The following code, scaled up to 500,000 filenames and 100,000 IDs, takes 2 seconds to run and uses 90Mb approx.

use warnings;
use strict;

# create some sample data
my (@pathnames, @safe_list);
push @pathnames, sprintf "C:/abc/abc1/GS%06d", $_ for 1..500_000;
push @safe_list, sprintf "GS%06d", $_ for 1..100_000;

my %safe_hash;
$safe_hash{$_} = 1 for @safe_list;

my @list;
for (@pathnames) {
    # (adjust regex to match whatever the filename/ID format is)
    /\/(\w+)$/ or die "bad pathname: $_";
    push @list, $_ unless $safe_hash{$1};
}
[download]

Dave.