Just a quick 90 frame test deforming a parametric cylinder with twist deformer, a cached jiggle deformer, and hypernurbs / subdivision surfaces. Same project file rendered in both Maxon Cinema 4D R14 and R15. Both using Physical renderer with "Adaptive" sampler and "Automatic" sampling quality, 1 infinite light with no shadow, 1 omni with soft shadow, and 1 area with area shadow.
Render times for above samples: R14 - 9:05 vs. R15 - 4:44
A few other test results (using an average of time for rendering the first 15 frames ONLY):
PHYSICAL (same as above + phys. motion blur) - R14 - 18.1 sec/frm vs. R15 - 17.27 sec/frm
STANDARD render, results were more equal - R14 - 16 sec/frm vs. R15 - 15 sec/frm
PHYSICAL + Ambient Occlusion (default settings) - R14 - 7.73 sec/frm vs. R15 - 3 sec/frm
Clearly, straight-up physical render wins, even with AO added on, but motion blur and standard render are FAR less dramatic.
All rendered using OS 10.8.4 Retina MBP with 2.6 i7 and 16GB, and C4D as only running application.