Hi,
The obvious benefit is speed. The hardware SPI is (or can be) SO much faster than using the bit-banged routines. Then of course you can run it interrupt driven so it does its thing in the background instead "blocking" the processor. Then again, it's so fast that there might not be any benefit in using interrupts. It all depends on the amount of data to process etc etc...

Using SHIFTIN/SHIFTOUT you can still have both devices on the same clock and data pins and run separate CS lines to each, just as you would with the hardware SPI, no special need to use a separate set of pins.

/Henrik.