Overview

Teaching: 10 min
Exercises: 0 min

Questions

• What does it mean to profile a program?

• What are the different types of profiling techniques?

• How can profiling information be used?

Objectives

• Learn about different ways to profile a program.

• Understand how profile information can be used to improve a program.

In order to optimize a program, it is essential to understand where the bottlenecks are. These are the places where the program is spending most of its time. A profile is a set of statistics that describes how often and for how long various parts of the program executed.

Deterministic profiling is meant to reflect the fact that all function call, function return, and exception events are monitored, and precise timings are made for the intervals between these events (during which time the user’s code is executing). Deterministic profiles are obtined by instrumenting the program - inserting intructions into the program that collect this timing information.

Statistical profiling is a process that randomly samples the effective instruction pointer, and deduces where time is being spent. This technique traditionally involves less overhead as the code does not need to be instrumented, but provides only relative indications of where time is being spent.

In Python, there is an interpreter active during execution, so the presence of instrumented code is not required to do deterministic profiling. Python automatically provides a hook (optional callback) for each event. In addition, the interpreted nature of Python tends to add so much overhead to execution, that deterministic profiling only adds small processing overhead in typical applications. The result is that deterministic profiling is not expensive, yet provides extensive run time statistics about the execution of a Python program.

Call counts (i.e. how many times a function is called) and profiling statistics can be used for a variety of purposes:

1. Unusual count numbers can help identify bugs in code.
2. High call counts can help to identify possible points where in-lining (unwrapping loops) might benefit.
3. Internal time statistics can be used to identify “hot loops” that should be carefully optimized.
4. Cumulative time statistics can be used to identify high level errors in the selection of algorithms.

Python has two standard modules that provide the same profiling interface: cProfile and profile (a third, called hotshot, is no longer maintained). The cProfile module has the lowest overhead, but because it is written in C, may not be as widely available. The profile module is written in Python, so has a much higher overhead, but is easier to extend. There is also a newer module, called line_profiler that profiles on a line-by-line basis. This module is not part of the standard distribution, so needs to be installed separately.

Key Points

• Profiling provides useful information on where to focus optimization efforts.

• Different techniques provide a variety of useful information.

• Profiling can have an impact on the performance of the program.