by Guest on 2019/03/22 03:34:01 PM
I'd like to add that if Your proposition would be implemented, the devs should also take care to not read from any single physical drive in multiple threads, even if logical partitions are different.
IOW tixati will need to distinguish that, say, C: and D: are on the same HDD platters, while E: is not, so it is safe to hash from E: and D: simultaneously in multiple threads, but concurrent hashing from C: and D: should not be possible by default, and should resort to single thread in latter case (if IO-bound and not CPU-bound, that is).
This whole affair might also need to be implemented by using separate threads for reading, buffering, hashing, and passing events regarding two.