some unexpected behavior from framedelay(ex9) with switch(node) as input
framedelay (ex9) problem.v4p (7.7 kB)
hey ggml,
just wanted to tell you that we are aware of that one.
the behavior is certainly not quite right.
anyway it turns out that the solution would more or less require a new resource & device management. the current one struggles with some
rare cases which typically are related to cyclic patch structures.
a rework is in the works but needs more time.
just wanted to give a well deseverd feedback (ah!!!) on that…