Hi, nice project. Could you please share a schematics with us? It's OK if it's drawn by hand on a piece of paper. I'm interested in seeing how you generated 10V for VHH and the part with the transistor to switch between 5 / 10 V. The software part seems to be clear to me: Address buss is generated via the 2 counters by toggling the 'clock' pin (I like that, it needs just one clock pulse to generate consecutive 12 bit address), the data bus is 16 bit wide in write mode ( top byte of the data bus is generated by the shift register, bottom byte uses 8 arduino pins D2-D9), and for reading only an 8bit data  mode is used where the data bits are again mapped to arduino pins D2-D9 (but this time switched as input). The rest of the code  is just sending and receiving the data between the PC and MCU.

My tips for speed improvement:

1) do not write data with value 0xFF (erased flash memory has already value 0xFF)

2) In flash_databus_output you always set the pinMode for data bits even if you write 128 bytes in one go - optimise it to set the ouput pin mode just once, before the first byte is written.

3) unroll the loops where 8 bits are sent  (the time spent calculating the 'for' loop is eliminated)

4) do not use digitalWrite function - although it is portable, it is also slow because it does several checks and also (if it is not inlined) there is an extra overhead of entering the function, passing the parameters, then exiting the function. Use a macro  that toggles just a single pin on a specific arduino port for maximum speed.

5) use some some sort of data compression (maybe it's an overkill for a simple project like this)

Would you share what software you used to program MX29F1615?

Have you the pcb layout with all components? And how to program all this?