Better yet, if you're going to try to do something with multiple children, here's an example to speed you up. Create a file for input with a number of sleep commands with random values, then test.

#!/usr/bin/perl

# multi.pl
# usage: multi.pl [max_tasks] < job_list
# max_task defaults to 4
#
# do all the jobs as quickly as possible in parallel, but no more than
#   max_tasks at a time.

use warnings;
use strict;

sub mysystem
{
    my $cmd = shift;
    my $pid;
    #print "forking: $cmd\n";
    if (($pid = fork()) == 0) { exec($cmd); } # child
    return $pid; # parent
}

# how many tasks can we do at once
my $max_tasks = 4;
if (defined($ARGV[0])) { $max_tasks = shift(@ARGV); }
# num of curr tasks
my $num_tasks = 0;
# pids running, and command run
my %pids = ();
my ($pid, $cmd);

# read tasks
while ($cmd = <>)
{
    chomp($cmd);
    # execute the task
    print "starting: $cmd\n";
    $pid = mysystem($cmd);
    #print "forked pid $pid\n";
    $pids{$pid} = $cmd;
    $num_tasks++;
    # if we're doing all we should, wait for 1 to end
    while ($num_tasks >= $max_tasks)
    {
        $pid = wait();
        #print "reaped pid $pid\n";
        last if ($pid == -1); # error?
        print "finished: $pids{$pid}\n";
        delete($pids{$pid});
        $num_tasks--;
    }
}

# wait for all children to end
while ($num_tasks > 0)
{
    $pid = wait();
    last if ($pid == -1); # error?
    print "finished: $pids{$pid}\n";
    delete($pids{$pid});
    $num_tasks--;
}

exit(0);
[download]

sleep  4
sleep  2
sleep  6
sleep  3
sleep  4
sleep  2
[download]