Description: aircrack-ng VS pyrit and why you need a video card even if you don't play games
Tags: aircrack-ng pyrit wpa ,
Disclaimer: We are a infosec video aggregator and this video is linked from an external website. The original author may be different from the user re-posting/linking it here. Please do not assume the authors to be same without verifying.
When I was looking at CUDA/OpenCL it looked like the bottleneck was going in and out of the card. I wonder if anyone has code that brute-forces all typeable passwords and speed tests it against using a dictionary?
I have to agree the bus is limiting factor. However still better than the alternative. This was done using openCL and not CUDA as I'm having problems getting CUDA to compile after I updated to gcc4.6.
The test you mention.. You mean something like john piped into pyrit VS pyrit reading a text file? I think I know where you are going with this though and I agree also. More people are becomming aware of password security every day and as such the dictionary isn't as useful as it used to be and brute forcing is going to be the way of the future.
What I mean is, suppose you take the possible password space...give a character limit. Divide by the number of GPUs in the box/farm to get starting points. Pass in only the starting point \x00 for example. Then within OpenCL go to the first typable character set after that. Get number of workers. Find the next typable character as a starting point. workerID+starting-point is the seed. run an iteration. check to see if next iteration starting point is typable...if so run iteration. if not go to next typable starting point. run iteration. make sense?
Think, when GPU farms are faster than dictionaries and it's fire and forget. With the right GPU farm, I think it could even be faster than precomputation lookups.
Thats brilliant... Unfortunately coding time is non-existent right now as I'm trying to launch a business which is absolutely consuming me. But I like where your head's @.
Thanks...now if only govt entity X wanted to fund a software project for GPU farm crypto bruting...
I'd suggest using oclhashcat-plus instead of pyrit especially if you have a high end AMD GPU. To fully utilize the power of the GPU it's mandatory to keep the GPU busy with work and that can be done by doing a rule based attack. I'll post some links that do a better job of understanding this. Another reason why oclhct+ is better than pyrit is because it doesn't require a CPU core for each GPU and maskprocessor is a lot faster than using pyrit + crunch.
http://hashcat.net/wiki/brute_force_in_oclhashcat_plus
http://hashcat.net/wiki/rules_with_maskprocessor
http://hashcat.net/wiki/cracking_wpawpa2
http://hashcat.net/wiki/rule_based_attack