phc-discussions - Re: [PHC] EARWORM speed and improved defense

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] [day] [month] [year] [list]

Message-ID: <CAOLP8p60Z9=TUkhUuE6q9eoW=UZ9tvx+6Kj5Jjbm0ww9mrUPLQ@mail.gmail.com>
Date: Mon, 21 Apr 2014 17:07:39 -0400
From: Bill Cox <waywardgeek@...il.com>
To: discussions@...sword-hashing.net
Subject: Re: [PHC] EARWORM speed and improved defense

On Mon, Apr 21, 2014 at 4:24 PM, Solar Designer <solar@...nwall.com> wrote:

> Why did you choose 4 lanes?  Current CPUs need 5 to 9 cycles to compute
> AESENC, so 4 lanes are probably (barely) enough when running 2
> threads/core, but not enough on Intel CPUs lacking HT (OK, these are not
> meant to be used in servers, and older ones of these lack AES-NI) and
> with HT disabled (OK, maybe it's those admins' problem).
>

The algorithm normally uses 4 lanes, so mainly out of respect for the
author, I didn't change it.  I assume he had his reasons.  By my estimates,
he should be able to generate closer to 25GiB/s on my box if he had no
trouble with memory bandwidth.  Somehow going massively parallel does gain
him a couple more GiB/s...

Bill

Content of type "text/html" skipped