There are 2 large families of 3D rendering engines: CPU and GPU based.
In the 1st one, the processor is responsible for the rendering process whereas, in the 2nd, the video cards are responsible.
The most obvious distinction between the 2 ways is that the number of operations that can be performed simultaneously: a CPU can do from tens to hundreds of operations at the same time, whereas a GPU will do up to thousand different ones.
The second distinction is about the utilization of memory. this is often the weakest point of GPU engines.
They use VRAM video cards that reach a max. of 32 GB in the case of the most expensive nVidia.
CPU engines, on the other hand, benefit from the of RAM memory. Even a daily computer are often easily expanded up to over 64 GB.
This can cause some limitations: the more memory is accessible, the greater polygons can be rendered within the scene; the higher the quality of the textures, the greater the size of the rendering and also the wider the quality of solutions like fur, displacement, motion blur, etc.
Even for compatibility reasons, render CPUs have some benefits. they can be adapted to any kind of processor on the market. On the contrary, some GPU renderers use only CUDA technology which can only be applied to nVidia video cards and not to AMD / ATI cards.
In the end, we are able to say that for a CPU based rendering engine you must have a processor, a motherboard and a high-performance RAM.
However, you'll be able to save on the video card deals basically only with the viewport view. within the case of GPU based engines, the processor is used to load scenes and geometry.
The motherboard might not be the best, however, the recommendation is to shop for one with many PCIexpress slots, just in case you would like to upgrade the system later on by adding new video cards.
For this reason, investing in a GPU engine is generally more expensive, at least initially we've considered 2 engines so far, VRAY that is CPU based and octane which is GPU based.