|On this page|
By default PDG runs in-process work items automatically as part of the graph cook, and will run as many in-process items as it can based on the value of
HOUDINI_MAX_THREADS. This scheduler node can be used to manually control how many in process work items can run at the same time. For example, it can be used to limit how many work items in an Invoke cook simultaneously.
The settings on this scheduler are still limited by the value of
HOUDINI_MAX_THREADS. For example, if Houdini itself is configured to use 16 threads then this scheduler is also only able to cook at most 16 work items. This node is mostly used to reduce the number of concurrent in-process work items to a value less than the number of threads in Houdini.
If you want to use an instance of this scheduler to control in-process work items, this scheduler must be set as the TOP Scheduler Override on that node. Otherwise, the node will continue to use the default in-process scheduling. Examples of TOP nodes that create in-process work items include the Invoke, File Copy, and Python Script when Evaluate In Process is toggled on.
This scheduler cannot be used to run out of process work items.
Global parameters for all work items using this scheduler.
The relative directory where the work will be generating intermediate files
and output. The intermediate files will be placed in a subdirectory.
For the Local scheduler or Hqueue, typically
$HIP is used. For other
schedulers, this should be a relative directory to
Local Shared Root Path
Remote Shared Root Path; this path is then appended to these root paths.
Max Work Items
The maximum number of work items that can scheduled at the same time by this scheduler. This number of work items that can actually execute concurrently is affected by the number of threads available in the Houdini session, e.g based on the value of
HOUDINI_MAX_THREADS. When this parameter is set to Equal to CPU Count in-process work items are scheduled automatically as part of the PDG graph evaluation. Otherwise, if a Custom Count value is specified, that value is used to limit the number of concurrent in-process work items.
You can override these parameters per node with the Edit Parameter Interface. For more information, see Scheduler Job Parms / Properties.
When this job parm is enabled, the scheduler will only be able to run one "single" work item at a time. Work items in other nodes without the Single flag set will still be able to run concurrently.