[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20131030151745.GB28785@openwall.com>
Date: Wed, 30 Oct 2013 19:17:45 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: ZedBoard: bcrypt
On Wed, Oct 30, 2013 at 02:09:26PM +0100, Katja Malvoni wrote:
> On Wed, Oct 30, 2013 at 10:21 AM, Solar Designer <solar@...nwall.com> wrote:
> > In the utilization numbers above, do Blowfish's P-boxes count towards
> > "Register: 19%"?
>
> No, since I'm using P, expanded key and salt as 2D arrays, they are
> implemented as distributed RAM - I need to change this so that all of those
> are implemented as registers.
Oh, I forgot distributed RAM. This might actually be good. It is
non-obvious which approach is going to work better. Maybe registers
for P, but distributed RAM for expanded key and salt?
Since distributed RAM is using LUTs as tiny RAMs, this may be a reason
why our LUT count is so high (not the reason, just one of many reasons).
On the other hand, if you move that data to using registers instead, it
would probably only increase the number of slices per core (and thus
reduce core count). Yet you may try...
We're probably spending most LUTs on something else.
Alexander
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.
