I know Pyrit has support for clustering machines over a local network. Just listen on port 19935 after a serve command and your set. This is great if you got like 10 machines at home with a bunch of CUDA cards. Thing is that I don’t have a bunch of Nvida boxes at home.
What I do have is 100/100 internet and one CUDA card that is very fast. It would be nice to see this working as a distributed internet cluster. Before people say “hell no, his card sucks” think about this. Some people are willing to leave their computers on for days at a time. A system could be set up that the TOTAL amount of “computing power” gives you a certain amount of points.
You login a week later after your week long trip to Uganda and see what your point count is. Maybe a 4:1 ratio would be good? You could use these points for your own endeavors. Some criteria would be needed. A fast internet connection, at least a semi fast card and an account. Good tables could also be made by seeing patterns in essid and the broken keys. More importantly optimal pass phrase lists too! Maybe I missed something? Is this even possible?
All the best,
Its a great idea and pure_hate and i had a very brief talk about this, the problem is apparently the data "wordlist" to send and distribute is huge when a new node is added.
I already had the idea that each user is responsible for his own section of the wordlist but the problem then lies what if that user drops for whatever reason, that means his section of the wordlist is left out of the crack.
If we can ever get wordlists HIGHLY compressed this would be a great idea ... atleast something to throw at the pyrit guys to mull over anyways.
@onryo-- Hey man, how's it going?
These are good ideas, but it's clear that pyrit is a little ways off in supporting the global peer-2-peer cracking network you envision. Yes, it's very b/w intensive. Seems like it's possible when the cluster is co-resident.
Question back at you guys: Have you installed the latest pyrit svn on your boxes? I managed to get r120 installed and working with just minor debugging. The SSE2 support adds only about 1.1% in my tests, however.
Since I have one box, I cannot even attempt to exercise the 'serve' funtionality in pyrit, but man it's tempting.
Edgan and I have done some testing with the cluster. he has 2 box's with gpus and I have 2 boxes with gpu's. The problem , like I have already stated is the BandWidth.
On top of that when Edgan and I did it we lost quite a bit of power in the transaction, almost 33% and use a a$$load of bandwith.
Most people do not have 100 mb connection with unlimited bandwidth that they would be willing to donate to the project.
Me and KMDave were thinking about it and figured a central server would hold "the list" and then send out work units to connected clients. It wouldnt be hard but like I said there in the bandwidth issue. Im sure this will eventually come into play but not right now.
Where has the serve command gone? Is it out for lunch?
I would relly like to get those 157 PMKs my old laptop can provide...:)
I believe pyrit's author removed the feature hhaving decided it was not ready for prime-time. Check the projects's googlecode homepage and/or blog for details.
Ohh, im sorry too hear that..
Here in Dk, we all have >2 mbit, and many have 100/100 mbit..
I also have 3(soon 4) pc in my apartment..
Bring the power back..
bandwidth should be symmetric for this scenario to work, most residential networks today are asymmetric.