Wishlist for CUDA MultiForcer

Discussion and support for the CUDA Multiforcers (Windows and Linux)
  • Ads

Wishlist for CUDA MultiForcer

Postby ILoveBitweasil » Fri Sep 04, 2009 11:39 pm

Hey Bitweasil,

First, I wanna tell you how much we appreciate your time and effort into making this great product. It is really very much needed and we love you for it!

CUDA Multiforcer is simply amazing. I am really glad that you are planning for it to use all GPUs and CPUs available.

My wishlist is simple yet needed. In a way, I wish CUDA to be as close to what we are used to, which is John the Ripper, as possible. I hope you will look into the source of that and give us the best of both worlds.

I wish to see DES included in the choice of hashes. I wish CUDA Multiforcer to be able to use sessions so that we save out work. I wish we can create charsets based on the cracked hashes, or at least that CUDA Multiforcer would accept the charsets of John the Ripper, I wish the commands of CUDA Multiforcer to look more like those of John the Ripper, I wish it to continue and improve its autotune, and have a maxthread setting that is less extreme than the actual, maybe moderatethreads because maxthreads crashes my rig, I wish it to have pot files, I wish it to be able to extract the hashes left, and to do all those great things that John the Ripper, although stopped being developed in 2006, and its patches do so well, but bring that great world to CUDA. I also wish it to be able to have like two different cores in it, one optimized for 100 hashes or less, and the other for much larger hash list, like 50 000 or 150 000 [yes, some people collected uncracked hashes, and it is easy to end up with numbers like those]. I wish to see support for salted DES 64 and 128, and for salted MD5.

Thank you again Bitweasil. I know you are doing it for free, and I hope it didn't disappoint you that I compared your product to John the Ripper, but it is such a standard now that it would be great if you helped us migrate easily and only need your tool for DES and MD5 decrypting (etc), and I really hope you will use JTR and also look at its source file to have a better understanding of it.

Thank you for being such an amazing awesome guy! You know you are making out lives complete and we adore you for that.
ILoveBitweasil
 
Posts: 1
Joined: Fri Sep 04, 2009 11:29 pm

Re: Wishlist for CUDA MultiForcer

Postby gadelat » Fri Mar 11, 2011 4:50 pm

Hello, i really wish session support too!
gadelat
 
Posts: 1
Joined: Fri Mar 11, 2011 4:48 pm

Re: Wishlist for CUDA MultiForcer

Postby Bitweasil » Fri Mar 11, 2011 6:14 pm

ILoveBitweasil wrote:Hey Bitweasil,

First, I wanna tell you how much we appreciate your time and effort into making this great product. It is really very much needed and we love you for it!


Donations are welcome... half the reason I don't keep the Windows version up to date is because I have to reboot to do it.


ILoveBitweasil wrote:CUDA Multiforcer is simply amazing. I am really glad that you are planning for it to use all GPUs and CPUs available.


Yes, the 0.80 alpha does use all the GPUs, and I am working on integrating nein's code for hash cracking on the CPUs... SSE2 is a unique breed of ugly, though.

ILoveBitweasil wrote:My wishlist is simple yet needed. In a way, I wish CUDA to be as close to what we are used to, which is John the Ripper, as possible. I hope you will look into the source of that and give us the best of both worlds.


I don't think I can do John-style mutations. Are you talking about the command line options in general, or the mutations? The mutations and whatever unique brand of probabalistic engine it uses is not easy to port to CUDA - there are some fundamental differences between the architectures they are designed to work on, and CUDA simply does not work well that way.

ILoveBitweasil wrote:I wish to see DES included in the choice of hashes.


What variety of DES? DESCrypt? LM? There are a few different "DES" based hashes...

ILoveBitweasil wrote:I wish CUDA Multiforcer to be able to use sessions so that we save out work.


Yeah, should probably consider doing that. It's easy enough with the workunits... I'll see what I can do.

ILoveBitweasil wrote:I wish we can create charsets based on the cracked hashes, or at least that CUDA Multiforcer would accept the charsets of John the Ripper, I wish the commands of CUDA Multiforcer to look more like those of John the Ripper,


Why can't you create charsets based on the cracked hashes? That's something an external utility should do, not the main cracker.

I'll look into supporting John charsets, and see if I can modify the command line to accept John-style options, but they are fundamentally different tools built around different paradigms.

ILoveBitweasil wrote:I wish it to continue and improve its autotune, and have a maxthread setting that is less extreme than the actual, maybe moderatethreads because maxthreads crashes my rig,


Are you using 0.72? Try 0.80, it's better. Build the 0.80 out of SVN on sourceforge, it's better than any of the drops so far.

ILoveBitweasil wrote:I wish it to have pot files,


Ugh. Considering supporting that output style, I just haven't done it yet. Patches welcome... it's easy enough to add file support to the 0.80 branch.

ILoveBitweasil wrote:I wish it to be able to extract the hashes left,


As in give you a list of uncracked hashes after a run? 0.80 does this.

ILoveBitweasil wrote:but bring that great world to CUDA.


Again, John was written for a single threaded CPU world. Not a massively parallel SIMT implementation. Some of the concepts can be pulled around, but it's not possible to directly "port John to CUDA," which is what it seems you want.

ILoveBitweasil wrote:I also wish it to be able to have like two different cores in it, one optimized for 100 hashes or less, and the other for much larger hash list, like 50 000 or 150 000 [yes, some people collected uncracked hashes, and it is easy to end up with numbers like those]. I wish to see support for salted DES 64 and 128, and for salted MD5.


I support large hash lists for most of the hashes in 0.80. I also include some varieties of salted MD5 in 0.80 - specifically md5(pass.salt) and md5(salt.pass).

Again, what DES do you want? Saying "DES" is useless without details as to which implementation you are trying to crack!

ILoveBitweasil wrote:Thank you again Bitweasil. I know you are doing it for free, and I hope it didn't disappoint you that I compared your product to John the Ripper, but it is such a standard now that it would be great if you helped us migrate easily and only need your tool for DES and MD5 decrypting (etc), and I really hope you will use JTR and also look at its source file to have a better understanding of it.

Thank you for being such an amazing awesome guy! You know you are making out lives complete and we adore you for that.


What specifically do you want out of John? You keep saying "Make it more like JTR." What aspects? Command line? Mutations? Be specific...

And, seriously, donations. It's not cheap to keep myself fed with top of the line GPUs to dev on, and I'm already working on an ancient system. I'd like a bit more hardware to assist with Windows dev without having to reboot. Consider donating... I take money or hardware. :D

I'll see what I can do, but at this point, my interests are moving away from the multiforcer. It's a huge demanding project, and there's more to life than sitting in front of computers.

Take a look at oclHashCat. It's a better brute forcing tool in every way.
Bitweasil
Site Admin
 
Posts: 912
Joined: Tue Jan 20, 2009 4:26 pm

Re: Wishlist for CUDA MultiForcer

Postby stevem » Thu Mar 31, 2011 3:17 pm

It would also be nice if, for salted hashes, it stopped searching for hashes once it has identified the passwords for those particular hash. This would really help performance when there's a long list of hashes as the list would get smaller as the search progressed.

Regards,
Steve
stevem
 
Posts: 1
Joined: Thu Mar 31, 2011 3:12 pm

Re: Wishlist for CUDA MultiForcer

Postby Bitweasil » Sat Apr 02, 2011 4:37 pm

stevem wrote:It would also be nice if, for salted hashes, it stopped searching for hashes once it has identified the passwords for those particular hash. This would really help performance when there's a long list of hashes as the list would get smaller as the search progressed.


I have some plans to revise salted hash support soon, including adding that feature.
Bitweasil
Site Admin
 
Posts: 912
Joined: Tue Jan 20, 2009 4:26 pm


Return to CUDA Multiforcers

Who is online

Users browsing this forum: No registered users and 1 guest

cron