just a couple of things:

Min ray samples sets the sampling per shaded point in PBR, - NOT Max ray samples as stated in the docs (I believe Max ray samples is ignored completely since either flavor of PBR doesn't use variance)

another thing which I think could be clearer is that raytraced PBR mode also uses Min ray samples (ie total samples = pixel samples * Min ray samples) - the implied assumption from reading the docs is that its only MPBR that uses the ray sampling and that raytraced PBR is driven purely from pixel samples…

there also seems to be an apparent contradiction in how PBR is described as working and the actual settings on the PBR tab - ie according to Mark Elendt's otherwise excellent master class video here:

http://www.sidefx.com/index.php?option=com_content&task=view&id=1414&Itemid=132 [sidefx.com]

there is no way of differentiating between different kinds of indirect rays fired during PBR sampling, yet the PBR options panel on a Mantra ROP clearly has bounce settings for 3 different types of rays fired (Reflect limit, Glossy Limit and Diffuse Limit)

additionally what these 3 limits actually control could be also be better documented

from limited testing it seems that indirect “diffuse” rays seem to rely on BOTH reflect and diffuse limits, indirect “specular” rays rely on the reflect limit only and that the glossy limit seems to only affect “specular” reflection of light sources, but it would be great to get more information on how this all works - the information that's out there doesn't really reference this at all (and please feel free to RTFM me!)

eg: how does a BSDF actually differentiate between “diffuse” and “reflection” ? (or even does it? is this somehow inferred through some other mechanism?)