Usage: jcmwave_daemon_add_slurm_nodes(options)
Purpose: registers computing nodes (tasks) when running within a SLURM session
Input:
options: key-value list (or matlab) structure to configure the output
options.TasksIds -> List of task identifiers that should form the computing resource.
default: use all tasks
Within a SLURM session a task might be considered as a resource for
starting a process. Besides the number of tasks the SLURM
controller also sets the number of actual (physical) compute nodes (nNodes).
We consider a distribution of tasks across the nodes as follows:
The task with taskId in [0, ... nTask-1] runs on node taskId % nNodes
options.Cluster -> use the tasks to form a cluster for running a single simulation.
Output: resource identifier (integer vector form multiple used resources)
Warning: Adding a machine recursively increases the number of
simultaneously running jobs on this machine.