Mask prepend string performance

[Sydnic]

Hi, There was another post about this, a while back, but there was no suggestion to a workaround. The problem is when I have a known string as part of the passphrase, i am including it in the mask like this: "pass?l?l?l?l". This however reduces my performance by a lot! From 11000 MH/s to 250 MH/s. If I were to append it after..like "?l?l?l?lpass" , this wouldn't reduce the performance at all. Is there a way to avoid this? And why does this happen..is it a bug?

[atom]

It's not a bug, it's how it's designed. Depending on the hash-type, which you didn't specify, you can workaround this easily.

[Sydnic]

It was md5 hash, sorry for not specifying it. I believe I tried this before for sha512 and it was similar decrease in performance, not entirely sure. Could you help me out by showing me how I could work around that decrease in performance? So i guess i want for example a brute force of all the combinations (of lets say 10 lower case letters) that start with the substring "pass" (e.g. pass?l?l?l?l?l?l as a mask).

[atom]

If it's MD5, just append the known substring as a salt. That means, have a hash file in this format:

hash:pass

And then use -m 20 to crack it with the mask ?l?l?l?l?l?l

That should give you a decent speed. Don't forget -w 3

[Sydnic]

That is definitely better, thanks a lot!

I can however now notice a difference between -m 10 (string is appended) and -m20 (string is prepended), as in -m 20 goes at half speed (5000 MH/s vs 10000 MH/s). Is there more I can do about that, or that's just how -m 20 works? Still, thats good improvement from 250 MH/s to 5000 MH/s.

[atom]

No, that's related to a cryptographic flaw in MD5. But you can exploit it only efficient in a salt-appending situation.