Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 25 Jun 2015 17:09:16 -0400
From: Matt Weir <>
Subject: Re: precomputed attacks for john: rainbow tables and other ways

I'd recommend checking out the following paper since it talks about some of
the trade-offs involved with various TMTO approaches:

I'll admit I'm not a huge fan of Rainbow Tables, or I should say I'm
skeptical of their general usefulness outside of some specialized use
cases. That being said, I'm more than happy to be proven wrong!

Side note, to back up what Aleksey said, Rainbow Tables, (or Hellman
Tables), don't have to be charset based. As long as you have a fast
indexing function you can create dictionary based rainbow tables too. The
method described by Aleksy is very much like the PRINCE attack. If you want
to look at some of the issues with other rule based rainbow tables you
might want to check out Chapter 4 of my dissertation where I did some
research on the subject.

The biggest issue is by going to a dictionary based approach you can
introduce collisions in the index_to_plain function which isn't a problem
in character based RTs. Collisions in Rainbow Tables are much worse than
collisions in a normal cracking session so it helps to put extra thought in
how to avoid those.

Good luck!

On Thu, Jun 25, 2015 at 2:26 PM, Alain Espinosa <> wrote:

> -------- Original message --------
> From: Aleksey Cherepanov <>
> Date:06/25/2015 12:59 PM (GMT-05:00)
> To:
> Cc:
> Subject: [john-dev] precomputed attacks for john: rainbow tables and other
> ways
> ...Ideas?
> Some weeks ago I research a little about rainbow tables. I need to
> research more. I will probably add support for Hash Suite in the near
> future. Some comments:
> 1- There are a lot of scientific research related to time - memory trade
> off, the majority unused. We need first to take a look at it and then
> develop new ideas. For example there are claims that Hellman tables with
> Distinguished Points are better than Rainbow tables. We can access papers
> from: . I recommend for example "Variants of the
> Distinguished Point Method for Cryptanalytic Time Memory Trade-offs", 2008,
> Jin Hong, Kyung Chul Jeong and others.
> 2- 2^31 is too small. In a high end GPU we can exhaust the key space in
> 0.14 second for NTLM hashes or 0.25 for MD5. Yes, less than one second.
> 3- One thing worth investigating is mix rainbow tables with John
> incremental or Markov mode, so we had a small rainbow table with the more
> probable candidates. We need to ensure the probability of repetitions
> remains low, but this is interesting, particularly for high password
> lengths where full rainbow tables are to big.
> Regards,
> Alain

Content of type "text/html" skipped

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.