probably the problem is the CV643D-A2032 companion.
Try without the A2032.
About the speed, i've used P96 for a life with the CV64 (not 3D) and it only feels faster but it isn't.
Do you mean the slow WB backdrop redrawn under CGX?
It is the only thing, that is slower in CGX respect P96.
Try P96Speed with CGX, and you will see how faster is CGX in all the operation.
Anyway as already said, use the RTG system that "feels" better for yourtaste.