Digital Signal Processing: World Class Designs

Chapter 9. Code Optimization and Resource Partitioning

DSP is all about algorithms. These algorithms place heavy burdens on systems that often have limited cost and power budgets. Thus, DSP code usually requires heavy optimization and careful use of memory and other resources. In the early days of DSP, there was only one way to meet these goals: Use DMA to bring data into internal SRAM (caches were rare) and then process it in hand-coded assembly (many early DSPs only had an assembler!).Today, DSP algorithms are implemented on many different classes of hardware, often with caches and multiple processing cores. You can program in high-level languages like C, low-level assembly, or a mixture of both. This presents developers with a much more complex ...