Search found 82 matches

by neinbrucke
Mon May 04, 2009 7:42 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Algorithm optimizations (reversing steps)
Replies: 19
Views: 23526

Re: Algorithm optimizations (reversing steps)

the thing is that you are constantly re-reversing the hashes, so you do have a crossover point. let's take MD5 as an example and do a rough estimation. A full MD5 hash takes 4 rounds, when reversing you do 3 rounds + 1 round for every re-reverse. with MD5, you need to re-reverse after every charset_...
by neinbrucke
Sun May 03, 2009 2:43 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Algorithm optimizations (reversing steps)
Replies: 19
Views: 23526

Re: Algorithm optimizations (reversing steps)

let's revive this one again ;) Thanks for all the explanations about reversing, I have built an MD5 and an NTLM (MD4) brute forcer using them :) http://blog.distracted.nl/2009/05/entibr-ntlm-password-brute-forcer.html http://blog.distracted.nl/2009/05/emdebr-md5-password-brute-forcer.html Not nearly...
by neinbrucke
Wed Apr 15, 2009 9:29 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Question from a newbie
Replies: 14
Views: 10808

Re: Question from a newbie

hmm, maybe i should have read the code first. but i still don't see how this is supposed to help Lie8. As he asked "i know it is "salt"ed ... so how is it possible to crack it using BarsWF??", shouldn't the answer just be "it is not possible with BarsWF, no salts supported" ? (or do i just fail to s...
by neinbrucke
Wed Apr 15, 2009 8:35 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Question from a newbie
Replies: 14
Views: 10808

Re: Question from a newbie

i think bars accidentally posted that message in the wrong topic ;)

looks like it's in reply to http://www.3.14.by/forum/viewtopic.php? ... 77&start=0
by neinbrucke
Tue Feb 24, 2009 9:08 am
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: How high can you get
Replies: 260
Views: 227811

Re: How high can you get

lol... that's some nice speeds :)

btw, how come md5 is about as fast as md4? or is that just because you didn't put that much effort in md4 yet? (such as reversing)
by neinbrucke
Mon Feb 23, 2009 3:05 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: about the algorithm used
Replies: 49
Views: 39355

Re: about the algorithm used

not sure if i have time (or want to make time) to further work on it in the near future... but i'll keep it in mind :)
(i just have some random periods that i actually find joy in programming :P)
by neinbrucke
Thu Feb 19, 2009 5:18 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: about the algorithm used
Replies: 49
Views: 39355

Re: about the algorithm used

tnx both of you... apparently i was already close, but not close enough :p i'll try and order things a little closer, so instead of (for example): SSE_EXPAND_3(t); SSE_ROTATE2_F( sse_A, sse_B, sse_C, sse_D, sse_E, sse_W[t] ); and then another such line for the second one. i'll try to have the two li...
by neinbrucke
Wed Feb 18, 2009 7:23 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: about the algorithm used
Replies: 49
Views: 39355

Re: about the algorithm used

aargh...
to do 2x SSE2 and such... is that a special trick? or am i too stupid to understand how you let 1 instruction apply to 2 variables at the same time?
Or do i really need to use intel compiler for it to work at all?
by neinbrucke
Wed Feb 18, 2009 6:37 am
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Some news about the project
Replies: 23
Views: 16805

Re: Some news about the project

well well, things are getting serious :)
is she going to join this forum as well?
by neinbrucke
Mon Feb 16, 2009 8:09 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Windows/Linux/Mac CUDA MD4/MD5/NTLM brute forcers
Replies: 17
Views: 14422

Re: Windows/Linux/Mac CUDA MD4/MD5/NTLM brute forcers

wow, seems like you are doing a great job... even though barswf keeps topping on single hash... it's great to have a multi hash cracker that gets development... and i'm really looking forward for code releases and the windows version :)
by neinbrucke
Mon Feb 16, 2009 9:10 am
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: Windows/Linux/Mac CUDA MD4/MD5/NTLM brute forcers
Replies: 17
Views: 14422

Re: Windows/Linux/Mac CUDA MD4/MD5/NTLM brute forcers

LM, for what ? when frt finishes creating the huge tables, you won't be needing a bruteforcer anymore. (ah...for 100-99.97% maybe..) Will the LM tables support all possible foreign characters? I believe they are only supporting a few foreign characters. the new table will support most of the foreig...
by neinbrucke
Mon Feb 16, 2009 8:52 am
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: about the algorithm used
Replies: 49
Views: 39355

Re: about the algorithm used

BarsMonster wrote:
yes 2xSSE2 mean 8 plaintexts. What do you mean "modifications" :-) Just 4Ghz C2D :-)
*back from vacation :P*

with "modifications" i meant changes in code, as you said "2xSSE2 = 55Mhash", i have no clue what your speed was with just 1xSSE2 :P
by neinbrucke
Mon Jan 19, 2009 9:24 am
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: CUDA NTLM multi-hash bruteforcer
Replies: 29
Views: 44082

Re: CUDA NTLM multi-hash bruteforcer

Bitweasil wrote:I will also be releasing the code once it is completed and cleaned up, for those curious.
awesome :) i bet i can use that to when i decide to go and learn cuda :)
by neinbrucke
Mon Jan 19, 2009 9:22 am
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: about the algorithm used
Replies: 49
Views: 39355

Re: about the algorithm used

those speeds are probably on your own system? what do you get without your own modifications?

what do you mean by 2x SSE2? You mean 8 plaintexts being hashed at the same time?

anyway, i'm off for a vacation now... cya in 4 weeks :P
by neinbrucke
Sun Jan 18, 2009 12:22 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: punbb hash format
Replies: 8
Views: 7746

Re: punbb hash format

no... that's what i said :p
one of the combinations comes close, but it's useless unless it's exactly the right algorithm-combination :)
by neinbrucke
Sun Jan 18, 2009 12:20 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: What hash to support next?
Replies: 51
Views: 36073

Re: What hash to support next?

"NTLM, SHA-1, Vanilla MD5 are all easily cracked using rainbow tables"

with gpu you can already cover a larger keyspace then with rainbow tables... also you can do many hashes at the same time, without multiplying calculation time that much...
by neinbrucke
Sat Jan 17, 2009 6:47 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: punbb hash format
Replies: 8
Views: 7746

Re: punbb hash format

maybe that was the old punbb, what i said came out of the code :P

symbol, i don't see the right combination in the list, this is closest:
sha1($salt.sha1($salt.sha1($pass)))

(but is useless for you)
it is a strange one really... sha1 with pass, then 2x sha1 with salt...
by neinbrucke
Sat Jan 17, 2009 3:26 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: punbb hash format
Replies: 8
Views: 7746

Re: punbb hash format

return sha1($salt.sha1($str));
so take the salt, concat sha1(password), and sha1 this.

salt is generated: $salt = random_key(12);

random_key:
for ($i = 0; $i < $len; ++$i)
$key .= chr(mt_rand(33, 126));

so, a 12 character salt of readable characters
by neinbrucke
Fri Jan 16, 2009 7:58 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: about the algorithm used
Replies: 49
Views: 39355

Re: about the algorithm used

Hm, I'll look at porting that to CUDA here shortly. Thanks! good luck :) it might not be necessary to reverse the steps i did, it's probably enough to just reverse the value that is A (preCheck) before step SSE_ROTATE4(76)... don't think the code behind that check gets executed that often. i might ...
by neinbrucke
Fri Jan 16, 2009 6:33 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: about the algorithm used
Replies: 49
Views: 39355

Re: about the algorithm used

i used threads for the first time when i build rcracki_mt (rainbowtables stuff)... i first did it using windows threads... that's not portable at all :P then when i was looking for something else, i found pthreads, and a little later boost... what i remember was something about pthreads being a litt...
by neinbrucke
Fri Jan 16, 2009 1:46 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: about the algorithm used
Replies: 49
Views: 39355

Re: about the algorithm used

well, I think it's done now... So I present SHAbr: using SSE2 and it's multi threaded. As far as I know, it's currently the fastest CPU brute forcer for SHA-1. Now close to 60 Mhashes/ using 4 threads. Feel free to make it REALLY fast, probably possible to do things more efficient :) http://blog.dis...
by neinbrucke
Tue Jan 13, 2009 4:59 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: about the algorithm used
Replies: 49
Views: 39355

Re: about the algorithm used

well, i'll release source code anyway, so if anyone feels like compiling with intel c++, he's free to do so...
i think i'm done with optimizing it now, i'll just make it a real brute forcer instead of a benchmark... and it should be done :)
by neinbrucke
Tue Jan 13, 2009 1:20 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: about the algorithm used
Replies: 49
Views: 39355

Re: about the algorithm used

BarsMonster wrote:Try Intel C++ and #pragma unroll 2 or 3, also try higher numbers :-)
intel compiler isn't free and it doesn't seem to support Visual C++ Express... :/
by neinbrucke
Mon Jan 12, 2009 10:10 pm
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: about the algorithm used
Replies: 49
Views: 39355

Re: about the algorithm used

http://msdn.microsoft.com/en-us/library/5hah127h(VS.71).aspx this transpose does the following: #define _MM_TRANSPOSE4_PS(row0, row1, row2, row3) { \ __m128 tmp3, tmp2, tmp1, tmp0; \ \ tmp0 = _mm_shuffle_ps((row0), (row1), 0x44); \ tmp2 = _mm_shuffle_ps((row0), (row1), 0xEE); \ tmp1 = _mm_shuffle_p...
by neinbrucke
Mon Jan 12, 2009 7:29 am
Forum: BarsWF, Cryptography, Security, GPGPU and supercomputing
Topic: about the algorithm used
Replies: 49
Views: 39355

Re: about the algorithm used

I do _mm_set_epi32 out of int* pointer, no need to copy to union. oh, now i remember why i used the union... i used it to set padding byte at array_of_chars[length]... but since i'm now doing a brute forcer, i can just fixate that one in my input generator :) btw, just an idea, not tested yet; usin...