Encrypted bcrypt - Protect your hashes from a SQL injection only attack.

Found a bug in the SWAP_ENDIAN function in SHA1's SSSE3 version. This shouldn't affect the speed. Download page.

Moved the LHT server to a Linux box and I finally got off my ass and finished the SSE2 code (because I lost the original code and never compiled it for Linux).

ARTGen 0.1a
This just does MD5 for CUDA (no hybrid) and CPU (hybrid/hybrid3). Max length is 15 characters for CUDA. A GTS 450 averaged 273 MLinks/second when generating md5_mixalpha-numeric#1-9_10000x1000000.

XSHA1 Reverser 1.1
Version 1.1 is out and has a huge speed increase almost 7 times faster than version 1.0.1. Also I basically removed the rate limit for the online version.

XSHA1 Reverser 1.0.1
Found and fixed two bugs while making it around 4 times faster (not done yet). This is just a bug fix release. Minor bug with -f|--output-not-found and half of all 13-16 character passwords (-c|--characters=16) and half of all 17-20 character passwords (-c|--characters=20) were not found because of copy and paste error with the mask in functions finishRev16() and finishRev20() in xsha1help.cpp.

XSHA1 Reverser 1.0
Reverses XSHA1 hashes of up to 16 character passwords in 0.1 second and up to 20 characters in 3.3 seconds. Now with friendly arguments. For people using Linux, if you want to use AVX or XOP you will need to compile it with a new-ish compiler like GCC 4.6.1. I only have GCC 4.4.4 :(.

XSHA1 Reverser 0.1
Here's the program and code for reversing XSHA1 hashes of up to 16 character passwords in 0.1 second and up to 20 characters in 10.8 seconds (right that's why I don't have benchmarks using SSE2 the function xsha1Reverse_20_sse2() has a bug that I never fixed. So don't modify the code to use that function instead use xsha1_rev_20() since that works but is slower).

Beta
I've been working on DP stuff and I thought I'd share some results. It's going to be an interesting week... Hopefully. Also I moved the old beta (Rainbow table benchmark) to tools if you were wondering where that went.

Advanced RT Calc
Added step generation, optimized for imperfect tables for IRT, support for imperfect tables for IPRT (I should really change that name since P stands for perfect... "DIRT" Destructive Indexed RT [maybe]), and calculations for the expected time spent checking for false alarms. Also fixed a few minor bugs.