phc-discussions - Re: [PHC] Argon2 version 1.2.1 release

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]

Message-ID: <CALW8-7Kx_ZSykbVWdosW0XaErv+FGUnW7rcNbq-FJMCgTzX4og@mail.gmail.com>
Date: Wed, 26 Aug 2015 15:04:12 +0200
From: Dmitry Khovratovich <khovratovich@...il.com>
To: "discussions@...sword-hashing.net" <discussions@...sword-hashing.net>
Subject: Re: [PHC] Argon2 version 1.2.1 release

OK, I added this warning.

On Wed, Aug 26, 2015 at 2:54 PM, Solar Designer <solar@...nwall.com> wrote:

> Hi Dmitry,
>
> That's great news.  Thank you!
>
> Since this version isn't necessarily "the final PHC-winner-Argon2", I
> think it's preferable that you add a statement/warning to this extent to
> the start of README.  As it currently is, people coming to the GitHub
> page might think it's the final Argon2, and they would be upset if you
> have to break compatibility later.  As JP wrote:
>
> On Wed, Aug 26, 2015 at 12:27:03PM +0000, Jean-Philippe Aumasson wrote:
> > Thanks! We're gonna review the changes done, and we'll discuss on how to
> > move from this version to the final PHC-winner-Argon2.
>
> As to GPU implementations, we only have some minor news:
>
> - Some of the GPU performance figures for the original Argon2 have been
> improved slightly since my last posting.
>
> - Exploiting the original Argon2's 8-way BLAKE2b parallelism on GPUs
> turned out to be non-trivial because of the SIMT vs. SIMD aspect.
> Unfortunately (for the original Argon2's security), this may be more of
> an implementation difficulty than a fundamental limitation of GPUs.
> We might give some non-trivial approaches a try, but this isn't going to
> be as simple as programming with vector types in OpenCL (those vectors,
> so to speak, get rotated 90 degrees against our desired placement).
>
> - We should be able to test the existing OpenCL kernels on Titan X soon.
>
> Thanks again,
>
> Alexander
>



-- 
Best regards,
Dmitry Khovratovich

Content of type "text/html" skipped