Other Versions

pythonperformanceoptimizationprofiling

Abstraction: Practical Python performance optimization techniques and profiling tools

Key points:

Profile before optimizing; use cProfile (C implementation, faster than profile) and timeit module introduced in Python 2.3
String concatenation: use "".join(list) instead of repeated +=; immutable strings make concatenation O(n^2) otherwise
Loop optimization: map() and list comprehensions push loop into C; cache method references outside loops to avoid repeated attribute lookup
Generator expressions (Python 2.4+) avoid building full list in memory; use with sum(), min(), max()
xrange over range in Python 2 avoids allocating all integers at once; Python 3 range is lazy by default
Import statements inside functions add overhead when called repeatedly; lazy imports with module-level sentinel avoid redundant work

Connections: Python · Python Performance · Profiling · Optimization