1 7 = lengt of pasword the 0 = l = rainbow table count : is the table you are on, but how to calculate how many you need? 5700 = t = rainbow chain length. but again, how to calculate...
The command line output and the documentation clearly show what to do. Or you are going to use other algorithms than e.g. MD5? Then you would need to recompile the whole thing and insert the specific algorithm.
The command line output and the documentation clearly show what to do. Or you are going to use other algorithms than e.g. MD5? Then you would need to recompile the whole thing and insert the specific algorithm.
This is an unfair statement. The algorithm is quite complex and I doubt that most people fully understand it. If you know it, than please explain, cuz I would like to know as well.
Sure, I can go to the site and type in their samples, but I'd like to have a rainbow table for alpha-numeric-symbol up to 12 characters in length. I already have the simple example, and its great for security audits and the like, but it would be nice to have a full blown version for hacking attacks. The commands are simple, but determining how many rainbow tables and rainbow chains you need is not simple. At least from what I saw. A good question!
nulladd
Jan 19 2004, 05:26 AM
SyN/AcK and FiNaLBeTa try this tool, it will tell u how many rainbow tables are required for a good success rate, just enter in the required values and some default ones are provided just to give u the idea of how it works http://members.lycos.co.uk/nulladd/tools/RainbowCalc.exe right click->save target as (cause lycos is silly)
also SyN/AcK
QUOTE
alpha-numeric-symbol up to 12 characters in length
that will take a long time
NOTE: alternativly if u still dont understand or cant use my program then u can post the number of characters and charset to be used and i will make a .bat file for u to use
FiNaLBeTa
Jan 19 2004, 01:10 PM
Thnx for the tool. but the problem stands, the tool dos not help me on what the chain lenght ore count chould be.
In youre app , the chainlenght seems to have no effect on the file size ore generation time, but onely on the crack chanses.
I need the commands, file size, and estimated time for (yep, i know it's a big one) lowandupalpha-numeric = [ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789]
62 chars = key space of 3579345993194
I've been playing around whit youre program. And i found that when i use a chain lenght of 24000 and a chain count of 40000000, that i reach a 97% chance with 57 tables (16weeks 10hrs , 34,79 GB (610 MB per table))
Can i trust this info to be correct? If so, you've helped me alot. thank you.
//edit, how do i find the time to calculate 1 chain?
nulladd
Jan 19 2004, 02:31 PM
QUOTE
the chainlength seems to have no effect on the file size or generation time, but only on the crack chances
it affects the generation time (see below) and the crack chances, but not the file size
Because rainbowcalc cannot deternmine exactly how fast your computer is, you must do it yourself NOTE: i recommend that you close most programs whilst this is happening, otherwise you may slow your benchmark results.
Step 1. To get the minutes and seconds just run your desired configuration with rtgen.exe
Step 2. let that run for a while until you see something like this: (the time varies depending on chain length and speed of your computer, this example was done with a chain length of 5700 with the md5 hash (lm will be a bit quicker))
CODE
1 of 40000000 rainbow chains generated (10 m 4 s)
where m is minutes and s is seconds
Step 2. In rainbowcalc where it says "Time to Generate 1 Chain" you need to enter in the minutes and seconds obtained from Step 2.
IMPORTANT see how you have the value 24000 (the chain length), basically it takes about 4 minutes with a value of 2400 and about 10 minutes with 5700 on a moderately fast PC (when using md5, so lm will be a lilttle bit quicker), so i can only guess that for the amount of characters you want to do, this will take a very, very long time.
CODE
lm hash speed: 1402918 / s lm step speed: 821287 / s
also this is the speed ive been able to get rtgen to do, so your computer will take longer than 4 minutes with a chain length of 2400, and as from above a very long time with a chain length of 24000
i recommend you rethink how many characters you are to include in this, as a rough calculation it would take several months on my computer, which is faster than yours according to your benchmark
but if you want i have included a bat file for u anyway, it will calculate your config to a success rate of about 99% (rounded up)
temptation
Jan 19 2004, 03:11 PM
I did
CODE
rtgen alpha-numeric 0 2400 40000000 bla rtgen alpha-numeric 1 2400 40000000 bla rtgen alpha-numeric 2 2400 40000000 bla rtgen alpha-numeric 3 2400 40000000 bla rtgen alpha-numeric 4 2400 40000000 bla
And it worked for heaps of pw ... Its from A-Z and 0-9 ...
Ok, it would be better if there is a "." (dot) in it but it doesn't matter ... It is not perfekt, buts ok, for weak pw ...
then to make up for the extra keyspace i imagine youd only need to generate 1 or 2 more tables
FiNaLBeTa
Jan 19 2004, 05:42 PM
I know it's ambitious , but it can be donne, i tested it on a 900Mhz cpu. But in reality it will be executed by cpu's > 2Ghz (ofcource multiple ones) Now just to hope that you're program didn't gave an overflow anywhere cuz when i test it with 64 chars, youre program gave me an 99% sucsess rate after onely 10 tables :/.
QUOTE
Rainbow Charset = lowandupalpha-numeric : [ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789] Chain lenght = 24000 Chain Count = 44000000 Table Size = 671Mb Time to generate 1 Chain on a 900Mhz CPU = 111Min 12Sec Time needed for 1 table = 4weeks 5days 23hrs 28mins Sucsess wanted : 98% in 58 tables = 38,94 GB - 281weeks 3days 17hrs 4mins
50% after 10 tables (48 Weeks) 60% after 13 70% after 17 (82 Weeks) 80% after 23 90% after 33 (160 Weeks) 95% after 43 99% after 67 (330 Weeks 45 Gigs)
Needed commands =
rtgen lm lowandupalpha-numeric 1 7 0 24000 44000000 all ... rtgen lm lowandupalpha-numeric 1 7 67 24000 44000000 all
SyN/AcK
Jan 19 2004, 06:12 PM
Thanks guys, this will help alot!
vnet576
Jan 19 2004, 06:16 PM
Hmm..this stuff is kinda confusing, but lets say we wanna add all 32 symbolic chars and 1 space char to the mix. How would that change the situation.
FiNaLBeTa
Jan 19 2004, 06:53 PM
QUOTE (vnet576 @ Jan 19 2004, 06:16 PM)
Hmm..this stuff is kinda confusing, but lets say we wanna add all 32 symbolic chars and 1 space char to the mix. How would that change the situation.
For 33 chars you have a 99% chance of breaking the pasword when you take 3 tables
Plaintext lenght = 7 chain lenght = 2400 ( normal) rainbow chain count = 44000000 ( 671Mb so you can backup the tables on cd ) total = 2.6 gigs I think it would take about a week and a couple days to generate tables (900Mhz cpu)
Plaintext lenght = 8 You would need 69 tables here with same config
If you take my chain lenght = 24000 you would onely need 13 tables for 99% chance. But ofcource it would take much longer per table
in total it would be : (63weeks 7hrs 32mins )
//edit, message edited, some data was wrong. PS, all info comes from nullads tool. nice thingy
vnet576
Jan 19 2004, 07:12 PM
The thing is I'm having trouble with his utility...I don't want to do 33 chars..I'm thinking alot higher 94 chars (alphanumeric, upper/lower, symbols, space). Anyway, with a chain length of 2400, and plain text of 7, I'm getting 5000 tables just to get 63 %. Can it possibly be that high or am I doing something wrong. I tried to go higher, 10000 tables but my pc couldn't handle calculating that and it crashed.
FiNaLBeTa
Jan 19 2004, 07:27 PM
QUOTE (vnet576 @ Jan 19 2004, 07:12 PM)
The thing is I'm having trouble with his utility...I don't want to do 33 chars..I'm thinking alot higher 94 chars (alphanumeric, upper/lower, symbols, space). Anyway, with a chain length of 2400, and plain text of 7, I'm getting 5000 tables just to get 63 %. Can it possibly be that high or am I doing something wrong. I tried to go higher, 10000 tables but my pc couldn't handle calculating that and it crashed.
wel, read my example, that one is for 62 chars. going any higher just might be impossible.
calculate you're wishes with not with 2400 but with something much higher. like 10000 , but do note that generating tables will take ages,...
Read my example above please for Charset = lowandupalpha-numeric : [ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789
nulladd
Jan 20 2004, 01:03 AM
the success proababilty function it is ported from the matlab script on the rainbowcrack site, yes it will crash at incredibly large values (if this happens then the tables you are trying to make will take many years) which is a limitation of vb's number handling, however i intend work around this and eventually write a new program in c++ (hey PiP if ur readin this i need the code for a gui) but for now i'll add support for split tables (ie: using #0, #1, etc), some error checking and a tutorial (probly put that on my site)
QUOTE
10000 tables but my pc couldn't handle calculating that and it crashed
rainbowcalc was just taking a long time because i had not assumed that anyone would generate that many tables, nor do i suggest that you generate that many, instead try FiNaLBeTa's idea of increasing the chain length
vnet576
Jan 20 2004, 01:39 AM
As I found out the chainlenght has diminishing results as u go higher. At first it helps alot but then when u get past 1000 tables and the probability of success is still 50%<, increasing the chain length will not help anymore.
nulladd
Jan 29 2004, 01:38 PM
i dont mean to bump this thread but there was an updated matlab script on the rainbowcrack site which fixed a minor bug encountered during the calculation of the success rate
so ive updated rainbowcalc.exe and put it on my site so trial members can have access to it, i recommend that users of the old rainbowcalc get the newer version for better success rate accuracy
Thnx for the tool. but the problem stands, the tool dos not help me on what the chain lenght ore count chould be.
In youre app , the chainlenght seems to have no effect on the file size ore generation time, but onely on the crack chanses.
I need the commands, file size, and estimated time for (yep, i know it's a big one) lowandupalpha-numeric = [ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789]
62 chars = key space of 3579345993194
I've been playing around whit youre program. And i found that when i use a chain lenght of 24000 and a chain count of 40000000, that i reach a 97% chance with 57 tables (16weeks 10hrs , 34,79 GB (610 MB per table))
Can i trust this info to be correct? If so, you've helped me alot. thank you.
//edit, how do i find the time to calculate 1 chain?
I am not 100% sure, but isn't it right that all passes are converted to uppercase anyways, so there would be no need for lowercase? Thought I read that somewhere in the RC doc.
nulladd
Jan 29 2004, 02:08 PM
yes that is correct for lm hashes
FiNaLBeTa
Jan 29 2004, 02:24 PM
QUOTE (Jeeve5 @ Jan 29 2004, 02:05 PM)
QUOTE (FiNaLBeTa @ Jan 19 2004, 01:10 PM)
Thnx for the tool. but the problem stands, the tool dos not help me on what the chain lenght ore count chould be.
In youre app , the chainlenght seems to have no effect on the file size ore generation time, but onely on the crack chanses.
I need the commands, file size, and estimated time for (yep, i know it's a big one) lowandupalpha-numeric = [ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789]
62 chars = key space of 3579345993194
I've been playing around whit youre program. And i found that when i use a chain lenght of 24000 and a chain count of 40000000, that i reach a 97% chance with 57 tables (16weeks 10hrs , 34,79 GB (610 MB per table))
Can i trust this info to be correct? If so, you've helped me alot. thank you.
//edit, how do i find the time to calculate 1 chain?
I am not 100% sure, but isn't it right that all passes are converted to uppercase anyways, so there would be no need for lowercase? Thought I read that somewhere in the RC doc.
You have got to be kidding me. haha No, thnx for saying this, I already used up several weeks of cpu time. Gotta start over again.
Thnx for this info.
FiNaLBeTa
Jan 29 2004, 08:10 PM
QUOTE (nulladd @ Jan 29 2004, 02:08 PM)
yes that is correct for lm hashes
I pmed you, dont know if you ignored it ore something, but i'll trye once more here.
The new rainbowcaltool gives totally difrend result then the first. The first looked like it worked logaritmic, this one dossend, somethimes the result don't even change when you change parameters, like number of chars. Can you check this?
nulladd
Jan 30 2004, 01:57 AM
sorry about the late response, i think invision was a bit shy on informing me of PMs
the update was based on the new matlab script from the rainbowcrack which supposedly increases the accuracy, however things still seems to be under development but this matlab script is the best ive seen so far and works well for me. yet there stil may be bugs to iron out With the different keyspace issue i am looking to see whether different mulitples of certain numbers give better results even if they are higher, tho i may be barking up the wrong tree.
having said that i would like all values you have entered on each version so i can check this and try and fix any errors. heres some useful data that u can provide (say which version of rainbowcalc) keyspace rainbow chain length rainbow chain count number of tables the success rate
FiNaLBeTa
Jan 30 2004, 06:56 AM
QUOTE (nulladd @ Jan 30 2004, 01:57 AM)
sorry about the late response, i think invision was a bit shy on informing me of PMs
the update was based on the new matlab script from the rainbowcrack which supposedly increases the accuracy, however things still seems to be under development but this matlab script is the best ive seen so far and works well for me. yet there stil may be bugs to iron out With the different keyspace issue i am looking to see whether different mulitples of certain numbers give better results even if they are higher, tho i may be barking up the wrong tree.
having said that i would like all values you have entered on each version so i can check this and try and fix any errors. heres some useful data that u can provide (say which version of rainbowcalc) keyspace rainbow chain length rainbow chain count number of tables the success rate
I kindoff imediatly saved the new app over the old, so if you can hand me the old one again. I'll do that.
nulladd
Jan 30 2004, 02:19 PM
QUOTE (FiNaLBeTa @ Jan 30 2004, 05:56 PM)
I kindoff imediatly saved the new app over the old, so if you can hand me the old one again. I'll do that.
There is a problem in script "calc_success_probability.m". When key space is small, the calc_success_probability.m script works well. However, the formula in the script is not in a good form and the error accumulates so much when key space is large.
so at the moment i can assume that the updated version is more accurate, however i will do a lot of testing of my own code to make sure it is ok and will post my results here
GhostCow
Jan 31 2004, 10:04 AM
i got a question: if i have regular rainbowcrack v 1.2 for windows, which version of rainbowcalc is better and more accurate for me? i understand that the newer version has been updated based on the matlab scripts, so finalbeta: why did you ask for the older version?
FiNaLBeTa
Jan 31 2004, 10:28 AM
QUOTE
if i have regular rainbowcrack v 1.2 for windows, which version of rainbowcalc is better and more accurate for me? i understand that the newer version has been updated based on the matlab scripts, so finalbeta: why did you ask for the older version?
Because something is wrong.
If the new app is true, it would mean that all rainbowproject in the past are worthless. They would have less then half of the tables they need. Diskspace will olso get absurtly high.
I hope there is a mistake somewhere. The new formulla dossen't look like a bugfix. The first one was logaritmic ore simular, the new one :/
I'm stopping my project. Cuz i can nolonger trust the results, i'm not just gonna waste cpu cycles and HD space.
nulladd
Jan 31 2004, 01:47 PM
QUOTE (FiNaLBeTa @ Jan 31 2004, 09:28 PM)
I hope there is a mistake somewhere. The new formulla dossen't look like a bugfix.
i would hope there is a mistake too. with smaller charsets either version would be fine, athough as it states on the rainbowcrack site that the newer version should be more accurate
FiNaLBeTa - thanks for ur input, its a pity about your project but i guess we are still in the development stage here. hopefully it will get sorted, just needs more work and thinking
i am currently downloading the free program scilab so i can run a full test of rainbowcalc and scilab results (from the matlab scripts). i will post results here
The command line output and the documentation clearly show what to do. Or you are going to use other algorithms than e.g. MD5? Then you would need to recompile the whole thing and insert the specific algorithm.
This is an unfair statement. The algorithm is quite complex and I doubt that most people fully understand it. If you know it, than please explain, cuz I would like to know as well.
Sure, I can go to the site and type in their samples, but I'd like to have a rainbow table for alpha-numeric-symbol up to 12 characters in length. I already have the simple example, and its great for security audits and the like, but it would be nice to have a full blown version for hacking attacks. The commands are simple, but determining how many rainbow tables and rainbow chains you need is not simple. At least from what I saw. A good question!
Actually it's a matter of a few calculations; very simple read the paper on time-memory cryptoanalytic trade off to acquire the formulas
maxmouse
Feb 1 2004, 04:36 AM
First of all I would like to thank everyone who posted all this information about rainbowcrack the program makes a lot more sense now. I'm wondering if anyone has a complete rainbowcrack table for all printable characters and perhaps ASCII. I have an LM hash D967ED15958F0B997CA65F36030673DD. If anyone that has a table could find the plain text value I would be greatful.
SWEET THANKS! many many many thanks fly out to aberrant gzzah drccjsp fat raph BrianR numatrix for helping out lazy ass fools like me ))))
DiJiTooL
Jul 20 2004, 09:21 AM
Hi, If i use a 90000 as chain lenght, the time to create table will be very long.. but the max cryptanalysis time will be long too or it same for all configurations?? thanks for your response
nuorder
Jul 20 2004, 01:36 PM
QUOTE
Hi, If i use a 90000 as chain lenght, the time to create table will be very long.. but the max cryptanalysis time will be long too or it same for all configurations?? thanks for your response
shouldnt make too much difference
ZeroQool
Sep 10 2004, 03:39 PM
the download link is dead what a pity could you reload it plz?! thx
yep looks good you could save around 6GB and bit of precompution time by using 50000000 for example instead of 67108864 at the loss of only 0.3% success probability
Alex[RM-UK]
Sep 26 2004, 10:52 AM
Hi
I am new to Rainbow Tables And I am very stuck!
I've read the tutorial on there site about 10 times and yet I still do not understand a word they are saying.
Can someone write or find a tutorial on it please and how to set up tables and chain count and length, I just don't understand.
Thanks
nuorder
Sep 26 2004, 01:05 PM
is there a problem with using the default config, what sort of tables do you want to generate?
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
, but it can be donne, i tested it on a 900Mhz cpu. 
