First page Back Continue Last page Overview Graphics

Computer Architectures


The traditional computer organization (as described in early lectures) uses von Neumann architecture. In von Neumann architecture, all information is transferred between CPU and memory through one set of address, data and control buses. When the data transfer rate becomes the bottleneck for fast microprocessors, the Harvard architecture is adopted. The Harvard architecture uses separate program buses (program address and program data buses) and data buses (data address and data data buses). The memory is separated into program memory and data memory. Most digital signal processors adopt the Harvard architecture.