Accelerating Python

On Linux and MacOS we can count the number of lines in a file using the wc command.

wc cyslow.c

   6967 cyslow.c

• cyslow.c has nearly 7000 lines of C code!

We can then measure the speed of the functions in a Jupyter notebook via;

import cyslow

(ids, varieties, data) = cyslow.load_and_parse_data(5)

timeit(cyslow.load_and_parse_data(5))

3.09 ms ± 19.4 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)

timeit(cyslow.calculate_scores(data))

3.1 s ± 66 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)

scores = cyslow.calculate_scores(data)

timeit(cyslow.get_index_of_best_score(scores))

3.63 µs ± 19.1 ns per loop (mean ± std. dev. of 7 runs, 100,000 loops each)

• The speed of the cython-compiled functions is almost identical to that of the original Python functions from slow.py. cython does not appear to have sped up the code at all.

time python cyslow_main.py

python cyslow_main.py  3.38s user 0.14s system 111% cpu 3.141 total

• The total runtime of the script is almost identical to that of the original Python script.

Back