Skip to content

Latest commit

 

History

History
99 lines (62 loc) · 2.51 KB

README.md

File metadata and controls

99 lines (62 loc) · 2.51 KB
Raw

Wordlist Mask Generators Benchmarks

Results

As of writing this, Cracken is probably the world's fastest wordlist generator:

bechmarks results

Cracken has 25% increased performance over hashcat's fast maskprocessor (written in C).

Cracken can generate around 2 GiB/s per core.

Benchmark System Info:

  • Ubuntu 20.04 5.11.0-38-generic x86_64 Intel(R) Core(TM) i7-9850H CPU @ 2.60GHz
  • gcc 9.3.0
  • rustc 1.56.1
  • cracken v0.1.8
  • maskprocessor 0.73
  • crunch 3.6

Why Cracken is Fast?

Cracken uses multiple small optimizations but the most noticeable one is something i call a jump table.

A jump table is an array of 256 chars that maps each char to the next on the charset saving some memory access and a branch or two for every password generated.

jump table array for lower alpha looks like this:

idx val
0x0 0
0x1 0
... ...
'a' 'b'
'b' 'c'
... ...
'y' 'z'
'z' 'a'
... ...
0xff 0

whenever next_char <= char we reached the last char on the charset.

for more details see src/generators.rs:CharsetGenerator on the source.

Why Cracken is Fast (Hybrid Masks)?

Cracken can generate combinations of any wordlist charset and static chars (e.g. ?w1?w2?d?w1?l?u hello ?w3).

To make it fast Cracken does:

  • load wordlists into memory
  • sort wordlists by size and store them as big a buffer - Vec<u8> instead of Vec<Vec<u8>>. it is done to save memory (and thus cpu cache) and avoid another pointer indirection.
  • iterate wordlists by length - when same length no suffix copy needs to be done

for more details see src/generators.rs:WordlistGenerator on the source.

Other Generic Optimizations

  • fill the output buffer with '\n' chars once at start to save some mem writes
  • pre-compute wordlist batch size, simplify main loop if statement
  • use of stack pre-allocated buffers on main loop (no dynamic allocations on main loop)
  • tuning output buffer size

hope this info will help future implementations of wordlist generators.

Running Benchmarks

1. build from source

first build these from source:

2. run benchmarks

run the benchmarks script:

$ python run_bench.py

this will generate results.json file

3. plot the benchmarks

run create-bench-graphs.ipynb jupyter notebook