Problem

I use Matplotlib often to make animations with thousands of frames from a data source such as a Pandas Dataframe where each row of the DataFrame corresponds to state in one frame of the animation (very similar to what is done in the pandas_alive package). Sequentially rendering these animations is at times painfully slow. I would like a method to split the rendering across multiple processes to speed up the render time.

Proposed Solution

I would like a new animation API - something like a map function that takes a render function, and iterable container of animation state such as a Dataframe - and a number of jobs/processes to split the rendering across.

Additional context and prior art

I'm mainly inspired to make this request by the joblib package and scikit-learn which has a number of methods that integrate with joblib which can be configured by setting an n_jobs parameter, e.g. GridSearchCV