Cryptohaze.com

Linux, 64-bit:

http://www.cryptohaze.com/releases/Cryp ... 50.tar.bz2

Working on other OS releases, but that should work fine on 64-bit Linux. Who's up for some testing?

GRTCrack WORKS!!!!

OS: Ubuntu 10.04
GFX: Nvid GTX260 216SP

ran with:

Code: Select all

./GRTCrack -s 11111111222222223333333344444444 -h MD5 "/media/Main Storage/GRT/MD5-idx1-chr95-cl200000-perfect.grt" -t 512 -b 27 -m 500

I'm surprised it didn't take too long to scan through one index considering its 200k clength.



step rate was about 418M

One question, how come the above done% aren't at 100% and what are they for?

The run groups could be better documented...

For the candidate hash generation, I break it up into groups based on total number of threads. The run groups handle that. I... should probably just merge those into a single "completed" percent for the whole task. I'll make that change for the next release.

What was the total time per table you were seeing?

i think it was roughly 3mins

Thats running off a Green WD 640GB (the file should have been heavily fragmented) so i guess it could be better.

I have a few other questions
What are the specs of that MD5 Len7 table and how long did it take you to make?
Also if i'm generating tables i'm assuming same index = same seed or is there some math involved to calculate the next seed? and chain count / table number doesn't matter since i can just merge them all providing seed value is same and chain length is same?

I'm seeing some weird behaviour on Slackware64 13.1 box with 9600GT, latest nvidia cuda drivers.

Nothing works twice in a row - by that I mean that if I ask the system to generate a table and then to do it again, nothing comes up, generator hangs at "Creating table 1 of 1" forever, no completion display, no nothing. I have to issue some other command in between, clear or the like.

Same goes for cracking, only a little bit more drastic - it hangs after the first group is done in a similar manner to the above. By changing blocks/threads flags (not sure which made a difference, blocks I suppose) I managed to make the table into one group - no dice, group was done, but I got a segfault at 80% while "Searching table" was running.

How much RAM does your 9600 have?

My code is really designed for GT200 series cards or above - it *should* work on lower end stuff, but I've not tested it on anything below an 8800GTX.

I've seen the same "hanging on launch" issues at times, but it eventually clears up, so I've not worried about it. I assumed it was due to me hammering on the box with a ton of different code and fragmenting the video RAM or something like that.

Do you get the same behavior if you reboot, kill X11, and run the code over SSH?

blazer wrote:i think it was roughly 3mins

Still fast for MD5...

I have a few other questions
What are the specs of that MD5 Len7 table and how long did it take you to make?
Also if i'm generating tables i'm assuming same index = same seed or is there some math involved to calculate the next seed? and chain count / table number doesn't matter since i can just merge them all providing seed value is same and chain length is same?

I need to write the tutorial...

The MD5 len7 tables are 200k chain length, and took me a few days to generate.

You do NOT want to use the same seed value for different table parts, or you'll just generate the same part over and over again! Let it choose a random seed for the table gen.

The point behind seed values is to allow deterministic generation for the purposes of testing, verification, and distributed generation (assign the same table chunk to different systems and compare the results). You really shouldn't be using it for normal use.

Bitweasil wrote:How much RAM does your 9600 have?

512 MB

Bitweasil wrote:Do you get the same behavior if you reboot, kill X11, and run the code over SSH?

Now you're just talking technical gibberish, I'll try to answer it to the best of my knowledge: I never used it with X11 in the first place (that's what GUI is based on I think? just command line here), rebooting changes nothing, running code over ssh? Would that be to see what happens with command line execution or is there some other hidden benefit to SSHing loops (since it's my home computer)?

I forgot to mention another artifact, this time quite literal - display artifacts come out, thin line (10 px or so) of red-white-black pixels all across the screen.

EDIT:

Well paint me surprised, it works when I'm using it from within an xwm (xfce in my case).

Hm...


I suppose it wouldn't hurt to run it through valgrind and see what memory abuse I'm likely committing. Glad to hear you got it working, though!

hi everyone
I tried to generate NTLM Rainbowtables but all i get is this error:

Hash ID 0 not yet supported!

Any Sugestion?