I have noticed that some operators run on single core in GPT when used in a graph, while they run on multiple cores when called from SNAP or as a single operator in GPT (e.g. ‘gpt operator_name’).
I have found this issue on the ‘c2rcc.msi’ and the ‘BiophysicalOp’, unfortunately I did not have time to test other operators. When using the --diag option for the gpt command, I can check all the cores on the machine are available.
Initially I though the problem could be related to the resampling operator I have included in the graph before the above mentioned operators. After removing all the operators from the graph, in order to have it clean like Read -> Operator -> Write, still a single core was used.
Hope the developers may find a solution for this issue.