[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20150819153703.GA16854@openwall.com>
Date: Wed, 19 Aug 2015 18:37:03 +0300
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: PHC: Argon2 on GPU
On Wed, Aug 19, 2015 at 07:10:42AM +0300, Solar Designer wrote:
> For Argon2i, you should be able to easily optimize this overhead out,
> since all the indices are known in advance (they are the same each
> time, by design, as required to avoid cache timing leaks). You should
> also be able to optimize out the hashing that produces those indices
> (before the modulo division), but that's relatively minor (yet by all
> means make this optimization as well if you do precompute the indices).
>
> This means you will need some memory to store those indices in (1536 of
> them for our current benchmarks? meaning something like 3 KB?), but this
> memory can be shared between different concurrent hash computations.
Well, not 1536, but 3 times that, for t=3. Still easily affordable.
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.
