Re: brute-forcing public keys at amazing speed 2.2 PH/s on CPU [malware warning]
Nice job! Thank you.
You're welcome
So, your total time for 16 keys is 216 minutes, including 9 minutes for 2^30 precomputations. That means that for 16 keys search you need 207 minutes, or approx. 13 minute per one 64bit key. As i saw from your results, the time per one 64 key varied from 2 min to 19 min.
Avg time 13minute is very good. I guess it is faster than the result reqched by Pollard Kangaroo method sahred in BurtW topic.
Avg time 13minute is very good. I guess it is faster than the result reqched by Pollard Kangaroo method sahred in BurtW topic.
If you look at the key who has been solved in 19min, you see that ...7D0E6081C7E0E865 is close to ...8000000000000000 (end range of thread #3) and that it requires a total of 2^33.86 giant steps (max 2^34) so we are close to the end of the range.
It takes ~20 minutes to browse the full 2^64 range, so the theoretical average time to solve a key should be 10min without taking into account the baby step precalculation.
I try to implement hashtable, because hashtable is more fast then binary search.
So, i use 24bit mask
total baby steps = 2^24 = 16777216
and i get a lot of collisions when fill table:
Code:
----------HashTable Info----------
Total hashes:10604491x12=127253892 bytes
Total items:16777216x4=67108864 bytes
Total 194362756 bytes
Total colisions:6172725
Max. colisions:10
----------------------------------
What i see from info:Total hashes:10604491x12=127253892 bytes
Total items:16777216x4=67108864 bytes
Total 194362756 bytes
Total colisions:6172725
Max. colisions:10
----------------------------------
there were unique 10604491 24 bit combinations from total of 16777216 combinations
24 bit combinations that hasve collision was 6172725
and maximum of collision was 10 on some of the hashes
i think it is terrible result of collisions
