l33tforza :
I would suggest a much more powerful GPU for a render system and possibly look at going SLI - it will help with render times especially if you are using software that can take full advantage of SLI like Octane. you can support to 980 Ti's with the 850 PSU -
In commercial non-realtime 3d products, the GPU is useful for rendering the user interface alone. The GPU is responsible for drawing those wireframes and the preview images in the viewport.
If you want to edit manipulate heavy scenes, model dense meshes or sculpt with high levels of subdivision then a fast graphics board is essential. Because a slow card results in a sluggish user experience for the artist.
But when it comes to the actual rendering process, the GPUs are typically not used at all. In fact the stacks of machines seen in commercial render farms might not have any GPUs installed at all. Just fast multicore processors, and lots of RAM.
So why are GPUs not used?
There's a lot of reasons for this.
Scenes can have arbitrary complexity, which makes writing robust shader code difficult. To efficiently ray trace a scene, a variety of optimisations will kick in depending on the context.
Reflections and refractions require the shader to recurse. I think the stack size on GPUs has limitations.
Textures for GPUs have size and shape limitations.
And rendering add ons benefit from the sheer flexibility of conventional programming languages.
So hardware based rendering is not something you tend to see used in many commercial projects.
There are some products which do exploit GPUs for rendering and post-processing tasks. But these tend to be the exception rather than the rule.
A workstation card would speed things up though