Download as PDF, PPTX
Uploaded byAnoop Thomas Mathew
Faster Python
The document provides 14 tips for optimizing Python code for speed. It recommends profiling code with %timeit and %prun in IPython, using iterators and generators when possible, using list comprehensions over for loops, xrange over range, map, filter and reduce, minimizing function calls and global variables, using threads for I/O-bound processes, and using the C versions of Python libraries when performance is critical. It also provides links to additional resources on Python optimization.
More Related Content
Faster Python
- 1. Faster Python Code in 5minutes Anoop Thomas Mathew Profoundis Labs @atmb4u 31st August, 2013 PyCon India 2013
- 2. ➢@atmb4u ➢ Co-Founder andCTO @ Profoundis Labs ➢ Avid user of open source since 2003 ➢ Python and Django for last 4 years ABOUTME
- 3. Quick Profiling ● %timeit and%prun in ipython ● Try line_profiler ● Not premature optimization TIP1
- 4. use Iterators aspossible ● Iterators and Generators are memory efficient and scalable ● use itertools as much as possible TIP2
- 5. listcomprehension > for loop> while loop consider numpy and tuples for huge data TIP3
- 6. xrange instead ofrange C version of range, designed to work with int TIP4
- 7. map is awonderful idea, so is reduce and filter map(function, data) TIP5
- 8. Function calls arecostly ● pass entire list to a function than iterating on each element and calling the function ● Recursive loops are dangerous with quadratic complexity TIP6
- 9. Local variables arefaster they are faster than global variables, builtins and attribute lookups TIP7
- 10. delete the unnecessary ● useproper namespacing ● weakref is a good idea ● use gc.collect() to expicitly collect the unused variables TIP8
- 11. Threads for I/O ctypesfor CPU(GIL) ● use threads for I/O bound processes. ● For CPU bound with GIL lock, use gevent or C extensions to override it TIP9
- 12. cProfile cStringIO cPickle ... use Cversions of library. They can be 100x faster. TIP10
- 13. Standard Library & builtin functions arealways faster TIP11
- 14. use try thanif; given mostly True (exeptions exists: builtin functions are always faster) NB: Bad practice! TIP12
- 15. immutable types whenpossible TIP13
- 16. ● ''.join('<string>') is fasterthan + ●insert in a list is O(n) ● Faster lookup in dict, set ●deque for double sided operations TIP14
- 17.
- 18.