Originally Posted by ViN86
That's cool and all, but 80,000 isn't nearly enough.
If we have an 8 letter phrase with letters a-z and numbers 0-9, that's 36 characters. With 8 characters, we have 36^8 different keys. That's roughly 2.8E12. Divide that by 80,000 per second and we are looking at roughly 400 days.
And that's not even considering special characters (!,@,#,$,%,etc.) and different letter cases (a,A,b,B,etc.). Throw those in there and you're nowhere near cracking an 8 character key.
Of course I am talking about brute force. If you can make educated guesses (i.e. eliminate characters) then you will have a power law decay of required guesses. Also, again, I am talking about pure brute force decryption.
True, something most people don't realize is that bruteforcing is still impossible in a realistic time-frame. However.... until these video cards starting doing the work running a moderately sized wordlist was out of the question too, at least that is now an option.
I'd be somewhere around 2,900 PMK/sec without my videocards and 22,000+ with them. ~9x a performance increase is great, but not ground breaking.
I have heard though that 5870's are pushing 30,000 PMK/sec and pyrit is claiming with some new code that they can push into the 50,000 + range on a single card. If that's the case we're starting to see scaling to the point of making bruteforcing short passwords almost possible with a small cluster of servers. It wont be much more than a few years before we have the power to really make waves