Is there a straightforward way to assign a pandas dataframe to a multiprocessing SharedMemory buffer? Converting the dataframe to numpy arrays first or pickling it is not an option. Also, the dataframe is in the order of gigabytes.