According to the documentation, max_parallelism when set to 0 or a negative value should set the parallelism to the number of cores in the machine. This doesn’t seem to be the case. By default it’s set to 1, and when I run a query that is roughly reflected in the CPU usage. If I then manually set it to 4, the query speed increases, as does the CPU usage. If I set it to 0, or -1, the same query speed and CPU usage as max_parallelism=1 occurs. Is the documentation correct, is this a bug, or is my interpretation incorrect? It’s a useful feature.