A. Digital Signal Processors (DSP’s)
DSP’s have began to appear around 1979. Since then, several DSP generations with increasing performance have
been introduced by many manufacturers. These chips have been developed specifically for real-time computing in digital signal processing applications. Many DSP’s function as embedded real-time processors in special-purpose hardware such as modems, speech coders, speech synthesizers, speech recognition systems, and image processing systems.Table 5 shows popular recent DSP families offered by different manufacturers. I) DSP Internal Structure and Features: Most DSP’s are
built with a Harvard architecture, where data and instructions occupy separate memories and travel over separate
buses as shown in Fig. 4. Because of this dual bus structure, the processor can fetch simultaneously an instruction and a
data operand. Pipelined operation of instructions and data transfer is thus possible, resulting in a higher instruction throughput rate. The pipeline can be from two to four levels deep, depending on the architecture. In order to optimize the processing speed, important operations such as multiplication and shift are implemented in hardware instead of using software or microcode. In recent DSP’s, the execution speed is further enhanced by using several independent units, multiple bus sets, and additional units such as instruction cache, register file, and dual-access memories.
The operation of DSP’s is optimized so that most of the instructions are executed in a single cycle. Third-generation