phc-discussions - Re: [PHC] Initial (non-proof-read) NeolKDF paper

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date: Sun, 26 Jan 2014 01:38:44 -0500
From: Bill Cox <waywardgeek@...il.com>
To: discussions@...sword-hashing.net
Subject: Re: [PHC] Initial (non-proof-read) NeolKDF paper

On Sat, Jan 25, 2014 at 11:53 PM, Dennis E. Hamilton
<dennis.hamilton@....org> wrote:
> The multiplication is kind of a congruential PRNG step modulo whatever the multiplier word size is.  My experience is that you need both terms of the multiplication to be odd to work well to avoid the poisoning effects of lots of trailing zeros in either multiplication operand.  (How that fits with regard to reversibility is not something I've thought about.)

This is interesting, because only one operand has to be odd to be
reversible.  However, if both are not odd, the output may depend
weakly on one of them.  I found that I get better scores in the
dieharder tests when I OR 3 into the low bits of one operand rather
than 1.  Is this a similar effect?

Bill