Fast Python, Slow Python by Alex Gaynor

Transcript

1. Fast Python, Slow Python Alex Gaynor

2. About me • Core developer of Django, PyPy, CPython, etc.

   • Director of the Python Software Foundation • Work at Rackspace (Thanks!)

3. A talk about two things

4. What is performance?

5. Why do we care about performance?

6. Lies, damned lies, and benchmarks

7. Performance is specialization.

8. Systems performance

9. What is Python?

10. What Python Isn’t • Cython • C • Numba •

    RPython

11. Excuses time

12. Dynamic typing

13. You can monkey patch anything

14. Slow vs Harder to optimize

15. PyPy

16. $ python! >>> print 2! 2! $ pypy! >>>> print

    2! 2!

17. Let’s make a deal

18. How can we specialize our code?

19. Let’s talk about C

20. struct Point {! double x;! double y;! double z;! };!

21. class Point(objet):! def __init__(self, x, y, z):! self.x = x!

    self.y = y! self.z = z!

22. point = {! "x": x,! "y": y,! "z": z,! }!

23. Dictionaries vs. Objects

24. std::unordered_map<std::string, double> point;! point["x"] = x;! point["y"] = y;! point["z"]

    = z;!

25. Why don’t we care?

26. if [hex, bytes, bytes_le, fields, int].count(None) != 4:! raise TypeError('need

    exactly one argument')!

27. if [hex, bytes, bytes_le, fields, int].count(None) != 4:! raise TypeError('need

    exactly one argument')!

28. if (! (hex is None) + (bytes is None) +

    (bytes_le is None) +! (fields is None) + (int is None)! ) != 4:! raise TypeError!

29. Python makes everything easier

30. Let’s talk about strings

31. char *data = malloc(1025);! while (true) {! size_t n =

    read(fd, data, 1024);! data[n] = '\0';! char *start = data;! while (start < data + n) {! if (isspace(*start)) {! break;! }! start++;! }! printf("%s\n", start);! }!

32. while True:! data = os.read(fd, 1024)! print data.lstrip()!

33. Zero Buffer

34. from zero_buffer import Buffer! ! b = Buffer.allocate(8192)! with open(path,

    "rb") as f:! b.read_from(f.fileno())! for part in b.view().split(b":"):! part.write_to(sys.stdout.fileno())! sys.stdout.write('\n')!

35. https://zero-buffer.readthedocs.org/ https://warehouse.python.org/project/zero_buffer/

36. A few more myths • Functions calls are really expensive

    • Using only builtin data types will make your code fast • Don’t write Python in the style of Java or C

37. One Python

38. I hate heuristics

39. Let’s take a lesson…

40. Thank you! Questions? https://speakerdeck.com/alex/