4. Improving performance

This chapter describes various techniques to improve the performance of C code on the C28x CPU. One or more of the techniques described below or a combination of techniques can be used to improve performance. Techniques that are beneficial will depend on the nature of the application and can vary from application to application. For example:

• Executing from RAM illustrates improving performance by placing frequently executed functions in RAM.

• Optimization levels provides an overview of compiler optimization levels.

• Inlining can improve performance in an application with multiple levels of function calls.

• Compiler annotations such as Pragmas, Assertions and Restrict provide additional information to the compiler to help improve the performance of generated code.

• Loop unrolling is a technique to improve the performance of small loops.

• Leveraging DMAC instructions describes how to leverage the Dual Multiply Accumulate (DMAC) instructions from C source.

• 4.1. Memory
  • 4.1.1. Executing from flash
  • 4.1.2. Executing from RAM
  • 4.1.3. Other considerations
• 4.2. Optimization levels
  • 4.2.1. Examples
  • 4.2.2. Code size vs. speed tradeoffs
  • 4.2.3. Optimization levels and debug
  • 4.2.4. Optimizer interlist
• 4.3. Inlining
• 4.4. Pragmas
  • 4.4.1. MUST_ITERATE
  • 4.4.2. UNROLL
  • 4.4.3. FUNC_ALWAYS_INLINE
• 4.5. Assertions
• 4.6. Restrict
  • 4.6.1. Overview
  • 4.6.2. Example
  • 4.6.3. Usage
  • 4.6.4. Incorrect Usage
• 4.7. Loop unrolling
  • 4.7.1. Benefits
  • 4.7.2. Performing unrolling
• 4.8. Leveraging DMAC instructions