The scaleup...

[Bitweasil]

I've decided that I really need to scale things up.

The goal is a stack of systems in a colo that handles all the table gen, and more importantly helps out with distributed gen. I'm going to be making a tighter table format for nothing but distributed gen data.

Y'all will be able to help gen stuff soon! Significant contributors to generate will be able to get a discount on drives loaded with tables as well... (up to and including free, if you do enough work).

First on the list:

- 16 drive storage server. Will be ~25TB total in a colo facility for collecting bits and pieces.
- Put a generate server in a 4U case & throw it in the colo with 3x6970s.

[Byte/\Puker]

Sounds great Bitweasil

I'm trying to get some 6970 or 6990 for a good price. My current "cracking" system has a 6970 and a 5770.
I think ATI cards are better for cracking and cheaper as Nvidia cards.

I'm interested in you 1-8 lenght tableset. What is the price for it? How long would it take to crack ~100 Hashes with a 6970 ?

Good luck in your OpenCL implementation

PS: I think in a week I can do the next step in my C++ improving (Then I will learned the basics). What would you recommend? Don't say my english I'm still learning it

Best wishes Byte/\Puker

[Bitweasil]

I'm trying to get some 6970 or 6990 for a good price. My current "cracking" system has a 6970 and a 5770.

Watch out on the 6990s. I don't yet support them, and they are apparently very, VERY touchy to support (moreso than standard ATI cards).

I think ATI cards are better for cracking and cheaper as Nvidia cards.

Well, better is a matter of what they're better at. Hashes? Yes. Large numbers of hashes? Not so much. Remaining working with driver updates? Not nearly as good as nVidia.

I'm interested in you 1-8 lenght tableset. What is the price for it? How long would it take to crack ~100 Hashes with a 6970 ?

$500 shipped on a 2TB external drive for NTLM, 100 hashes is on the order of a few hours. Depends on disk speed, mostly - I'm working on some read ahead magic to help improve table search time, which is the bulk of the time right now.

Good luck in your OpenCL implementation

Thanks... nobody implements the OpenCL spec. They just implement what they care about from it. It's always fun developing to an unknown environment - is my code faulty? Is the SDK faulty? Is the compiler faulty? I really don't like OpenCL...

PS: I think in a week I can do the next step in my C++ improving (Then I will learned the basics). What would you recommend? Don't say my english I'm still learning it

Find a project & start in on it. You'll learn as you go.