Re^19: Module for 128-bit integer math? (updated)

I haven't tried to build this yet, ((I think?)I would need to use MPIR on 64-bit), so I haven't seen the full output, but am I missing something with regard to your use of ior rather than xor?

sub FNV_1_128_gmpn {
    my $s = shift;
    mpn_set_str(my $h, '144066263297769815596495629667062367629', 10, 
+128);
    mpn_set_str(my $p, '309485009821345068724781371', 10, 128);
    for (unpack 'C*', $s) {
        my $h2 = $h;
        mpn_mul($h, $h2, $p);
        mpn_ior_uint($h, $h, $_);
    }
    return $h;
}

sub FNV_1a_128_gmpn {
    my $s = shift;
    mpn_set_str(my $h, '144066263297769815596495629667062367629', 10, 
+128);
    mpn_set_str(my $p, '309485009821345068724781371', 10, 128 );
    for (unpack 'C*', $s) {
        mpn_ior_uint($h, $h, $_);
        my $h2 = $h;
        mpn_mul($h, $h2, $p);
    }
    return $h;
}
[download]

Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.

"Science is about questioning the status quo. Questioning authority".

In the absence of evidence, opinion is indistinguishable from prejudice.

Comment on Re^19: Module for 128-bit integer math? (updated) Download Code

Replies are listed 'Best First'.
Re^20: Module for 128-bit integer math? (updated) by salva (Canon) on Feb 23, 2011 at 22:35 UTC
but am I missing something with regard to your use of ior rather than xor? Oops, that was an error on the benchmark script. It has been corrected now (and a new version of the module released).	[reply]