Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 22 Dec 2011 17:21:51 +0530
From: piyush mittal <piyush.cse29@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: Bit slice implementation of DES based hashes

Hello,


Sorry for replying too late. I was busy in some urgent work.


 Each bit of what?
>

Each bit of processor is what I am talking of,like in 64 bit processor each
64 bit will behave as different processor.

According to Eli Biham's paper,In Des Bitslice implementation he considered
processor as a SIMD computer,ie as 64 parallel one bit processor computing
the same instruction.


"To further increase the speed"?  Would there be any speed increase at
> all (compared to a non-bitslice implementation) without this?  And what
> alternative(s) did we have?
>
>

Yes.Speed has been increased,like in his paper he told:

In DES, the input of each S box depends only on the output
of only six S boxes in the previous round. Thus, the code can be optimised
to start computing the next round while still computing the preceding one.
This can speed up implementations on pipelined processors, where we can
compute several instances in parallel.


These two are not valid answers to my questions.
>


Truly speaking I am not getting how B and K is being initialised in JTR
code according to the architecture of processor specified.And if I will be
able to get it then i can also  complete my conversions.However I read its
corresponding header file but then also I am not getting.Even I tried to
take some idea from DES_bs_get_hash() function where we are fetching hash
from B but I am not getting what "B[0] DEPTH" represents and why we have
taken b[0] to b[19].


I am sorry for asking those questions like that, but I think that you're
> missing some of the basics and I am trying to get you to realize what
> exactly you're missing and learn that.
>


No need to say sorry because in this way I can learn only.And yes its true
that I am missing some basic things as I explained you above.


Thanking You


On Tue, Dec 20, 2011 at 9:34 PM, Solar Designer <solar@...nwall.com> wrote:

> On Tue, Dec 20, 2011 at 05:56:15PM +0530, piyush mittal wrote:
> > >
> > > Can you please describe what bitslicing is, in your own words?
> >
> > In Bitslicing ,all combination of bits operates simultaneously and
> > parallelly  i.e  each bit will behave as a different processor.
>
> Each bit of what?
>
> > The same
> > concept we are using here in DES.Like in 64 bit machine 64 plain text's
> > encryption can be done simultaneously and parrallely.
>
> OK.
>
> > However here we have
> > used non standard implementation of DES i.e structure of DES is broken
> down
> > into AND,OR NOR.. gates to further increase the speed.
>
> "To further increase the speed"?  Would there be any speed increase at
> all (compared to a non-bitslice implementation) without this?  And what
> alternative(s) did we have?
>
> > > What portion(s) of the DES block(s) does each element of B[] hold?
> > >
> > > What portion(s) of the DES key(s) does each element of K[] hold?
> >
> > 64 elements of  block is holded by each element of B[]
> > 56 elements of block is holded by each element of K[]
>
> These two are not valid answers to my questions.  Here are some
> ridiculous questions in line with your answers above:
>
> What is an "element" of a block?
>
> Are you implying that one element of B[] has a different size than one
> element of K[] (64 "elements" in it vs. 56 "elements")?  Is this
> discrepancy also seen in the data types used in the program?  If not,
> what does that mean?
>
> I am sorry for asking those questions like that, but I think that you're
> missing some of the basics and I am trying to get you to realize what
> exactly you're missing and learn that.
>
> > Now I am starting NETLM_fmt.c and Oracle code I will also give you after
> it.
> > Now I will not inquire further questions to you.
>
> OK, that's your choice.
>
> Alexander
>



-- 
Piyush Mittal
Department of Computer Science and Engineering
National Institute of Technology,Rourkela
INDIA

[ CONTENT OF TYPE text/html SKIPPED ]

Powered by blists - more mailing lists

Your e-mail address:

Powered by Openwall GNU/*/Linux - Powered by OpenVZ