Basic reimplementation of md5sum that computes hashes in parallel.
The naive way to run md5sum in parallel is to run multiple processes, e.g. via xargs. Example:
find /path/to/files -type f -print0 | xargs -0 -P 2 md5sum
where the -P 2 argument to xarg causes two md5sum processes to run in parallel.
The problem with this naive approach is that the multiple md5sum processes issue reads concurrently. On hard disks (but not SSDs) the concurrent reads cause extra seeks and can reduce the disk read throughput.
The seekmin implementation allows parallel hashing while keeping sequential reading.