Harvard Architecture: In von Neumann architecture, same memory is used for storing instructions & data. Similarly, single bus called address or data bus is used  for reading data and instructions from or writing data and instructions to the memory.  Later during the development of newer technology was found that this features of Von-Neumann architecture limited the processing speed of computer.

Harvard architecture based computer consists of separate memory for program & data. Each memory has its own address & data bus . As a results of this both instructions and data can be fetched from memory concurrently. 

This feature provides significant processing speed.  There are two address and data bus for program and data memory. Separately, the program memory data bus and data memory. 

Data bus are multiplex to from a single data bus. Where as program memory address bus and data memory address bus are multiplex to form a single address bus. Hence there are two blocks of RAM chips, one for program memory and another for data memory, address bus. 

Data memory address bus where as carries memory address of data where as program memory address bus carries memory address of instructions. Program counter is used to address program memory. 

Program counter always contain the address of next instruction to be executed. Control units controls the sequence of operation to be executed. Data and control bus are bidirectional where as address bus is unidirectional.