tpetrick
Feb. 22, 2022 17:17:23
Yea, a batch will process all of its work items in one process. Rather than cooking each work item individually, the batch itself is scheduled as a single logical job. For both ROP Fetch and HDA Processor batches are processed sequentially, however that's not a strict requirement. For example, if you're writing a custom work item job script that supports batching (or a Python Processor's onCookTask hook) you can process the items in the batch in whatever order you want.
The ROP Fetch node cooks batches by calling rop.render(..) over the whole frame range, so if an error occurs on a frame the entire operation ends early. Likewise for HDA Processor -- the work items are cooked in a loop, and a failure will cause the process to exit immediately.
galagast
Feb. 23, 2022 09:38:49
thank you for the answers and confirmation ^_^