2.4_formats_and_performance energy monitoring

[jzvolensky]

Looking into codecarbon as a tool to collect data about the energy use of functions/processes in the exercises.

Codecarbon collects metrics into a csv like so

@track_emissions(offline=True, log_level="error", country_iso_code="ITA")
def whatever(size: int, iterations: int) -> np.ndarray:
    """
    Multiply two random matrices of size x size

    Args:
    size (int): size of the matrices
    """
    for _ in range(iterations):
        matrix1 = np.random.rand(size, size)
        matrix2 = np.random.rand(size, size)
        result = np.dot(matrix1, matrix2)
    return result

Now we can have a util to read this csv and output HTML in the notebook. Something like this but with Images or icons.

This requires an extra function run to read and generate the emission output.

Discussion point 1: I think that to make this seamless we could fork codecarbon and add a parameter to print this html directly:

https://github.com/mlco2/codecarbon/blob/master/codecarbon/emissions_tracker.py#L1023

This would mean that the only change from the user perspective is the decorator on top of a function

Point 1 has been dropped as it is not an issue

Discussion point 2: I think we could also make the comparisons fun using unusual items or such. Coffee cups is a fine baseline but I think we can make this more entertaining. Suggestions welcome!