3.14.by forum

You can easily test it by compiling an AVX only version. AVX uses the same intrinsic functions as SSE2 and stuff. So you just have to tell the compiler to use AVX instructions.

For the Intel compiler:
icl /QxAVX
icc -xAVX

MS Visual Studio compiler flag:
/arch:AVX

GCC compiler flag:
-mavx

So does this mean you're going to add AVX support?

Windows task manager. In the processes tab right click on the process for BarsWF and click set affinity then select which cores it can run on. The CPU version is really old 0.8. The source code from http://3.14.by/forum/viewtopic.php?f=8&t=1333 says has a version of 0.c (12). You can download the CU...

With the Core i7 you might want to try 8 CPU threads since it has hyper threading. 25 * 2.93 / 2.4 = 30.5. Nvidia cards have a lower clock-core vs ATI but Nvidia cards are more efficient per clock-core. 5570 is 650 MHz * 400 cores = 260,000 Mclock-cores 460 is 1300 MHz * 336 cores = 436,800 Mclock-c...

Gate counts: 36 33 33 26 35 34 34 32
Average: 32.875

http://www.openwall.com/lists/john-users/2011/06/22/1

Oh oops the average is around 5.63 Thash/s. Going with 10% extra so 23k 6870's so 46 PFLOPs... Hmm I think I was off by a zero on that. Since the fastest supercomputer is the Tianhe-1A which is only 4.7 PFLOPs. The sum of the top 500 is only 64.7 PFLOPs (11/2010). Hmm the Cray XK6 can scale to 50 PF...

Just letting you know that any government can take out Bitcoin or well any large entity. * Given $10 million USD, bot net, or commandeer/rent a super computer or two: 20000, 6870's (20000 * 272 MHash/s = 5.440 Thash/s > "Network total 5.152 Thash/s" http://bitcoincharts.com/about/) is more then the ...

Happy PI Day!

mysql323 beta 4 is up: * Bug fix buffer overflow mysql323 collider beta 1 is up: * 1.83 Quadrillion passwords/second on a Q9300 @ 2.5GHz (DDR2 800 CL5) * requires 1.5 GiB ram * Can only crack [a-z]{7}["$&(*,.02468:<>@BDFHJLNPRTVXZ\^`bdfhjlnprtvxz|~]{3,9} where [a-z]{7} has a sum of 736 and the next ...

beta 2: * on exit it displays stats * ctrl+c displays resume code and exits * xp support (sorry about that one I totally forgot that I was using SleepConditionVariableCS) beta 3 is up: * Bug fix there was a race condition that I didn't know could happen. Apparently pthread_cond_signal has a short de...

Beta release. For me it tops out at just over 5 Trillion passwords / second on a Q9300 (2.5GHz). :( I was expecting closer 8 Trillion. * Windows and Linux (I know it runs in CentOS) * Multi threaded :) * Key space file for several sub key spaces with per position character sets. http://www.tobtu.com...

The CPU code that does over a trillion passwords per second for MySQL323 has divergent code so it will be hard to get nice performance with GPUs or SSE2. Also it only works for a single hash at a time so multiple hashes will slow it down until it gets to a point where it makes more sense to just gen...

HD5970 has two GPUs in it.

4 HD5970 * 725 MHz * 1600 cores * 2 GPUs = 9,280,000
vs
8 HD6970 * 880 MHz * 1536 cores = 10,813,440

8 HD6970 is 1.1652 times faster than 4 HD5970.

BarsWF does work with md5(unix) hashes just plain unslated md5 hashes. Interesting it appears that he's only checking to see if it is 32 characters long. This is a bug you should only be able to input characters that are hex ("012345789abcdef" or "0123456789ABCDEF"). md5("david") is 172522ec1028ab78...

I'm going to guess you mean Radeon 6970 not Radeon 6790. A Radeon 6970 58.3% the speed of a Radeon 5970 (assuming the stats of 880 MHz and 1536 cores are correct). So 8 Radeon 6970 would give you 33.3 billion/second (assuming they can do the same amount of work per clock-core). I assume the 6970 is...

I'm going to guess you mean Radeon 6970 not Radeon 6790.

A Radeon 6970 58.3% the speed of a Radeon 5970 (assuming the stats of 880 MHz and 1536 cores are correct). So 8 Radeon 6970 would give you 33.3 billion/second (assuming they can do the same amount of work per clock-core).

Your DNS is broken!

What happened?

My point is that if the required hash keeps changing, then brute forcing is only worth while if you can complete all the permutations in less time than the hash is changed? Wrong. It's a probability game let's say you can only check 10% of the key space before it changes. You'll have to guess 65.92...

Are you talking about this?
http://en.wikipedia.org/wiki/SecurID

And no it is not uncrackable. Well something that has a 256 bit key I'd say is uncrackable (well depending on the algorithm such as a 256 bit RSA private key is crackable in like a hour on a new computer).

Ah ok i get you, so on a simple level imagine you have two 'threads'. One thread could be tasked with ensuring there is always 5000 hashes available and the other thread could check your hash with the 5,000 target hashes? Therefore there's never any overhead lost on creating the 5000. I don't know ...

This has nothing to do with GPUs or CPUs this is the big picture. He's saying if you have 5000 MD5s that it is faster to generate an MD5 and check it against a list of MD5s you're trying to crack than generate an MD5 and check it against the first MD5 you're trying to crack then regenerate the same ...

Does anyone know the speed difference between BarsWF Brook version and ighashgpu?

I'm pretty sure ighashgpu is less than 23% faster than BarsWF Brook version. So 9.87 / 1.23 = 8.02 Billion PWs/sec hey it's a tie.

blazer wrote:impressive

Would 4x 5970 beat that?

Two 5970 would beat that. (11.73 Billion PWs/sec)
http://3.14.by/forum/viewtopic.php?f=8& ... 8509#p8509

Three 5870 would beat that. (9.87 Billion PWs/sec)
http://3.14.by/forum/viewtopic.php?f=8& ... 8449#p8449

Those numbers are from ighashgpu.

Dynamic creation of an array of classes using a non default constructor. class SomeClass {public: SomeClass(int);}; ... int useNonDefaultConstructor = 0; SomeClass *sc = new SomeClass(useNonDefaultConstructor)[5]; ... ------------ I'm pretty sure there is no way to do this in c++. Well besides using...

Yeah but it only saves: 3 out of 64 steps for MD5 if you use the first 32 bits or less or 2 out of 64 steps for MD5 if you use 64 bits or less (the first 32 bits and last 32 bits) You can save 1 more step if all the passwords are less than 36 bytes, but then you need to use bits 32-63 or bits 0-63. ...