To give you a general idea:

#!/usr/bin/perl -w
use strict;
use Time::HiRes qw/gettimeofday tv_interval/;

my(@data, %seen);
$data[$_] = int rand 100 for 0 .. 1e6 - 1;  # 1,000,000 pieces of data

my $t0 = [gettimeofday];

for (@data){
    1 unless $seen{$_}++;   # Perform a no-op the first time we see a 
+given number
}
print "Elapsed time: ", tv_interval($t0, [gettimeofday]), " seconds\n"
+;
[download]

Running this 1 million-item data list with a hash lookup for each item on my little laptop (Intel Atom 1.6GHz) gives the following:

Elapsed time: 1.466658 seconds
[download]

Reasonably fast, I'd say.

-- 
Education is not the filling of a pail, but the lighting of a fire.
 -- W. B. Yeats