Visualize profiling result in Python

Optimizing your program is important. However, knowing where to optimize differentiate good developers and bad ones.

Premature optimization is the root of all evil (or at least most of it) in programming.

— Donald Knuth —

In Python programming language, one can use cProfile module to profile a program to know where the bottleneck is. You can either export the profiled result to a file or have it printed directly in the terminal. The two options are shown below.

python -m cProfile -o result.pstats hello.py
python -m cProfile hello.py

However, sometimes what cProfile shows can be confusing (to me).

ncalls  tottime  percall  cumtime  percall filename:lineno(function)
1127/1    0.007    0.000    8.355    8.355 {built-in method builtins.exec}
     1    0.000    0.000    8.355    8.355 <string>:1(<module>)
     1    0.000    0.000    7.388    7.388 /home/quanhoang/hello.py:1(func)
...

That’s where some visualization tools come into play. There are two available tools for this task: gprof2dot and snakeviz.

Gprof2doz

Installing required packages for gprof2dot.

sudo apt install graphviz
pip install gprof2dot

One can create a graph of execution time either in png or pdf file format, then preview them by appropriate software programs.

gprof2dot -f pstats result.pstats | dot -Tpng -o result.png
eog result.png

gprof2dot -f pstats result.pstats | dot -Tpdf -o result.pdf
evince result.pdf

Snakeviz

Snakeviz, which can be installed via pip, is a browser-based visualization tool. Its usage is shown as follows:

pip install snakeviz --user
snakeviz result.pstats

References

Narendra Kumar Vadapalli, Profiling and visualization tools in Python