Hi there,
I've now 2 rendes slaves, one is a Threadripper 2970wx, the new one is a Ryzen 3950x. Theoretically, the threadripper is faster in multithread, and the ryzen should be faster in single core calculations. Measuring with cinebench this is the case.
Now, I'm rendering an animation. After ca. 20 frames the threadripper is much faster, even at single core, at scene parsing. For one reason, the scene parsing time on the ryzen is gettin longer with every new frame. This is weird, since the ryzen should be much faster in single core.
(The scene pasring time is also getting longer on the threadripper too, but not no much, as on the ryzen.) My worstation is an Intel, the are the parsing times constant.
Does any of you encountered something like this?
EDIT: CPU usage on the 3950x is REALLY low, when scene parsing: i get ca. 5-15% during the parsing process. On the 2 other CPUs, the usage is 45-55%.
Thanks :)