Code profiling for runtime

Writing Efficient Python Code

Logan Thomas

Scientific Software Technical Trainer, Enthought

Code profiling

  • Detailed stats on frequency and duration of function calls
  • Line-by-line analyses
  • Package used: line_profiler
pip install line_profiler
Writing Efficient Python Code

Code profiling: runtime

heroes = ['Batman', 'Superman', 'Wonder Woman']

hts = np.array([188.0, 191.0, 183.0])

wts = np.array([ 95.0, 101.0,  74.0])
Writing Efficient Python Code
def convert_units(heroes, heights, weights):

    new_hts = [ht * 0.39370  for ht in heights]
    new_wts = [wt * 2.20462  for wt in weights]

    hero_data = {}

    for i,hero in enumerate(heroes):
        hero_data[hero] = (new_hts[i], new_wts[i])

    return hero_data

convert_units(heroes, hts, wts)

{'Batman': (74.0156, 209.4389),
 'Superman': (75.1967, 222.6666),
 'Wonder Woman': (72.0471, 163.1419)}
Writing Efficient Python Code

Code profiling: runtime

%timeit convert_units(heroes, hts, wts)

3 µs ± 32 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)
Writing Efficient Python Code
%timeit new_hts = [ht * 0.39370  for ht in hts]

1.09 µs ± 11 ns per loop (mean ± std. dev. of 7 runs, 1000000 loops each)

%timeit new_wts = [wt * 2.20462  for wt in wts]

1.08 µs ± 6.42 ns per loop (mean ± std. dev. of 7 runs, 1000000 loops each)

%%timeit
hero_data = {}
for i,hero in enumerate(heroes):
    hero_data[hero] = (new_hts[i], new_wts[i])

634 ns ± 9.29 ns per loop (mean ± std. dev. of 7 runs, 1000000 loops each)
Writing Efficient Python Code

Code profiling: line_profiler

Using line_profiler package

%load_ext line_profiler

Magic command for line-by-line times

%lprun -f
Writing Efficient Python Code

Code profiling: line_profiler

Using line_profiler package

%load_ext line_profiler

Magic command for line-by-line times

%lprun -f convert_units
Writing Efficient Python Code

Code profiling: line_profiler

Using line_profiler package

%load_ext line_profiler

Magic command for line-by-line times

%lprun -f convert_units convert_units(heroes, hts, wts)
Writing Efficient Python Code

%lprun output

%lprun -f convert_units convert_units(heroes, hts, wts)

alt=”Magic command lprun tabular output summarizing runtime profiling statistics”

Writing Efficient Python Code

%lprun output

%lprun -f convert_units convert_units(heroes, hts, wts)

alt=”Magic command lprun tabular output summarizing runtime profiling statistics with Line # and Hits columns highlighted”

Writing Efficient Python Code

%lprun output

%lprun -f convert_units convert_units(heroes, hts, wts)

alt=”Magic command lprun tabular output summarizing runtime profiling statistics with Time column highlighted”

Writing Efficient Python Code

%lprun output

%lprun -f convert_units convert_units(heroes, hts, wts)

alt=”Magic command lprun tabular output summarizing runtime profiling statistics with first line Time unit highlighted”

Writing Efficient Python Code

%lprun output

%lprun -f convert_units convert_units(heroes, hts, wts)

alt=”Magic command lprun tabular output summarizing runtime profiling statistics with line three highlighted”

Writing Efficient Python Code

%lprun output

%lprun -f convert_units convert_units(heroes, hts, wts)

alt=”Magic command lprun tabular output summarizing runtime profiling statistics with Per Hit column highlighted”

Writing Efficient Python Code

%lprun output

%lprun -f convert_units convert_units(heroes, hts, wts)

Magic command lprun tabular output summarizing runtime profiling statistics with line nine highlighted

Writing Efficient Python Code

%lprun output

%lprun -f convert_units convert_units(heroes, hts, wts)

alt=”Magic command lprun tabular output summarizing runtime profiling statistics with % Time column highlighted”

Writing Efficient Python Code

%lprun output

%lprun -f convert_units convert_units(heroes, hts, wts)

alt=”Magic command lprun tabular output summarizing runtime profiling statistics with Line Contents column highlighted”

Writing Efficient Python Code
%timeit convert_units convert_units(heroes, hts, wts)

3 µs ± 32 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)

%lprun -f convert_units convert_units(heroes, hts, wts)

alt="Magic command lprun tabular output summarizing runtime profiling statistics with second line Total time highlighted"

Writing Efficient Python Code

Let's practice your new profiling powers!

Writing Efficient Python Code

Preparing Video For Download...