[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 13 Feb 2013 20:20:34 +0100
From: magnum <john.magnum@...hmail.com>
To: john-users@...ts.openwall.com
Subject: Re: RAR Cracking with JtR Jumbo (Files found during forensics)
On 13 Feb, 2013, at 19:37 , magnum <john.magnum@...hmail.com> wrote:
> On 13 Feb, 2013, at 17:25 , Solar Designer <solar@...nwall.com> wrote:
>> Actual speeds I am getting for these are:
>>
>> HD 7970: 861 c/s
>> GTX 570 o/c: 1219 c/s
>>
>> I'm not sure why 7970's speed has dropped so much (vs. --test). magnum?
>
> RAR speed is highly dependant on password length. Length 4 is significantly faster than length 8. Our GPU benchmark use length 6. The speed suffers as soon as *any* password in a keys-per-crypt batch is longer. Unfortunately this is often the case for wordlist runs. You can mitigate it by sorting the file by length but adding --rules to the mix will render that effort pointless.
>
> Also, I had over 7000 c/s [for length 6] at one point, using 7970. I have tried to find out why it has dropped but can't see why. It might be a driver thing. The nvidia speed is better than ever (but still low).
You could try raising HASH_LOOPS to 64 (or some other power of 2) for a boost (but this is not the real answer for the performance drop). There is a hard limit for 200 ms anyway, but if that means you get higher speed at lower GWS it will mitigate the variable-length problem a little too.
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.
