NTLM-Multiforcer-Windows-0.51

[the_drag0n]

there we go :

C:\Documents and Settings\Administrator\Desktop\NTLM-Multiforcer-Windows>NTLM-Multiforcer.exe -c ./charset.txt -f ./ntlm.txt --min=8 --max=8 -v
Cryptohaze.com NTLM Multi-hash brute forcer by Bitweasil
Version 0.51 beta, length 0-14
Verbose mode enabled.
CUDA Device Information:
Device 0: "GeForce 9600 GT"
Number of cores: 64
Clock rate: 1.60 GHz

Target execution time: 1162039122 ms
Charset loaded (36 characters)
Charset: abcdefghijlkmnopqrstuvwxyz1234567890
Hashes loaded (152 hashes)
Please wait... finding optimum block count.
Testing 384 threads, 0 blocks
Null work test done: Reference time 0.531 ms
100 step test done: Reference time 10.757 ms
Per step adjusted to -2082016607
-2082016607 step test done: Reference time 9760.635 ms, rate 1392980.0M/s
Testing 384 threads, 0 blocks
Cuda error: the launch timed out and was terminated.

[Bitweasil]

Awesome. That gives me useful info to debug with. I can fix that.

[lemon]

Here's my result:

NTLM-Multiforcer.exe -ccharsetfile -fNTLM.txt --min=6 --max=6 -v
Cryptohaze.com NTLM Multi-hash brute forcer by Bitweasil
Version 0.51 beta, length 0-14
Verbose mode enabled.
CUDA Device Information:
Device 0: "GeForce 8600 GT"
Number of cores: 32
Clock rate: 1.51 GHz

Target execution time: 1129532718 ms
Charset loaded (63 characters)
Charset: ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijlkmnopqrstuvwxyz1234567890
Hashes loaded (5 hashes)
Cuda error: invalid device symbol.

CUDA Drivers seems OK, because it works with other GPU crackers.

[Bitweasil]

Thanks. VC2008 wasn't doing some implicit type conversions that gcc does, and this makes math go very wrong, which leads to (in some cases) crashes, or bad memory allocation.

I'm working on patching this tonight.