| 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
| ||
|
Date: Thu, 26 Mar 2015 07:18:14 -0700
From: Bill Cox <waywardgeek@...il.com>
To: "discussions@...sword-hashing.net" <discussions@...sword-hashing.net>
Subject: Re: [PHC] Another PHC candidates "mechanical" tests (ROUND2)
On Thu, Mar 26, 2015 at 6:55 AM, Bill Cox <waywardgeek@...il.com> wrote:
> The impact I'm seeing is that L3-bound SSE optimize memory-hard hashing
> does not work and play well with other L3 intensive tasks on the same
> server. It is literally faster to run Scrypt, then run another task, than
> to run the two in parallel, even when both tasks are single-threaded.
>
> This means there is money to be saved by running authentication servers
> that offload memory-hard hashing from other servers. I would think this
> would be good for your use case :-)
>
One more issue: If you're running an Scrypt service on a server, and some
other latency critical services gets assigned to the same server, it gets
ugly. A 1ms Scrypt hash can easily slow down the other service from a
typical 1ms to 2 or even 3ms. It happens randomly because they wont always
be running at the same time. It makes it difficult to manage services on
servers when you have this much random variation.
I personally am now a big fan of your authentication server concept. It's
more efficient, allows a large ROM to be used, and safer in general.
However, this is just my personal opinion, and is likely not to be
reflected by my employer.
Bill
Content of type "text/html" skipped
Powered by blists - more mailing lists