phc-discussions - Re: [PHC] A must read...

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: Wed, 15 Jan 2014 17:38:57 +0000
From: Peter Maxwell <peter@...icient.co.uk>
To: discussions@...sword-hashing.net
Cc: Solar Designer <solar@...nwall.com>
Subject: Re: [PHC] A must read...

On 15 January 2014 17:26, Krisztián Pintér <pinterkr@...il.com> wrote:

>
> Solar Designer (at Wednesday, January 15, 2014, 3:54:15 PM):
>
> > Besides integer multiply, I was considering floating-point
>
> since it came up, i use this opportunity to ask the list:
>
> how reliable floating point is from interop standpoint? i mean:
>
> 1, the in-memory storage of single and double types. can we expect
> different architectures having same bit pattern for the same value?
>
> 2, exact results. should we expect different architectures supporting
> the same type (like IEEE 754) to produce different results due to
> different rounding or otherwise?
>
> so at the end, my question for example is: if we take two 32 bit
> words, treat them as singles, convert them to doubles, multiply them,
> and then store the resulting double in a 64 bit word, can we expect
> different architectures end up with different bit patterns?
>

Well, using floats in cryptography is not exactly a new idea but I suspect
it does have some portability problems, e.g. there's implementation of
Bernstein's poly1305 using floating point - http://cr.yp.to/mac/53.html -
although I haven't actually sat down to look at it... it's sitting fairly
far down on my to-do list.

Content of type "text/html" skipped