Piping speed from statsprocessor to oclhashcat
#1
Hi all. Tried searching but no one seemed to ask this question so here it is:

I was trying a custom markov from a dictionary (e.g. linkedin) and ran statsprocessor.

Is there a speed difference between piping in Windows or in Linux? I am roughly getting 80k tries on Windows (which was relatively slower than dictionary) so I'm wondering if it's a Windows thing or it's about the same number of attempts in Linux/*nix as well?

Also, would it boost up the speed more if you just store the output of statsprocessor into a file and run it as a dict in the end (assuming a decent amount of HDD capacity with a limited threshold and length)? Or in totality the amount of time you need to save sp output to a file will end up as the same as piping it to oclhashcat directly?

Pardon the noob questions, and thanks for your time
#2
Depending on the algorithm you use, this speed sounds to low. For example if this is MD5, you should be in den 500M or more range. If thats so, you might have forgotten to add rules as a multiplier. This is required to get maximum speed.
#3
(09-05-2013, 05:46 PM)atom Wrote: Depending on the algorithm you use, this speed sounds to low. For example if this is MD5, you should be in den 500M or more range. If thats so, you might have forgotten to add rules as a multiplier. This is required to get maximum speed.

Hi atom!

All the hashes I'm trying to crack are on MD5...

This is what I'm getting after I add a rule (best64)

Starting attack in stdin mode...


Session.Name...: cudaHashcat-plus
Status.........: Aborted
Rules.Type.....: File (rules\best64.rule)
Input.Mode.....: Pipe
Hash.Target....: File (C:\Users\eXPeri3nc3\Desktop\md5.txt)
Hash.Type......: MD5
Time.Started...: Fri Sep 06 00:18:12 2013 (5 mins, 44 secs)
Speed.GPU.#1...: 169.8 MH/s
Recovered......: 0/5556 (0.00%) Digests, 0/1 (0.00%) Salts
Progress.......: 53715517440
Rejected.......: 0
HWMon.GPU.#1...: 6% Util, 63c Temp, N/A Fan

Started: Fri Sep 06 00:18:12 2013
Stopped: Fri Sep 06 00:23:56 2013

Much faster! However, like what you've said, I normally run a dictionary cracking at the speed of 450.0 MH/s so this is still 1/3 of it.

Can you please explain why the rule is needed as a multiplier?

--

Also, one question on statprocessor. If you run the same output twice, will the output be statistically the same?

Thanks a lot!
#4
add more rules
#5
(09-05-2013, 06:46 PM)undeath Wrote: add more rules

If I add more than 2 rules, the speed decreases. I tried just now, if I add like best64, toggles2, toggles3 it went from 170MH/s to 150MH/s and so forth
#6
Try passwordspro.rule. It should be big enough. If you do not get more speed with it than with best64, then you're already running at maximum speed. Also note that rule-based attacks are slower than brute-force since the candidate generation takes more time, too.
#7
(09-06-2013, 10:01 AM)atom Wrote: Try passwordspro.rule. It should be big enough. If you do not get more speed with it than with best64, then you're already running at maximum speed. Also note that rule-based attacks are slower than brute-force since the candidate generation takes more time, too.

With passwordspro or d3ad0ne rule, I can hit 220MH/s, which is my highest so far. Thanks.

Can you please kindly share why rules are needed as a multiplier? I can't seem to find it in the wiki...
#8
your cpu simply is too slow to supply enought words in time.
#9
(09-07-2013, 06:07 PM)undeath Wrote: your cpu simply is too slow to supply enought words in time.

Intel(R) Core(TM) i7-3610QM CPU @ 2.30GHz

Shouldn't be too slow I guess
#10
it's not that the cpu is too slow, it's that you're constantly having to make large transfers from the host to the device.