05-29-2012, 11:31 AM
Yikes !!! I get some sleep and return to this thread and it has moved on more than I could imagine !!
@Pixel and ntk you are doing some great work and have some good idea’s, I will have to answer briefly I’m afraid as there is too much to write about this and if I don’t get a reply in quick things will move on so fast I may as well not bother !!!
@Ntk
I understand your concerns about trying to predict what users would choose as a password but this thread was not considering “usersâ€, it has a more specific target machine chosen default SKY passwords. However I think the principle of a less brutal password list is a good one for other applications.
@Pixel
I personally believe so, I think we are making a huge assumption based on a few known keys. You made a good observation but it may be down to chance that the keys we have do this. If we don’t add this extra filtering then the lists may be more use for other things also.
Your calculations are very interesting, they seem a little more accurate than the ones I did when I first thought of this. It is a significant drop in cracking time and I hope atom adopts this idea as it will save days !!!
I am generating Z at the moment and it has been running for 14 hours or so, the output file is only 3.76MB !!!!!!!!
I don’t think SED is fast enough, this needs to be on GPU.
Another interesting thing to consider is I used ULM’s regular expressions to do some of this filtering on a text file and it does it much faster. I am purely guessing it would be done in 5 hours or so.
I think filtering an existing list may be faster than generating one then pushing it through SED.
I think none of us should seriously get into generating these lists yet until we are all happy that the filter is ok and we have heard from atom. Think about it, it takes many hours to generate possibly days, we will then have to upload / download all the lists and then find somewhere to store them. Unless you have a huge empty drive they will have to be stored zipped which will mean unzipping them every time you need a character then move on to the next.
This filtering is a sound idea and worth pursuing but not without GPU or being able to do it on the fly, without atom I believe my idea will die.
@Pixel and ntk you are doing some great work and have some good idea’s, I will have to answer briefly I’m afraid as there is too much to write about this and if I don’t get a reply in quick things will move on so fast I may as well not bother !!!
@Ntk
I understand your concerns about trying to predict what users would choose as a password but this thread was not considering “usersâ€, it has a more specific target machine chosen default SKY passwords. However I think the principle of a less brutal password list is a good one for other applications.
@Pixel
(05-29-2012, 09:00 AM)Pixel Wrote: So, how would sed delete all lines that don't have at least one or more duplicate character any where in the whole line?
Hash-IT what do you think to this. too much? you did say...
I personally believe so, I think we are making a huge assumption based on a few known keys. You made a good observation but it may be down to chance that the keys we have do this. If we don’t add this extra filtering then the lists may be more use for other things also.
Your calculations are very interesting, they seem a little more accurate than the ones I did when I first thought of this. It is a significant drop in cracking time and I hope atom adopts this idea as it will save days !!!
I am generating Z at the moment and it has been running for 14 hours or so, the output file is only 3.76MB !!!!!!!!
I don’t think SED is fast enough, this needs to be on GPU.
Another interesting thing to consider is I used ULM’s regular expressions to do some of this filtering on a text file and it does it much faster. I am purely guessing it would be done in 5 hours or so.
I think filtering an existing list may be faster than generating one then pushing it through SED.
I think none of us should seriously get into generating these lists yet until we are all happy that the filter is ok and we have heard from atom. Think about it, it takes many hours to generate possibly days, we will then have to upload / download all the lists and then find somewhere to store them. Unless you have a huge empty drive they will have to be stored zipped which will mean unzipping them every time you need a character then move on to the next.
This filtering is a sound idea and worth pursuing but not without GPU or being able to do it on the fly, without atom I believe my idea will die.