If you're doing a scientific workload where you need to process all your vectors before moving on to the next part of the simulation, you're still better off mu

If you're doing a scientific workload where you need to process all your vectors before moving on to the next part of the simulation, you're still better off multithreading your vectorized operations even with the throttling.