The Fiper DRM system distributes workload from the SIMULIA Execution Engine to the SIMULIA Execution Engine stations, which must be running and awaiting work items sent from the SIMULIA Execution Engine server. Each station is assigned a number of available slots based on the concurrency settings, with the default concurrency for each station equal to twice the number of CPUs. Each running work item is assigned to a specific slot on a specific station and, in most cases, will hold the assigned slot for the duration of the work item. Certain work items that are waiting for other work or work items not using local resources on the station (such as design drivers waiting for subflow completion or OS Command components running a grid plug-in) may temporarily relinquish a slot so that it becomes available for other work. If no slots are available on any station that matches the work item’s affinity requirements during the server’s dispatch process, the work item will be queued on the server until an acceptable slot becomes available. The advantages of using the built-in Fiper DRM include very low scheduling overhead, which is particularly valuable for fast-running work items and for mixed workflows using the mixed-mode DRM (see Mixed-Mode DRM and the SIMULIA Execution Engine). However, Fiper DRM does not consider relative machine speed or dynamic scheduling parameters such as machine load, available memory, or available disk space when selecting a station; it considers each available slot to be equivalent. As a result, Fiper DRM does not always provide ideal scheduling for workflows containing long-running, resource-intensive work items. For such workflows, the Platform LSF DRM option (see LSF DRM and the SIMULIA Execution Engine) or the Mixed-Mode DRM option (see Mixed-Mode DRM and the SIMULIA Execution Engine) may be appropriate. | |||||||