Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 20 Sep 2012 10:50:35 +0200
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: 1.7.9-jumbo-7

On 20 Sep, 2012, at 2:14 , Solar Designer <solar@...nwall.com> wrote:

> magnum, Jim, Claudio, Lukas -
> 
> On Wed, Sep 19, 2012 at 02:19:12AM +0200, magnum wrote:
>> I hope all is well now.
> 
> Thank you!  Here are a few more things I noticed:
> 
> dumb16.conf and dumb32.conf started to use cipher_limit, but it is not
> defined anywhere.  I'll drop these changes.

Right, that shouldn't be in -fixes.

> macosx-x86-64-gpu, macosx-x86-64-cuda targets have been added - do they
> actually work?

All CUDA formats work fine. Many OpenCL formats too, unless you revert the added "inline" to some kernel functions as discussed in other thread. They can be changed to "static" but it has to be either (or perhaps both).

> alghmac.h change:
> 
> +#ifndef __APPLE__
> #include <malloc.h>
> +#endif
> 
> Wouldn't it be more correct to #include <stdlib.h> (on all platforms)
> instead, or something like that?  (Note: I did not review this in context.)

Perhaps. I just got it working on OSX and forgot about it.

> Please help me describe the changes to common-opencl.[ch] - what are the
> additions?  Do I say this is "better autodetection" or something?  I'd
> appreciate a properly worded paragraph describing this.

From memory the most important fixes are OSX fixes, better auto-tuning to device, and automatic selection of a GPU, if present, unless you explicitly state --platform and/or --device.


> "RAR: better early rejection (speedup)" - correct description?

Well that about explains it. The speed for cracking large -p archives went from very poor to much closer to -hp archives. But you could just say "performance fixes" or whatever.

> Lukas, magnum - how come Lukas' phpass optimizations, which Lukas
> extensively tested prior to CMIYC 2012 (in case we'd be given these
> hashes in the contest), are not in the fixes branch?  I'd like to add
> them now.

Last patch I saw, he stated "not good enough to be placed in repository", so I didn't:
http://www.openwall.com/lists/john-dev/2012/08/22/1

magnum

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.