But with a 40 MB inputfile (8 digit numbers from 11111111 to 19999999, excluding the ones with a 0 in it) oclvanitygen already consumes 2 GB memory, and it tells me WARNING: better match throughput is possible using vanitygen on the CPUvanitygen might be faster for low difficulty.
I tried to upload my input-file to pastebin, but my laptop ran out of memory copy/pasting it.
To summarize, using an input-file might much faster if you want addresses with 9-10 (or more) digits at the beginning of the address, but your system needs to have a lot of memory.
I was thinking of using an input file method but I thought it would be alot slower as there are alot of possibilities, especially as you start to search for over 10 numbers. I am also quite limited with the space it takes up so this probably isn't the best way for me. I'm using the method mentioned previously and have managed to find an address with 12 numbers in a row and I am gradually increasing the amount of numbers I'm searching for to see how long it will take.