the numba in your name is convincing me to give this a try! But good idea - I see the point, If I understand you correctly you say with n_cores==10, doing 50 ba

the numba in your name is convincing me to give this a try! But good idea - I see the point, If I understand you correctly you say with n_cores==10, doing 50 batches that a pool schedules on any free core will reduce the chance of one core taking much longer. I will try this out.