I am running some fairly large gam models and don’t want to parallelize the computations, or at least want to be able to control the degree of parallelization. (Besides not wanting to fry my machine with long stretches of full CPU utilization, I also suspect that over-parallelization is actually hurting performance.)