I've a question on brute force and encryption

[afdude]

Hello folks, just joined the wonderful community here. I do have a question concerning brute force and encryption. Would it be possible to use a Geforce cards CUDA cores for possible brute force to gain keys from an encrypted flash dump? Cuda cores as some of you know have much more uses than just in the graphic department. For example, in satellite hacking/unlocking they can be used to brute force crunch for keywords and keys in determining CSA tables. These cores are blazing fast and work relatively well crunching data in other fields. I'm simply wondering if that approach can be applied here? Thanks.

[indirect76]

This is an interesting topic for me. The first ECC scheme publicly broken was done with a cluster of 200 PS3s in 2009. Let me do some research and I'll come back after I crunch some numbers...

[indirect76]

OK, so based on the previously discovered key, the key size is 160 bits (2^160). This is equivalent to a number between 1 and 1.45*10^48. If you could get a single GPU to process a billion (10^9) keys in a clock cycle, and the GPU ran at one Ghz. that would allow for one quintillion (10^18) key s to be processed in one second. Let's say we have a billion computers crunching away, that would increase it to one octillion (10^27) keys being process a second.

Given this very generous situation it seem like it would take roughly 10^21 seconds to run a complete brute force attack. How long is that? Well there are 31,536,000 seconds in a year. Simple divison gives us 31,709,791,983,764 years. 31 trillion years.

That's a few thousand times the age of the universe.

I didn't double check the math, but I'm sure I'm in the right ballpark. I'll follow up after lunch.....

[frankplummer]

Hi guys,

Just thought I would stick my oar in here.

I'm a systems administrator and a large proportion of my job revolves around security. I've used nVidia GPUs to test hashes in the past using an application called Hashcat. This allows GPU-powered hash cracking and at the same time, clustering ability. I'm sure you know of this already but, it may be of use in your 31 trillion year project..

[JOshISPoser]

well, we better get started then :P

[indirect76]

Well in the 31 trillion year scenario, I'm imagining we are looking for a unique key for decryption. If the keys we were looking for were for a hash check then I would imagine we could just modify the firmware not to do that check. I'm a little fuzzy on the exact problem we have in the scene right now about not having keys. I think they would be for decryption.

[kavitha]

hi thnaks for your information