That is why it's not time for the BF still. We need to gain information about how the nck is generated and we can only do that by having a lot of NCK-IMEI pairs. As far as we know Geohot has 6 pairs so you guys should use your 4-quad core computers to find more people that want to help by giving their NCKs instead of trying to brute force it.
Actually I agree w/ deco...
PS: Welcome me :)
There are coutnless way to math it, you don't know how, and it won't be easy if it is. Second, it's VERY likely, IMEI does not have a connection with the NCK, thats technical and security wise recommended and if you loock into the algorithm, try to think about it, really.
The algorithm is know for a while now, to those who worked on the iphone unlockign and seriously, there is very little space for guess immo. A bruteforce will take incredible much cycles and it's questionable if this way is even of any use for an unlock, unless you have the math powers and programing skills/time to do, creating a real distributed network. Even than the input/output and use of it is questionable again.
In the end, or at the current level take it as a POC game and don't have to high expectations towards this method of possible unlock for the masses.
Agree you but one note besides, that was not _only_ a code optimizing. The run was done on a singel system with two Xeon DP 5130 and 8GB RAM ;)
Originally Posted by Sproglet
Yes thats true! But it would be nice if anyone can release a method which can be used by people who had access to a lot (and i mean a lot) of math power eg. CPU's. So these people can spend a lot of CPU time for weeks (or months, so time is not my problem for a real legit unlock) like me. This method is not for the big public community, sure, but for some peoples it could be interesting also to do this. The main target should be a regular unlock based on anySIM but the other "solution" should be coexistent in my opinion.
Originally Posted by sam
Just a few calculations I made:
Let's we can get a total of 500 machines working on this(I believe we can get more than 1000)
And that we are satisfied with a solution that unlocks a phone every 2 weeks?(not really that good...)
Right now the algorithm(optimized with compiler option) running on a single maching will take 117 years to crack the nck(worst case senario always)
To get an unlock(for just one phone) in 2 weeks with 500 machines we would need an the algorithm to be optimized to finish on the 1/6 of the time it finishes now.
Lets say we can find 1000 machines then we can either optimize it to 1/3 or if we can opt to 1/6 then we would get an unlock every week.
At least for me this is exactly the reason I'm doing it. There are plenty of people out there who do have access to some substantial machinepower and can run this BF once it's optimized enough and there are ways to extract the relevant ID-s and tokens from BL4.6 phones.
Originally Posted by Shade.sh
The point is that this will quite likely NEVER be a solution for the masses unless we do stumble upon a relationship between the IMEI and the NCK, but it will be a solution for a certain % of people who do have the tools (similar to hardware unlock method, but likely a bit more people than that even).
The approximate speed of 60k keys/sec/core is what we see at the moment depending on the core clock, with that one needs 6430 cores for the key to be unlocked in one month. That number can likely be brought down a bit further through optimizations as well as assuming that it'll in reality take you a lot less time to stumble upon the correct NCK, but it will not be a solution for the masses as you can understand from here.
So people who expect that they can BF on their home PC/PC-s, not gonna happen...
How? This would be very usefull...
Originally Posted by toruonu
I'd rather have my PC run and don't find anything, that don't let it run and just wait...
User grey (maybe) made a POC exploit to do this. But don't blame me if this is not true, i think i read it somewhere....
Originally Posted by Dive
I think we have a better shot finding the weakness in the NCK generation system than attacking how the NCK is related to the unlock.
In case we can't find any relation, we'll still be able to shorten the search down as Geohot said.. if the first number is known for each area.. it's a 10 fold reduction in our search... so any information gained from the NCKs is extremely important even to the BF project.