Don’t think the gap is that big, this test estimates 9.83mm^2 for B35AM, and it’s reasonable to estimate around 4mm^2 for dedomed 519A. So the actual ratio is closer to 16/9.83 = 1.63.
I am very curious where the efficiency differences come from, and how much of it is at the emitter vs at the driver. The emitter should certainly account for some, since 4x519A has more LES (thus lower power density) AND better heatsinking per unit area.