In the early days of the 8051 it was thought that there would be quite a bit of packed BCD arithmetic done with the MCU. Packed BCD arithmetic allows for fairly effcient algorithms to be written that can support more than just the 4-5 equivalent digits you get when working with 16-bit binary entities. The XCHG instruction has good use in algorithms that scan through buffers of packed BCD digits in memory.

Once I wrote an application that read data from mag cards using an 8051 type MCU. All the data from the card was able to be stored in 4-bit chunks and to make efficient use of RAM the buffers that held the data were packed two nibbles (the 4-bit chunks) were packed together at two nibbles per byte. I made good use of the XCHG instruction in the low level routines that were used to store and fetch nibbles to/from the storage buffers.

Michael Karas