Problem with multi-core in scan1 on Windows

[kbroman]

A user on the rqtl2-disc list mentioned a problem with multi-core calculations in scan1():

I've been bench marking the time it takes for me to run a genome scan (scan1) by increase the
number of cores that I want to use. I was quite surprised to find that the quickest was running on 1
core. In short, I ran scan1 with one phenotype and then increasing the number of cores. I got the
following times:

1 core 1 phenotype: 29 sec
2 cores 1 phenotype: 46 sec
10 cores 1 phenotype: 2 min 51 sec.

Is this something you have seen before and are aware of?

I'm doing the analysis on a 32-core processor with 128 GB RAM.