[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20120620161504.GC21372@openwall.com>
Date: Wed, 20 Jun 2012 20:15:04 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: EPiServer
On Wed, Jun 20, 2012 at 09:16:21AM +0200, Lukas Odzioba wrote:
> I do not know exactly how john's benchmark procedure looks like, but
> can't we use only first test vector for benchmarking, and all of them
> for correctness checking?
We could, although we need at least two salts for the "many salts"
benchmark.
> Or we could add another test vector group to fmt interface only for
> benchmarking purposes.
>
> Of course additional code is required but we could gain:
> - benchmark results comparability between different code versions
> - utilization all of those commented tests vectors
Yes, there's definitely room for improvement here. In the core JtR tree,
there are no formats where we had to comment out any tests to get proper
benchmarks. Such commenting out pretty much started with the GPU stuff.
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.
