Performance question on rendering to 4 separate monitors vs one wide one?

Consider a very wide angle head mounted display like the StarVR. Imagine it has two screens per eye, so total of four (like StarVR).

From the Panda3D side and from the experience of a software developer, which is more computationally intensive? Rendering to 4 separate screens and feeding to the head mounted display via DisplayPort MST (which supports connecting 4 screens via one cable), or feeding one wide screen and letting the ICs inside the HMD do the cropping before sending to the 4 screens?

If there is not much difference in rendering time, first option is more attractive as saves the head mounted display developer from using expensive video cropping ICs.
Asking because I’m working on a DIY headset and need t build a prototype board for it.

You did not specify, the images of the four screens are different? In fact, the view from the camera is rendered for each monitor. And not the principle of the entire screen.

I am only speculating, but I suspect rendering to four screens/framebuffers is slower than a single screen/framebuffer. However, I believe shaders and fillrate would play a much larger role in performance than the number of screens, which may make any performance impact from four screens fairly negligible. I would recommend doing some profiling/experimenting to find out for sure. One issue you might run into with a single screen is the resolution may exceed your maximum texture size.