1

I got 2 functions which essentially do the same thing, and I want to compare their runtime performance.

def calculate_bonus_by_table(data, value):
    cdf = data["cdf"]

    # CAUTION: the following loop doesn't do bound checking
    i = 0
    while (value > cdf[i]): i += 1

    return data['bonus'][i]

def calculate_bonus_by_regression(data, value):
    max_slope = len(data['bonus']) - 1
    slope = int((value - 1) / 6)
    if slope > max_slope:
        slope = max_slope

    slope += 1
    return (0.205078125 * slope**2) + (0.68359375 * slope) - 70.888671875

data = json.load(open('bonus.json'))

A snippet of the JSON file used above

{ "cdf": [6, 12, 18, 24, 30, 36, ...], "bonus": [-70, -68, -66, -64, -62, ...] }

In iPython, I time both the function seperately

%timeit calculate_bonus_by_table(data, 199)
1000000 loops, best of 3: 1.64 µs per loop

%timeit calculate_bonus_by_regression(data, 199)
The slowest run took 7.99 times longer than the fastest. This could mean that an intermediate result is being cached.
1000000 loops, best of 3: 604 ns per loop

running timeit for both functions multiple times always gives me similar results. The .._by_regression function result always gives warning about caching.

How can I compare the 2 if one is cached while the other isn't? Why the _by_regression function is cached while _by_table isn't? Will _by_regression performance and caching hold when used in production environment, or should I assume the worst performace (from above, 7.99 * 604ns = 4.83us)?

Thanks

Rick Manix
  • 445
  • 3
  • 7
  • Mostly I've ignored this warning. Newer `ipython` versions` have changed the message to something like `1.05 ms ± 498 ns per loop (mean ± std. dev. of 7 runs, 1000 loops each)`, letting us interpret the `std. dev.` for ourselves. – hpaulj Apr 16 '18 at 23:07
  • Related: https://stackoverflow.com/questions/48733545/ipython-timeit-magic-changing-output-from-mean-std-to-best-of-3 – hpaulj Apr 16 '18 at 23:09
  • Are there anyway to change the message to the new format (mean ± std. dev) with python 2.7? – Rick Manix Apr 20 '18 at 15:56

0 Answers0