AZdecrypt BETA 0.94

[Jarlve]

Update 04/06/2019:



AZdecrypt is a fast multi-threaded homophonic substitution cipher solver with a Windows GUI.

The latest AZdecrypt can be found here: http://www.zodiackiller.net/viewtopic.php?f=81&t=3198

[Jarlve]

Fixed an issue where I was not ending the threads properly leading to increased memory and cpu build up when processing many ciphers.

[Jarlve]

Recompiled with optimizations resulting in a big speed increase. Keys checked per second is now in the millions for my system.

[Quicktrader]

Great approach, checking it out asap,

QT

[Jarlve]

Thanks Quicktrader,

I'm still working on it, let me know if anything.

I just released a new version which has a big change: each thread now handles it own cipher, in previous versions all threads worked together on one cipher. I've also done some minor optimizations here and there. And I retuned the hill climber for ciphers with a multiplicity (63/340) of that around the Zodiac 340.

The "time per cipher" setting has been replaced with "keys per cipher". I recommend that for serious use you put this at 20000000 (20 million). I can currently not guarantee score stability below this point, I'll see if I can get this down. This is about 5-10 seconds per cipher on my system (i7 930).

[Jarlve]

0.4

Nearly doubled the speed of the program. A serious bug was fixed that let some threads load the same cipher as the first thread, if you have been using this program I recommend retesting everything you did so far. Some problems considering file input and output are fixed.

I've been working on score stability as a top priority. If for example you score 100 copies of a cipher, I would like the difference between the minimum and maximum score for each copy to be within 1%. Currently this is more like 2-3% for 10 to 20 million keys per cipher. Score stability for lower amount of keys has been increased significantly, I find the program useable starting from 5 million keys per cipher for 340 like ciphers.

[doranchak]

Jarlve, I just wanted to say I haven't had time to review your program but I am excited to hear about your progress. Please keep up the wonderful work you are doing!

I will soon start experimenting with the automatic solvers again to help me explore some ideas. It is nice to have several different programs to pick from now!

[Jarlve]

Thank you doranchak,

0.5

Added a colorful UI with some basic options, plus/minus keys to increase/decrease threads, 0=pause. u/i keys to increase/decrease keys per cipher. It shows all kind of handy stats, pressing the x key will exit the program after letting the threads finish and generating a text log in the results directory + a screenshot of the window in the root. Fixed some minor problems here and there.

I have stress-tested the program up to 125.000 ciphers (30+ hours of computing) and nothing went wrong. I'll update the main post with some extra information about the UI as soon as I have some more time.

[Jarlve]

I also want to note that this program is designed to seperate the wheat from the chaff and to get the job done for many and not necessarily provide the best solve quality. Though it should mostly be more than adequate but ZKDecrypto will probably provide a higher quality, better readable plaintext.

Here is an example for the 408, notice that some letters (k,x,etc) are missing. And that the solve is not perfect.

isihehalsingpeopl
ebecauseitissomuc
hfunatiumoretunth
anhillingwasdgime
intheforrestbecau
semanisthemoatcan
gertueinamalotall
tohillsomethanggi
vesmethemoatthril
lingedperenceatis
evenbetterthinget
tingyourrochsoftw
ithagarsthebestpa
rtofitiathiewheni
caeiwillbereborni
nparidaceuncassth
eihavehissedwillb
ecomemyslavesiwil
snotgaveyoumyname
becauseyouwisltry
tosloicownorutopm
ycollectingotssav
esformyafterlitee
beorietemethhpati

[Mr lowe]

JarIve can you run the 340 through your system but first remove all of the first and last lines.
cheers