Skip to content

Profiling in Python⚑

Generate data⚑

Pythons integrated cProfile⚑

python3 -m cProfile -o profile.pstats -m my_module <args>

Yappi⚑

Yappi supports asynchronous and multithreaded profiling, which is not supported by the built-in profiler.

pip3 install -U yappi

import yappi
from my_module import my_function

yappi.start()

my_function()

yappi.stop()

yappi.get_func_stats().save("profile.pstats", type="pstats")

Visualisation⚑

gprof2dot (Dot Diagram, SVG)⚑

Transform a .pstats file with gprof2dot into a dot graph as a svg file.

pip3 install -U gprof2dot

gprof2dot -f pstats profile.pstats | dot -Tsvg -o profile.svg

Snakeviz (Interactive)⚑

Snakeviz is a web-based profiling tool which allows users to analyse their code by filtering data by module, function and file, and sorting it according to different criteria such as the number of calls or cumulative time spent in a function.

pip3 install -U snakeviz

snakeviz profile.pstats

flamegraph (SVG)⚑

Flame graphs are visual tools that show how much time is spent in each function call. The width of each bar in the graph represents the amount of time spent in that function, with wider bars indicating more time spent and narrower bars indicating less time. The main function is at the bottom, and the subfunctions are stacked vertically on top.

pip3 install -U flameprof

flameprof profile.pstats > profile.svg