Traditionally, the rendering engines of 3D modelling and rendering software programs uses the central processing unit (aka CPU) to do calculation for rendering. However, for the past couple of years, there seems to be a new trend emerging. It appears that there are now more 3D rendering applications taking the graphics processor unit (aka GPU) as a way of churning out calculations for renderings. GPU refers to the processor units on dedicated graphic cards. Most of the GPU cards supports either one of the following: CUDA, DirectX, Open CL. Currently Nvidia or AMD (Advanced Micro Devices) produces graphic cards that can be used to process calculations for 3D rendering.

The photo below shows a CUDA capable Nvidia GTX card being installed in a desktop computer.

When compared to the CPU, using the graphics processor unit (GPU) of a good video card can result in much faster rendering speed. One determining factor of rendering powess is the number of available cores on the video card. For example, a good Nvidia card can have thousands of cores.

One consideration to take note, though, is that the current limitation of GPU-based rendering technology is the amount of available rams on the graphics card. A CPU have the possibility of being able to utilise much more RAM. On the other hand, as of this writing, most graphics cards have eithor 2, 3 or 4 GB RAM. The highend ones (example, Nvidia GTX Titan Z) with 12GB rams are very expensive.

Even the open source arena, free programs are taking advantage of GPU as well. For example, Blender has Cycles Renderer which now can accept both AMD and Nvidia cards for GPU rendering. In another word, Cycles have take advantage of both CUDA and OpenCL to accelerate renderings.

As for commercial 3D programs, there are GPU renderers such as V-ray, IRAY, Octane Renderer, just to name a few. One can even experience some of these capabilities for free. For example, DAZ Studio, which is free as a base program, comes with IRAY in its latest version. IRAY is a renderer that can take advantage of Nvidia GPU for realistic renderings. Below video shows IRAY in action with Daz Studio.

As can be seen, the digital field and industry is clearly moving towards a GPU-based rendering paradigm.