>>720373226
NTA but by my understanding this is true on contemporary gpus as all the framegen work is offloaded to tensor cores which are *not* doing the actual upfront rendering save for raytrace slop