If you naively copy a file onto a destination on the same physical disk, the drive constantly has to seek back and forth between the location of the source file and the destination location. Drives are optimized to quickly seek to the next track, so adjusting the buffer size to allow the drive to read for a long time from the source and then write for a long time to the destination improves performance significantly because you wind up with mostly "track to very next track" seeks rather than "track to track on the opposite side of the disk" seeks.
This actually makes things worse if you're copying to a different physical disk. In that case, large buffers mean less overlap between reading and writing. If you read a large amount of data and then write a large amount of data, the disk buffer will rapidly fill and your write will occur at physical disk speed. This will cause you to lose the opportunity to start the next read while the write finishes.
TeraCopy is specifically designed to copy as quickly as possible. So it interrogates the configuration to determine whether the source and destination are on the same physical drive and then adjusts the sizes of its read and writes and how it overlaps them to find the optimum behavior for the exact situation it finds itself in.
Basically, there are two parameters it adjusts. One is how much data it attempts to read ahead of its writing. The other is how much it attempts to overlap reading and writing.