3.14.by forum

Hmm according to Wikipedia it took 8000 MIPS-years to factor a 512 bit RSA public key which is 2.92 days at 1 trillion instructions/second (Radeon HD 4850 X2). Well there is a chance that it doesn't run nicely on GPUs and it will take a lot longer than 3 days. 512 bit 8000 MIPS-years 463 bit 2000 MI...

Yes, how many bits is your private key and which algorithm is it (RSA, DSA, ...)?

for (int i = 0;i<deviceount;i++) { threadhandle[i]= CreateThread(NULL,0,(LPTHREAD_START_ROUTINE)GpuThread,this,CREATE_SUSPENDED ,&id); SetThreadIdealProcessor(threadhandle[i],i*2); SetThreadPriority(threadhandle[i],THREAD_PRIORITY_TIME_CRITICAL); ResumeThread(threadhandle[i]); } Probably using of T...

I did the same for LM and NTLM http://funprogs.topcities.com/hashgen.htm <-- wooohh best web site ever one of these years I'll update it.

Wait so this is legit?

neinbrucke wrote:

Sc00bz wrote:not yet but it's planed

does that mean you are somehow involved with barswf development?

No, I read it on this forum somewhere. Unless that was cryptohaze hmmm... Well either way it's planed for here or there.

not yet but it's planed

If it's 64 bit it should start slowing down after interlacing it three times since you need 5 registers to do MD4 or MD5 and you have 16 registers to work with. It might be faster interlacing it four times but it depends on your cache.

You forgot that with large lists you'll need to resort every time you reverse. Also with reversing you don't need to re-reverse the whole way you can cheat again. With MD5 you can reverse 2 steps save that point, reverse 7 more steps save that point, and then reverse 7 more steps save that point. Th...

I'm pretty sure the price at which people will be interested in your cluster it won't be profitable for you. $105 for a recertified 9800 GTX+ or $130 for a GTS 250 (same card different name). Prices are from newegg.com. 560 M/s vs 5 B/s So 14 hours for $10/hr at 5 B/s is 234 trillion passwords (2^47...

Don't you mean $20 USD a day. Since it would cost you at most $5 USD a day for power.

#include <stdio.h> #ifdef _WIN23 #define uint64 unsigned __int64 #else #define uint64 unsigned long long #endif void num2pw(uint64 number, char *buffer, const char *charSet, int charSetLen, bool incRight = true, int minLength = 1) { uint64 pow = 1; int len, pos; if (charSetLen <= 0 || minLength < 0...

Use LHC for now it's free and fast, but it only works in Windows.

Huh? Are you still stuck on trying to reverse engineer the brute forcer. It's not that hard if you try to optimized md5 by writing it in x86-64 it looks almost the same as Bar's brute forcer. I don't know why he uses Intel's compiler because he's forced to either spend $1000 on a license or never as...

So why is there a limit when you can just dynamically allocate enough threads?

I am going to go with it's definitely not reversed, but just think if you do you'll get around 9191.42 MHash/sec (5553.15 * 48 / 29 = 9191.42).

2x SSE2 just means to interlace (doing your algorithm twice):

algo:
a += b;
b ^= c;
c &= d;
b += d;

"2x" algo interlaced:
a1 += b1;
a2 += b2;

b1 ^= c1;
b2 ^= c2;

c1 &= d1;
c2 &= d2;

b1 += d1;
b2 += d2;

Here's LM and NTLM implemented in javascript http://funprogs.topcities.com/hashgen.htm

Ohh oops I feel like a noob I forgot to tell you where to find the code for that program that does 1.6 trillion/sec on a 2.4 GHz. http://oxid.netsons.org/phpBB2/viewtopic.php?t=2562 I also posted it to free rainbow tables first but this is nicer since there's not many posts everywhere. If you can't ...

I can get over a trillion/sec on a 2.4 GHz P4 what the hell are you run a 8086?

Radeon HD 4870 X2 for $539.99 GeForce GTX 295 for $499.99 1213.36*2/539.99= 4.49 MHash/sec/USD 706.32*2/499.99= 2.83 MHash/sec/USD 59% more cost effective to get a Radeon HD 4870 X2. Anyway the price point is at Radeon HD 4850 which is around $150 and should get around 1000 MHash/sec (6.67 MHash/sec...

DES will probably run faster on a GPU than a CPU per FLOPS. This is because each MP has 8,192 (8000s and 9000s) or 16,384 (GTX 200s) registers vs 16 for a CPU. The best bit slicing implementation is optimized for least amount of instructions (but it might not be the must optimal) this does not assum...

dJe781 wrote:Edit : Erm, I guess that 700 Mhash/s means that both of my GPUs are working

You need to disable SLI for both of them to work in CUDA. Once you do that you'll get 1600 Mhash/s instead of 900 Mhash/s.

10268 million chains per table so like 77 GiB per table. The first table will be twice as big. So the total size is 385 GiB. Unless http://www.freerainbowtables.com/en/tableprogress/ is messed up or something. I think it's reporting perfect chains which is why it looks like there's less than a day l...

oops that was for a 4830 but then I changed it to 4850 because I was like wait you can't get that speed with a 4830.