Are you absolutly sure you are digesting exactly the same sequence of characters, including line end characters? One way to check would be to create a test file using a Perl script then digest it with both systems and check that result:

use strict;
use Digest::MD5;

open OUTFILE, '<', 'temp';
print OUTFILE <<TEXT;
It doesn't matter what the text is it should
getthe same treatment from both digests
TEXT
close OUTFILE;

my $md5 = Digest::MD5->new();
$md5->add('TEXT');
my $digest = $md5->hexdigest;

print "Perl: $digest\n";

`systemmd5util temp > temp.out`;

print "System: ";
local @ARGV = ('temp.out');
print while (<>);
[download]

Untested code.