Hi Guys,

Most of you have started or atleas once thought of generating Rainbow tables. But some of us might have dropped the generation of tables in between bcoz of time and system considerations.

Here is an idea, instead of individually generating rainbow tables lets starta project. A Distributed table generation Project.

One person will moderate, people join togather and generate parts of tables and in the end the moderator will join all the tables and we can have a DVD of Rainbow tables written for us.

well i have a 2gb alphanumeric LM hashes....but at my work place though

how about this idea, if many of us are intrested in this very soon we can get huge tables generated for us.

- Onlinepass

First of all, the tables will take more than a mere dvd size if you want them to be good. I assume that you want Alpha-Numeric lower/upper case tables. That would take several hundred gbs. Which means somebody would have to host that.

A DVDs worth of rainbow tables (4gb) will prolly only get you numbers...not much cracking power if all you can do is numbers.

i have an 1.5 gig rainbow cracker and it cracks letters+numbers.

just modify rainbow a bit to let it be good

That would be cool, its hard to use the big groups rainbow crack tables. You have to give to receive...

I have an Alpha-Numeric LM. Not too hard, it took like 2 or so days with 2 AMD 2800's and an AMD 1800.

I am very interested in this project. It would be nice to have a Alpha-Numeric-Symbol for each of the hashs supported by RainbowCrack. Maybe we could also have some Alpha-Numeric tables, much smaller, thus easer to download.

I am willing to host this project on a VERY fast server. I am thinking people can upload it, then I setup a BT tracker and the server also shares the file.

PM me if you are all interested.

I have a question..according to that rainbowcalc tool..rainbow chain length has a very high effect on the probability of success without impacting table size at all.

For example:

1 table of 2400 chain length results in 610 mb and 60.55% success:
rtgen md5 loweralpha-numeric 1 7 0 2400 40000000 all

On the other hand watch this:

1 table of 100000 chain length results in 610 mb and 99.77% success:
rtgen md5 loweralpha-numeric 1 7 0 100000 40000000 all

Now my question is..is that a bug in the rainbow calc program or will higher chain lenghts result in higher success while at the same time retaining the same size. Are there any adverse effects with such a large chain length. Seems too good to be true...

it does seem to good to be true, but there is 1 thing u missed - TIME!
it takes longer to generate larger chains
for a brief explanation (which was to the best of my knowledge) read the info on rainbowcalc values on my site, it will tell u what increases certain things like file size, time and success probability

SecureIt.Co.Il has already initiated a "RainbowCrack Project", which aims to generate the RainbowCrack "all" characterset, by dividing CPU load amongst members.

for more info : SecureIT.Co.Il

you have to work the time out yourself READ the article on the site again

talking about the success rate and size
something doesn't work out for rainbowcalc .... have read the site 3 times, maybe I (filtered) up or misunderstood something but look at this.

If you build 1 Rainbowtable with chain count of 40000000 successrate is 99,44%
If you build 4 tables with chain count of 10000000 successrate is 99,99%

same total table size, same generation time. both completed tables will contain the same charcombination when done so they should have same successrate, or am I wrong?

you have to modify the parameters a bit to get optimum performance (PiP seems to have had some success with this)

as of yet i have not added a feature to rainbowcalc to optimise rainbowcrack parameters due to a lack of time
the matlab parameter conguration scripts on the rainbowcrack site can be tested with a free program called scilab

wow nice program, free and open source. and I wasted all my time with 100 of other tools to get my math functions working(mathlab,labview,mathcad...doh).

I own you one, thanks

ive done some tests and vs using longer chain lengths vs more overall chains, the cpu time is roughly equivalent for me at least. for a particular % chance of working, using longer chains vs more chains, very long chains only take me 25% longer to generate than more chains, and using significantly less disk space. I would say we can probably generate a 7 character length upper/lower/numeric md5 set with a good success rate in 40-50 gigs. Computation time is about 2ghz-years. So, 200ghz + high bandwidth server = 4 days generation time

I was thinking that we could set up a server to search the tables online, and report the password back to the users. It couldn't be done with PHP, because mysql would not be able to handle that much information. I was thinking about downloading xeneo server 2.2 and using it's ASP functionality, but it costs 100 dollars, and I spent most of my money on 2 120 gig drives. I need to get more processing power, but other than that I'm set. I'm generating the tables on a different computer, so after that's done I'll copy them back. I have about 300 gigs of open space (I also have the root drive which is 80 gigs), so I deffinitly have the space, just not the power, yet. I'll see what I can do with setting up this idea in the meantime.

Has anyone found a resolution to this as far as MD5 goes? I think whitehat.co.il is a pay site now..

take a look here =]

http://www.governmentsecurity.org/forum/in...opic=9702&st=15