It is actually potentially more difficult than that. If you're actually using threads, then one iteration should start before the other finishes.

What you need to do is use threads::shared then share $i across threads. That way no matter what thread updates it, the others will see the update. With one caveat. There might be a possible race condition where 2 threads try to ++ at the same time and you create a problem internally. This should be rare, but to be safe you should lock. So that looks something like this:

use strict;
use threads;
use threads::shared;
my $i :shared;
$i = 1;

# Time and code passes.
for my $id (@article_id) {
    lock($i);
    print "$i\n";
    $i++;
}
# More code.
[download]

Note that if you want to put long-running code in that loop you will lose the benefits of threading because different threads will block on the lock on $i. If you want to solve that you can do this:

use strict;
use threads;
use threads::shared;

{
    my $i :shared;
    sub get_i {
        lock($i);
        return ++$i;
    }
}

# Time and code passes.
for my $id (@article_id) {
    my $i = get_i();
    print "$i\n";
    # Do lengthy processing.
}
# More code.
[download]

Update: johngg pointed out that the modules were named threads but my code called them thread instead. Fixed.