09-01-2019, 08:13 PM
Hello,
I've came across a new hash algorithm based on MD5 recently which was not supported by Hashcat so I set out to write my own OpenCL kernel for it.
After a lot of (frustrating) time researching and trial and error, I finally managed to get it working.
My kernel was based on the algorithm 3710 - md5($salt.md5($pass)). I've only implemented the pure version and only for straight mode attack to use wordlists.
However while it works fine at this point, it only seems to work when I use the flags "-T 1 --force" which I'm sure it is limiting my cracking speeds a lot. If I remove these flags, Hashcat uses the 64 kernel threads by default (I think) and it doesn't seem to be able to crack any passwords.
Any hints on what could be the problem? I'm really clueless on this one.
Let me know if you want to check my code, I would be happy to share it.
I'm using the source code of Hashcat 5.1.0 running with a single OpenCL device (nVidia).
Thanks.
I've came across a new hash algorithm based on MD5 recently which was not supported by Hashcat so I set out to write my own OpenCL kernel for it.
After a lot of (frustrating) time researching and trial and error, I finally managed to get it working.
My kernel was based on the algorithm 3710 - md5($salt.md5($pass)). I've only implemented the pure version and only for straight mode attack to use wordlists.
However while it works fine at this point, it only seems to work when I use the flags "-T 1 --force" which I'm sure it is limiting my cracking speeds a lot. If I remove these flags, Hashcat uses the 64 kernel threads by default (I think) and it doesn't seem to be able to crack any passwords.
Any hints on what could be the problem? I'm really clueless on this one.
Let me know if you want to check my code, I would be happy to share it.
I'm using the source code of Hashcat 5.1.0 running with a single OpenCL device (nVidia).
Thanks.