Has anybody tried these???

[NC009]

I am looking to build a new workstation specifically for brute-forcing mainly WPA/WPA2 handshakes/hashes (I would also use it for other things.. ). Take a look at this nifty piece of hardware AND THE POWER...I wonder how much PMK/s you would get if you ran 4 of these in parallel with enough RAM to support? I know BackTrack has drivers available specifically for the GPU CUDA software (for PMK/s acceleration). Thoughts anybody?

Here is the link:

http://www.nvidia.com/object/personal-supercomputing.html

[Mellow]

That's one power piece of machinery. Pretty price too. What's the MATLAB option though?

[xor]

You can build better computers with more power for less money than buying a Tesla.

[FuyuKitsune]

NO NVIDIA.

For hyperparallel processing you would be better off with several AMD/ATI cards. AMD's "more cores!" mentality is really badass for cracking. NVidia designs cards that are fairly balanced and even the Tesla series don't match AMD's core count. NVidia's balanced style is good for graphics (it stays on par or slightly exceeds 'more cores') but sucks for bruteforcing.
This style difference is extremely noticeable to anyone who has mined bitcoins which is basically bruteforcing SHA-256 hashes. NVidia cards get utter crap speeds while the AMD cards fly. Even the optimized CUDA SHA functions are lucky to get a third of the speed of OpenCL on AMD. Check this chart and see the diff:
https://en.bitcoin.it/wiki/Mining_hardware_comparison Leftmost column is the card model, 2nd from the left column is the hashrate in million hashes per second.
Mid-to-upper level AMD cards generally get 100-300 (depending on clock) and the 59xx and 69xx series gets 300-800. Nvidia's competing cards get around 1/4 of the AMD cards. The highest NVidia card, some Telsa model, only reaches 155Mhash/s.

tl;dr - want to crack hashes? Use AMD, more cores.

[NC009]

NO NVIDIA.

For hyperparallel processing you would be better off with several AMD/ATI cards. AMD's "more cores!" mentality is really badass for cracking. NVidia designs cards that are fairly balanced and even the Tesla series don't match AMD's core count. NVidia's balanced style is good for graphics (it stays on par or slightly exceeds 'more cores') but sucks for bruteforcing.
This style difference is extremely noticeable to anyone who has mined bitcoins which is basically bruteforcing SHA-256 hashes. NVidia cards get utter crap speeds while the AMD cards fly. Even the optimized CUDA SHA functions are lucky to get a third of the speed of OpenCL on AMD. Check this chart and see the diff:
https://en.bitcoin.it/wiki/Mining_hardware_comparison Leftmost column is the card model, 2nd from the left column is the hashrate in million hashes per second.
Mid-to-upper level AMD cards generally get 100-300 (depending on clock) and the 59xx and 69xx series gets 300-800. Nvidia's competing cards get around 1/4 of the AMD cards. The highest NVidia card, some Telsa model, only reaches 155Mhash/s.

tl;dr - want to crack hashes? Use AMD, more cores.

Wow, thanks! This is exactly the information I was seeking... Nice link. Good Stuff! I appreciate it.