Hi all,

While playing around with Parallel::ForkManager i stumbled over Re^3: Fork vs pThreads where BrowserUk wrote:

"...running 50 tasks concurrently on 4 cpus will take longer than running those same 50 tasks; but only 4 at any given time"

This sounds logical.

I wrote this:

#!/usr/bin/env perl

use strict;
use warnings;
use Math::BigInt;
use Parallel::ForkManager;
use Time::HiRes qw ( time );
use feature qw(say);
use Iterator::Simple qw(iter);

my @numbers  = ( 1 .. 2000 );
my $processes   = shift;
my $tmp      = q(/tmp/);
my $pm       = Parallel::ForkManager->new( $processes, $tmp );
my %data     = map { $_ => undef } @numbers;

say qq(numbers: ), scalar @numbers;

say qq(processes: $processes);

$pm->run_on_finish(
    sub {
        my ( $id, $result ) = ( $_[2], $_[5] );
             $data{$id} = ${$result}
    }
);

my $start=time;

my $iterator = iter(\@numbers);

while(defined( my $number = $iterator->())) {
    $pm->start($number) and next;
    my $factorial = Math::BigInt->bfac($number);
    $pm->finish( 0, \$factorial );
}
$pm->wait_all_children;

say qq(fork: ), time - $start;

%data = map { $_ => undef } @numbers;

$start = time;

for my $number (@numbers) {
    my $factorial = Math::BigInt->bfac($number);
    $data{$number} = $factorial;
}

say qq(for: ), time - $start;
[download]

I think the results are quite surprising:

karl@host:~$ ./forker.pl  50
numbers: 2000
processes: 50
fork: 22.5553870201111
for: 33.7985281944275

karl@host:~$ ./forker.pl  4
numbers: 2000
processes: 4
fork: 501.324076890945
for: 33.1997199058533
[download]

I assume (and hope) there is nothing wrong with my code but i have the vague feeling that i make some false assumptions or jump to some wrong conclusions.

What is my error in reasoning (if any)?

Update:

Many thanks to all for this interesting thread.

Update 2: Please see Allow for sub-second resolution #5 and Changes.

Thank you very much for any hints and best regards, Karl