aycwbatu - CSA bruteforcing tool

[servese43]

Hi,

Came across this CSA bruteforcing tool earlier and was wondering what you all make of it. It is open source and runs on both windows and linux. It can be found here:

https://github.com/malvanos/aycwabtu

This is a fork of the original version, and development has stalled since mid-2019. Have any of you tried out this software before? Looking through this forum, it seems that there are some CUDA-based tools that give a much higher Mcw/s speed.

 

[marrakr]

Yes, but it's very very slow in compare to CW Finder.
First you need to spend days or weeks to optimize it.
Sadly CW Finder owner does not want to publish its source code.

 

[campag5242]

CW Finder is not comparable: it searches a text file containing old keys, plus new keys based on previously observed patterns eg embedded dates etc.

Of course, it's best to try that tool first: as you have observed, it's quick. It makes sense to try old keys & often used patterns first. But that speed is simply a sign that it is searching only a tiny fraction of the 2^48 possibilities (the keyspace). Even if your key file has 100 million entries (all your best guesses), that's a tiny, tiny fraction of the 281,474,976,710,656 possible.

aycwbatu is different: it grinds its way through all those 281,474,976,710,656 possible keys. If rainbow tables aren't an option, that's the only published method to retrieve the key if it's entirely random.

@servese43: If you are going to search the complete keyspace, it appears FPGA > GPU > CPU. eg KT2SET's FPGA searches the keyspace in around 12 hours, so 6500 MCW/s. My own CUDA project on a 1070TI gets around 710MCW/s, or 4.5 days, & maybe ~1 day on the most powerful 30-series nVidia GPUs. I still hope to accelerate that further.

Full respect to the effort put in to these CPU based efforts, but after much effort optimising, 40MCW/s was reported on a 12-core 3.6GHz processor: using CPU is clearly not the way to go.

 

[servese43]

Nice! 6500MCW/s! Using the FPGA program only seems like a good option if you're really dedicated to this kind of thing and do it lot though. I am interested in CUDA project you've done because I'm running on a 1050ti myself. If you haven't already published it, could I please have a look at it?

Thanks, servese43

 

[Plasma70]

For the biss keys not found via video.ts \ crypt8 procedure, I use Cudabiss with my video card (Geforce Rtx 2070) and it has always found them all in a maximum time of 100 hours. Do you know any methods faster than Cudabiss? Thanks