I have read at various points that CPU encoding yields better results than GPU encoding, but the experiments I'd done in the past weren't conclusive enough to justify the extra encoding time. I've just started editing drone footage videos (2720x1530, 59.94 CFR) 9XMAS present!), some of which have extensive slow panning, and noticed a definite jerky/"non-smooth" motion. It was so bad that I was disappointed, especially with the TV display.
So as an experiment, I loaded that jerky file into Handbrake and exported it using the default x264 and to my horror, it was smooth.
I then loaded that jerky back into VDL (I'd deleted the source files) and exported using the UHD770, and lo and behold, it was smooth as silk.
So, for this type of source at least, it seems using the iGPU produced a much nicer export than the RTX GPU. I didn't try the CPU itself.
Has anybody found the same?