Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ab61e606d277b8df7ebd22f202d11245@smtp.hushmail.com>
Date: Wed, 14 Jan 2015 01:33:48 +0100
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
CC: atom <atom@...hcat.net>
Subject: Re: PRINCE

On 2015-01-14 00:58, magnum wrote:
> On 2015-01-13 19:49, Solar Designer wrote:
>> As a maybe better alternative, what about "double"?  Have you tried?
> 
> Good idea, that was totally trivial given my present code. The modulo
> operation now used fmod() and got very slow so I also tried a version
> that casted the modulos and divisions to uint128_t and otherwise use
> double. That one was actually fastest of them all with a slight margin:
> 
> Running in 30s against 1 fake NT hash:
> =========================================
> double w/ int128 mod/div: 8940 Kp/s  146%
> int128:                   8843 Kp/s  144%
> GMP:                      6141 Kp/s  100%
> double:                   5441 Kp/s   89%

Sorry, I screw up the double+cast build. It's not faster at all. Here's
a correct table:

Running in 30s against 1 fake NT hash:
=========================================
int128:                   8843 Kp/s  144%
GMP:                      6141 Kp/s  100%
double:                   5441 Kp/s   89%
double w/ int128 mod/div: 5253 Kp/s   86%

magnum

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.