Hi,
long time ago I have decided to never use the auto setting in the Parallel Chunkl nodes. During the summit I participated in the L3 training for the hub and the conversation came up which or how to chose the right setting.
A participant pointed out that chosing auto resolved any issues and resulted in a presumably balanced processing.
Now with retesting the auto-setting KNime almost immediately froze. I got a Ryzen 7950X with lots of cores but the rendering is still carried out on CPU.
Alongside the processing constrain I wonder if it might be an idea to consider to not trigger the rendering of anything wrapped in close / no opened metanodes respectively componets. All teh animations of the node status, connections getting created and destroyed might choke up the processing.
PS: Counting the prallel chunks + the main one I realized 48 chunks were running in parallel. Knowing I got a 16 code, 32 threads CPU I’d have never chosen anything beyond the available thread. How did / why were the 48 chunks determined in the first place?
Best
Mike