What are you after? Verify whether pairs of files (same name, same place in the directory hierarchy) are identical, but you don't care about the actual differences, only a single same/different indicator? And you need it to run really fast, about the speed the disk or file system can deliver data at? The most efficient approach is to calculate a good checksum of each (sha-256 or -512 will work well and runs fast), and compare the checksums. It is quite easy to wrap that in a script to run it over all files in a directory hierarchy, using find. The only kind of tricky part is to find files that are in the "other" directory hierarchy, but not present here, that requires a small amount of coding skills.
Let's not argue about the birthday paradoxon and the probability that the checksum approach will give the wrong answer; with large enough checksums (256 bits and so on) this is a non-problem in the real world.
If you are after comparing files and getting information about what the differences are: Why not just use the built-in diff, with the -r option? That will walk directory trees.
