For what it's worth, I have just conducted a small experiment to try to confirm the effect of (RAW only) x (RAW + JPEG) x (JPEG only), and the effect of SD card speed.

I took six samples, each of ten shots. I extracted the 1/100 second timestamps from the EXIF data, and plotted the nine inter-frame intervals, shown in the graph, as follows:
UHS-1: RAW+JPEG (full size), with a Sandisk Extreme 16GB card rated at 45 MB/s, U-1 C-10
UHS-2 (the 2 is a typo in the graph): RAW+JPEG (full size), with a Sandisk Extreme Pro 16GB card rated at 95 MB/s, U-3 C-10
1080P: JPEG only, 1920x1080, with the UHS-3 card
720x480: JPEG only, 720x480 (the smallest possible), UHS-3
RAW: RAW only, UHS3
R-UHS-1: RAW only, with the UHS-1 card
Conclusions?
There isn't a lot of difference between UHS-1 and UHS-3 SD cards -- none when JPEGs are involved, and a small difference in RAW-only once the buffer is full
The only way to get close to Canon's figure of 4.3 fps is to shoot RAW-only, and even then only for a burst of about seven frames
JPEG size doesn't appear to make much difference (JPEG only)
The only thing that I don't understand is the peak of the second interval when shooting RAW+JPEG