DBR and Vray performance loss: number of rendering threads

just wondering if anyone can shed light on doing DBR with machines with very different core count, also backburner non-DBR render too. What setting area best for example with a 12 or 16 core machine that is in a network with 4 core machines. Do I have to set it for the lowest core render slave?


thanks

sorry to I am talking about the number of passes in light cache which seems to need to be matched to core count to be optimized?

thanks!

thanks!