Developer Guide and Reference

Contents

Profile-Guided Optimization (PGO)

Profile-guided Optimization (PGO) improves application performance by shrinking code size, reducing branch mispredictions, and reorganizing code layout to reduce instruction-cache problems. PGO provides information to the compiler about areas of an application that are most frequently executed. By knowing these areas, the compiler is able to be more selective and specific in optimizing the application.
PGO consists of three phases or steps.
  1. Instrument the program. The compiler creates and links an instrumented program from your source code and special code from the compiler.
  2. Run the instrumented executable. Each time you execute the instrumented code, the instrumented program generates a dynamic information file, which is used in the final compilation.
  3. Final compilation. When you compile a second time, the dynamic information files are merged into a summary file. Using the summary of the profile information in this file, the compiler attempts to optimize the execution of the most heavily traveled paths in the program.
See Profile-guided Optimization Options for information about the supported options and Profile an Application for specific details about using PGO from the command line.
PGO provides the following benefits:
  • Use profile information for register allocation to optimize the location of spill code.
  • Improve branch prediction for indirect function calls by identifying the most likely targets. Some processors have longer pipelines, which improves branch prediction and translates into high performance gains.
  • Detect and do not vectorize loops that execute only a small number of iterations, reducing the run time overhead that vectorization might otherwise add.
Interprocedural optimization (IPO) and PGO can affect each other; using PGO can often enable the compiler to make better decisions about inline function expansion , which increases the effectiveness of interprocedural optimizations. Unlike other optimizations, such as those strictly for size or speed, the results of IPO and PGO vary. This variability is due to the unique characteristics of each program, which often include different profiles and different opportunities for optimiza