Cinebench is a great predictor of what you will see inside Cinema4D(with both AMD and Nvidia cards). Only things it can not predict is the behavior of the GPUs using large scenes/textures, but for average scenes or simple one like in the car benchmark what you see is exactly what you'll get in your viewport in everyday job.
It is funny to see how some people here prefers gaming benchmarks instead of professional bench for predicting OpenGL speed in pro apps. Gaming benchmarks just shows precomputed stuff, while in 3d editing softwares everything has to be processed by the CPU at first, and that's why OpenGl tests like Cinebench are so dependent by the CPU speed.
All GPUs installed on old MP are bottlenecked by the slow CPU performance for viewport tasks so they will only reach a fraction of their potential and it's no surprise that all reach about the same score when installed on the same machine. To see some real gains in your viewport you have to increase the single thread speed of the CPU(one of the biggest limits of the old MP). Of course this is only for viewport editing in 3d software, for most other usage(like running OpenCL/CUDA) their performance should be mostly unaffected by the slow single thread of old machines.
I've see plenty of people upgrading GPU hoping to get a viewport boost in C4d or other 3D packages, but eventually all of them do not get the big speedup they where hoping for..(except for the more memory of newer GPUs that let them handle larger scenes, but slowly). Also be aware that different drivers can speedup or slowdown things considerably inside Cinema4d, so better to choose recommended drivers.