I can't understand, is the search for private keys from the list of full addresses (100000+) on the processor and on the video card not the same (despite the speed)? And what is the difference if it is?
I just peeked at the code and it doesn't look like there's any. The GPU code is called from the main loop in CPU so the only things which could possibly be different could be the way it checks points/hashes and how it iterates through them.
I think it is very similar to these btc address generators. But underlying algo is different which is ed25519 in case of onion v3 addresses. How hard is it to port vanitysearch to do onion v3?
The Integer class is abstracted enough to let you make your own curve class and slot it right inside as a parameter, but writing ed25519 arithmetic in assembly will be one hell of a challenge.
... or you could directly use this expression -i input.txt > output.txt this way no extra work has to be done we can directly save it to text file whatever the command line is outputting
I don't like this particular approach. The console is just as slow as file I/O, which you can easily notice if you run any cpu-intensive program with its --verbose switch and notice how the time for the command to complete
becomes longer because so much time is being spent printing to the screen.
Redirecting output to a file from a shell doesn't make it any faster, it just "fools" the program into thinking its writing to the console.