Cryptohaze.com

I'm not dead. Just slow... life interferes with coding at times.

Current improvements/stuff in the works:

- Len48 support for MD5/SHA1, Len26 support for NTLM (longest you can go without two data blocks to hash), working on adding that to other hashes. This opens up some per-position charset attacks.
- Working on some wordlist stuff... should be pretty cool if I can get it working. Very flexible, comparable to some oclHashcat capabilities (and exceeding it in others)
- A resizable curses display is finished.
- Working on some auto-tuning default parameters for different cards, and per-card parameters, plus detection of non-display cards & using long kernel lengths by default for them.
- Working on some algorithm improvements. The next release might have fewer algorithms depending on how long it takes me to optimize each one. Others will be quickly added back. I'm hoping for between 5% & 10% across the board, and if I'm right, I may be able to get some more substantial speedups on Fermis.

So, yes, I still exist.

Glad to see we'll have new toys to play with soon. I'd like to add a possible request. I'm currently going through about 20k hashes for work (Multiforcer is holding up great with that many hashes btw) and I'm currently going through various charset combinations, the problem is that sometimes these charsets will overlap resulting in hashes being cracked multiple times. One thing I'd like to see is the ability to feed an already existing output file back into the Multiforcer when it's launched, from there it would parse out the hashes and ignore ones already cracked to prevent duplicate outputs. I thought I had read somewhere about this before, and thought it was an already existing feature...but apparently I'm going crazy.

well i guess you could just carefully design the bruteforce ranges eg if you are covering say ALL Length 1-5 then start everything else at length 6 etc to miniminize/reduce the number of dupes. Takes time and planning but then you save a heck of bruteforcing.

Picch wrote:Glad to see we'll have new toys to play with soon. I'd like to add a possible request. I'm currently going through about 20k hashes for work (Multiforcer is holding up great with that many hashes btw) and I'm currently going through various charset combinations, the problem is that sometimes these charsets will overlap resulting in hashes being cracked multiple times. One thing I'd like to see is the ability to feed an already existing output file back into the Multiforcer when it's launched, from there it would parse out the hashes and ignore ones already cracked to prevent duplicate outputs. I thought I had read somewhere about this before, and thought it was an already existing feature...but apparently I'm going crazy.

This feature exists, at least in the late 0.80 alphas.

Code: Select all

--notfoundfile=filename

This should, at completion or normal program termination, write out all the not-found hashes to the specified file. You *can* use this to overwrite the input file, but I wouldn't.

Chain this file through a number of instances and it should work properly. Let me know if this isn't working for you.

In other news, my documentation sucks.

Bitweasil wrote:Len26 support for NTLM (longest you can go without two data blocks to hash)

27 is the max length for NTLM [(64 - 8 - 1) / 2 = 27.5]

Sc00bz wrote:27 is the max length for NTLM [(64 - 8 - 1) / 2 = 27.5]

You're right. Code says 27, posting when tired said 26.