I agree your question is badly posed; I, at least, don't really understand the details.

See How to ask better questions using Test::More and sample data. You might start with something like

use strict;
use warnings;

use Test::More 'tests' => <number of tests>;
use Test::NoWarnings;  # adds 1 test

use Data::Dumper;  # for debug

my @al = (...);

my @distribution = (
    3,  4,
    5,  5,
    6,  6,
    8,  10,
    );

my $ar_expected_output = (
    [ ... ],
    [ ... ],
    ...
    );

my $ar_got_output = distribute(\@al, \@distribution);

is_deeply $ar_got_output, $ar_expected_output;

exit;

# subroutines #############################################

sub distribute {

    my ($ar_input,         # array ref.: input array
        $ar_distribution,  # array ref.: distribution to output arrays
        ) = @_;

    my  @output;

    ...

    return \@output;

    }
[download]

Construct the output arrays first and test the process. Once you have the output arrays constructed, it should be fairly straightforward to output them to files.

See also Short, Self-Contained, Correct Example and Perl Data Structures Cookbook.

See perlintro for basic file I/O stuff, open and perlopentut for more advanced stuff.

Maybe also see I know what I mean. Why don't you?