| Name: | NUM_CORES |
| Type: | integer |
| Default Value: | 1 |
| Range: | [1 – 2.e9] |
Number of MPI processes to spawn. If your problem is reformulated to a MIP, the MPI workers go to sleep and the engine will instead use this setting as the number of threads for CPLEX/CBC, etc.
Note that this also implies that if your problem is linearised you will not gain benefits from a distributed architecture, as MIP solvers don’t work over multiple machines.