You are wasting your time trying speed this up using multi-processing when a simple, single-threaded piece of Perl can process your 100,000 lookups far faster than your 30 second target.

Less than 2/10ths of a second in fact:

#! perl -slw
use strict;
use Time::HiRes qw[ time ];

our $N //= 1e5;

my @strings = map sprintf( '%18.18b', int rand( 2**18 ) ), 1 .. $N;

my %lookup = (
    '101010101010101010' => [ 1350, 9234, 8889 ],
    '010101010101010101' => [ 1345, 2234, 3689 ],
    '111111111000000000' => [ 2256, 3370, 1340 ],
);

my $start = time;

my @vals = @{ $lookup{ $_ } // [0,0,0] } for @strings;

printf "$N lookups took %f seconds\n", time() - $start;

__END__
C:\test>junk
100000 lookups took 0.137553 seconds
[download]