s          --        -36%        -47%
for_loop    184230/s         56%          --        -17%
map_void    221386/s         87%         20%          --
[download]

Ok....

use strict;
use warnings;
use Benchmark;

my @list = ( 1 .. 500 );
my $count = 25000;

timethese ( $count, {
    'Map'    => sub { my $result = 1; map { $result *= $_ } @list; },
    'For'    => sub { my $result = 1; $result *= $_ for @list; }
} );

__OUTPUT__

Benchmark: timing 25000 iterations of For, Map...
For: 11 wallclock secs (10.62 usr +  0.00 sys = 10.62 CPU) @ 2355.16/s
+ (n=25000)
Map: 11 wallclock secs (11.18 usr +  0.00 sys = 11.18 CPU) @ 2236.94/s
+ (n=25000)
[download]

The moral... There's no way to perform a mathematical operation based on the value of each element of a list of arbitrary values, without looking at each element of the list. And that implies (and requires) some from of iteration. map and for seem to be pretty close to each other in terms of efficiency in iterating over the list.

On the other hand, if the list consists of a known sequence, it is possible to break out the math-fu to derive a non-iterative solution. Brute force is necessary in the absence of a known mathematical sequence though.




Dave

I'm quite surprised by your results. It shows the reduce version to be the slowests, and it suggests that using a map with a block is faster than a for statement modifier. The former needs to enter/leave a block for each iteration, and the latter doesn't. A benchmark on my box shows different results:

#!/usr/bin/perl

use strict;
use warnings;

use Benchmark  qw /cmpthese timethese/;
use List::Util qw /reduce/;

our @list = 1 .. 50;

our ($map_b, $map_e, $for_m, $for_b, $red, $bc, $eval);

print "Perl version: $].\n";
system "cat /proc/version";

cmpthese -10  => {
    map_block =>   '$::map_b = 1; map {$::map_b *= $_} @::list',
    map_expr  =>   '$::map_e = 1; map  $::map_e *= $_, @::list',
    for_mod   =>   '$::for_m = 1; $::for_m *= $_ for @::list',
    for_block =>   '$::for_b = 1; foreach (@::list) {$::for_b *= $_}',
    reduce    =>   '$::red   = reduce {$a * $b} @::list',
    bc        => q !local $" = "*";
                    $::bc    = `echo '@::list' | bc`!,
    eval      =>   '$::eval  = eval join "*" => @::list',
};

print "Map block:    $map_b\n";
print "Map expres:   $map_e\n";
print "For modifier: $for_m\n";
print "For block:    $for_b\n";
print "Reduce:       $red\n";
print "bc:           $bc\n";
print "Eval:         $eval\n";
__END__

Perl version: 5.008003.
Linux version 2.4.18-3 (bhcompile@daffy.perf.redhat.com) (gcc version 
+2.96 20000731 (Red Hat Linux 7.3 2.96-110)) #1 Thu Apr 18 07:37:53 ED
+T 2002
             Rate       bc     eval map_block for_block map_expr for_m
+od  reduce
bc         2779/s       --     -55%      -88%      -94%     -95%    -9
+5%    -95%
eval       6199/s     123%       --      -73%      -87%     -88%    -8
+8%    -89%
map_block 22904/s     724%     269%        --      -53%     -55%    -5
+7%    -60%
for_block 49041/s    1664%     691%      114%        --      -3%     -
+9%    -13%
map_expr  50669/s    1723%     717%      121%        3%       --     -
+6%    -11%
for_mod   53638/s    1830%     765%      134%        9%       6%      
+--     -5%
reduce    56651/s    1938%     814%      147%       16%      12%      
+6%      --
Map block:    3.0414093201713378e+64
Map expres:   3.0414093201713378e+64
For modifier: 3.0414093201713378e+64
For block:    3.0414093201713378e+64
Reduce:       3.0414093201713378e+64
bc:           30414093201713378043612608166064768844377641568960512000
+000000000

Eval:         3.0414093201713378e+64
[download]

Abigail