Search found 110 matches

by Bitweasil
Wed May 13, 2009 2:42 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Updated CUDA Multiforcer, with SHA1 & per-position charset
Replies: 11
Views: 10371

Re: Updated CUDA Multiforcer, with SHA1 & per-position charset

shifter1 wrote:Just out of curiosity, how close are you to being able to release the source?
It's mostly a matter of cleaning it up and, ideally, releasing working source. I could dump out a copy now, but it would be a nightmare to dig through for people.

If nobody minds ugly source, though...
by Bitweasil
Tue May 05, 2009 1:03 am
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Algorithm optimizations (reversing steps)
Replies: 19
Views: 18370

Re: Algorithm optimizations (reversing steps)

You forgot that with large lists you'll need to resort every time you reverse. That was the thing I was trying to remember that made things horribly inefficient. Sorting on a GPU isn't particularly quick. Sorting a 5M element list isn't quick, period. I guess one could do the calculations to figure...
by Bitweasil
Mon May 04, 2009 9:18 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Could you make your Cracker support List-Cracking?
Replies: 1
Views: 3024

Re: Could you make your Cracker support List-Cracking?

If you're asking about BarsWF, I'm guessing you're running Windows. I have a multi-hash brute forcer available. http://cryptohaze.com/forum/viewtopic.php?f=4&t=57 You'll need the base files from http://cryptohaze.com/releases/CUDA-Multiforcer-Windows-0.61.zip then download the .exe from the previous...
by Bitweasil
Sun May 03, 2009 3:01 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Algorithm optimizations (reversing steps)
Replies: 19
Views: 18370

Re: Algorithm optimizations (reversing steps)

Is there an effective way to reverse things when dealing with large hashlists? It looks like 1 or 2 steps can possibly be reversed with MD5 (the null input bits past the length), but doing more significant algorithm reversal involves doing work for each step, which could get to be a lot of work for ...
by Bitweasil
Fri May 01, 2009 7:39 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Updated CUDA Multiforcer, with SHA1 & per-position charset
Replies: 11
Views: 10371

Re: Updated CUDA Multiforcer, with SHA1 & per-position charset

Yea, I'm going to release the code for it.

I'm totally redoing the design, yet again, to something a lot more solid and scalable (TCP-IP based, workunits, a variety of compute clients supported).

Keep an eye out at Cryptohaze & I'll have the source up soon.
by Bitweasil
Fri May 01, 2009 6:51 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Updated CUDA Multiforcer, with SHA1 & per-position charset
Replies: 11
Views: 10371

Re: Updated CUDA Multiforcer, with SHA1 & per-position charset

Quite fast. Im running a GTX 295 with the md5 option and getting stepping of 109M/s. Minor note: It doesnt tell you the command line option for charsets anymore. Working on 1171 hashes at once. EDIT: Adding threads = 256 made it jump to 295M/s The defaults don't work right on the GTX series cards, ...
by Bitweasil
Fri May 01, 2009 1:24 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Updated CUDA Multiforcer, with SHA1 & per-position charset
Replies: 11
Views: 10371

Updated CUDA Multiforcer, with SHA1 & per-position charset

I've released a new beta of the CUDA Multiforcer (for Windows & 64-bit Linux). http://www.cryptohaze.com/forum/viewtopic.php?f=4&t=57&p=264#p264 It's much, much faster on large hash lists. On 5M hashes, 216SP GTX260: Work: 222 Done: 0.07% Time: 481.00 ms Stp/srch: 60.5M/s 302471968.0M/s Also, it has...
by Bitweasil
Mon Apr 27, 2009 2:38 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Nvidia or ATI
Replies: 14
Views: 9488

Re: Nvidia or ATI

LordMike wrote:Prob. tells something about popularity or lack of logistics :P
Low yields. If The Register is to be believed, nVidia loses a lot of money on each GTX295.

Hopefully they'll start showing up more, but putting together a 1.4B transistor chip reliably is just difficult.
by Bitweasil
Mon Apr 27, 2009 1:03 am
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Nvidia or ATI
Replies: 14
Views: 9488

Re: Nvidia or ATI

BarsMonster wrote:Yep, nVidia is wider supported. i7 is faster :-)
That's largely because their programming environment isn't slopped together of "Oh... crap, we should probably do something about this... uh, here!"

i7 chips are <3 <3 <3, but GTX295s are really hard to find right now.
by Bitweasil
Sun Apr 26, 2009 9:02 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Nvidia or ATI
Replies: 14
Views: 9488

Re: Nvidia or ATI

Can you actually obtain GTX295s in bulk anywhere?

What are you planning to do? Pure MD5 cracking with BarsWF, or other products as well? nVidia appears to be currently supported with more cracking products.
by Bitweasil
Sat Apr 25, 2009 4:26 am
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: lease time on my High-End GPUs cluster for $$
Replies: 24
Views: 16415

Re: lease time on my High-End GPUs cluster for $$

YrFear wrote:Money is of no importance to me, only cracked hashes are
I will happily build, manage, and power a cluster of as many GPUs as you want, as well as write custom cracking code for you, full time, if money is no object. You may PM me to discuss starting salary.
by Bitweasil
Thu Apr 23, 2009 9:34 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: lease time on my High-End GPUs cluster for $$
Replies: 24
Views: 16415

Re: lease time on my High-End GPUs cluster for $$

Woah, overclocked GPUs?

Not interested anymore. Not at all.

And at $20/hr, that works out to right around $15000/month. That's a lot of very nice compute hardware.
by Bitweasil
Thu Apr 23, 2009 4:36 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: lease time on my High-End GPUs cluster for $$
Replies: 24
Views: 16415

Re: lease time on my High-End GPUs cluster for $$

What are the cluster specifications in terms of number of GPUs, GPUs per system, system interconnects, etc?
by Bitweasil
Wed Apr 22, 2009 2:20 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Some news about the project
Replies: 23
Views: 12900

Re: Some news about the project

Hi is it already working with BSD hashes? Like: $1$tZfyC3uY$vfu/iFI54YadXUIzZEeWd/ The primary problem is that pretty much everyone doing CUDA work right now has been poached by companies for various jobs - I believe pretty much all the non-company-aligned CUDA hash cracking types or CUDA rainbow t...
by Bitweasil
Fri Apr 10, 2009 10:39 am
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: use different gpus?
Replies: 3
Views: 3322

Re: use different gpus?

SDK sayd you cant, but in practice it works most of the time. We have some problems with GPUs of different generations. I would agree if the question were, say, an 8800 GTX and a GTX285. The GTX295 & GTX285 are the same core (GT200), and the same generation, even (55nm). One just happens to be two ...
by Bitweasil
Fri Apr 10, 2009 1:53 am
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: use different gpus?
Replies: 3
Views: 3322

Re: use different gpus?

You'll need to break the SLI (remove the GPUs from the SLI bindings - CUDA and SLI aren't compatible), and it will be seen as 6 GPUs (each of the 295s is two physical cards), but that should work fine.
by Bitweasil
Sun Apr 05, 2009 3:06 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: How high can you get
Replies: 260
Views: 177119

Re: How high can you get

The primary benefit of the GT200 cores (GTX*) over the G80/G92 cores (8000, 9000 series) is double the register count, and double precision floating point capabilities. Plus 192/216/240 stream processors vs the 128 max in everything previous (even the 9800GX2 was two 128 stream processor cards bound...
by Bitweasil
Sat Apr 04, 2009 1:48 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: How high can you get
Replies: 260
Views: 177119

Re: How high can you get

As soon as my 3 GTX295s show up, I might be able to run some benchmarks.
by Bitweasil
Tue Mar 31, 2009 2:00 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Some news about the project
Replies: 23
Views: 12900

Re: Some news about the project

Offshore commercial development shutdown?

What does that consist of? That's not a familiar term.
by Bitweasil
Tue Mar 31, 2009 12:00 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Some news about the project
Replies: 23
Views: 12900

Re: Some news about the project

BarsMonster wrote:I afraid to give any timeframes :-) I have 3 commercial projects competing for my time with free barswf :crazy:
Isn't that always the problem? Bills coming before play? I've run into that as well. :(
by Bitweasil
Tue Mar 24, 2009 7:16 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Multiple hashes
Replies: 12
Views: 7135

Re: Multiple hashes

Oh, if you want to PM me the hashes, I could run them easily enough. I'm hoping to finish coding on it soon, I've just been busy with paying the bills lately. :(
by Bitweasil
Tue Mar 24, 2009 5:07 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Multiple hashes
Replies: 12
Views: 7135

Re: Multiple hashes

I noticed that, for 100 hashes it runs at 200M/s per iteration and for 1000 it runs at 18M/s per iteration. I wonder why that is... Because it has to test each generated hash against the hash list, and version 0.61 is fairly inefficient at doing so. The version currently in development, 0.7, is muc...
by Bitweasil
Tue Mar 24, 2009 11:35 am
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Multiple hashes
Replies: 12
Views: 7135

Re: Multiple hashes

If you're dealing with multiple hashes, it will be substantially slower than working on a single hash. There's a much faster version I have that I just need to finish tweaking & release.
by Bitweasil
Tue Mar 24, 2009 1:25 am
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Multiple hashes
Replies: 12
Views: 7135

Re: Multiple hashes

http://www.cryptohaze.com/bruteforcers.php

Not quite as fast as BarsWF, but they're cross-platform multi-hash CUDA brute forcers.

Also the InsidePro stuff is very fast on multiple hashes, but not free.
by Bitweasil
Tue Mar 17, 2009 3:40 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: MD5, and 2 constants
Replies: 14
Views: 9113

Re: MD5, and 2 constants

Right, that's why I asked.

Also, >447-bit inputs are annoying, as you need a multi-pass implementation. Tip for those securing passwords: Just make sure you have a >55 character string at some point, and it's annoying to work with.