Archives for: February 2011


As you know atom is developing a new branch of oclhashcat called oclhashcat-lite. This version is dedicated to extreme easy handling, extreme optimizations of single-hash processing and it will be a research lab for atom.
The thought behind the lite version is that the user will soon get results by just entering the hash without taking care of masks, rules, gpu-loops or any of the advanced options that are available in the other versions. If the user then wants more options he will have the commitment learning to use the advanced versions.

One of the first questions that popped out of my mind as atom told me about the lite version was what will be the "predefined" attack mode? So basicly it will be a mask attack with followng features:
- 8 chars (cause that will hit approximitly 90%)
- lower chars
- digest chars
- some symbols

He constructed this because this will take approxmitly 1 hour one an ATI 5970.


atom has successfully enhanced the handling of salted hashes. It's now possible to run your attacks agains 32 byte salted hashes with no significant loss of speed!

root@sf:~/oclHashcat-0.26# ./oclHashcat64.bin -m 1 74685d01e04319787bdbedcd07369d88:5iaM55xWHMXcasPkehTpZk1v3FtphPJA ?l?l?l?l?l?l?l?l
oclHashcat lite v0.1 starting...

Platform: ATI compatible platform found
Device #1: Cypress, 512MB, 0Mhz, 20MCU
Device #2: Cypress, 512MB, 0Mhz, 20MCU


Status....: Finished
Length....: 40
Mask......: '?l?l?l?l?l?l?l?l' (208827064576)
Speed.GPU*: 9500.8M/s
Progress..: 108213043200/208827064576 (51.82%)
Running...: 11 secs

Started: Sun Feb 27 23:24:27 2011
Stopped: Sun Feb 27 23:24:40 2011


Trailer online:


Hashcat is a bitch that never sleeps!

In this spirit there are plenty of news round about the fastest cracking tool in world. At first there is a relative great story to write about. Atom focused on optimizing as you all know that there a major improvements in the current beta v0.26 and there will be more till the release of the version.

As a side project he decided to fork a new branch out of the v0.26 and call it oclHashcat-Lite. This new branch will, as the name is telling, be focused on usability and lightwight handling so it's ultimate goal is to make it easier for beginner get good results.

There are some details on how this will be approached:
- All options of the main branch will be available but optional so that the user only has to submit the hash itself
- The core cracking method will be the extrem fast bruteforcer
- Addionally to the algorythms oclhashcat support at this moment there will be MD5 and SHA1 salted

Atom's work on main branch of oclHashcat v0.26 now is based on the thesis that it was still extremly fast on the one side but on the other there was pretty much potential left unused because of the worldlists. Now with bruteforce there are much optimizations to tune such as reversing.
Here are some numbers to point this out (single hash):
- SHA1 3B/s
- MySQL-old 50B/s
- MD5 again 50M/s faster to the previous world record of 9637 M/s and is now 9687 M/s

He told me that his whole work is at the moment based on ATI and his private opinion is that he would kick the nVidia support out of the codebase because the performance of ATI compared is outstanding. Fear the mighty and make him sacrifices! Don't be a fool, he doesn't drop nVidia support, he loves you guys no matter what card you use!

Stay tuned!


Finally atom decided to go public with the v0.26 beta of oclhashcat.
The major improvement of v0.26 is the use of BFI_INT as described previous (hack of the binary kernels). The other great change of this version is that he managed to optimize his single hash reversing MD5 kernel in about two days so that it sets new records in performance!

To impress this here are the numbers of the benchmarks performaned on hd5970 at stock clocks:
- oclhashcat v0.26 9637 M/s
- ighashgpu v0.92.17.2 8561 M/s
- whitepixel-2 8275 M/s

VERY VERY INTERESTING: As addition i'm very proud to announce that atom found a weakness in SHA1 that exploitation will signifcantly improve the performance and will be implemented soon!

Benchmark of oclhaschat v0.26:

For more details take a look in the forum: