Puget Systems print logo

https://www.pugetsystems.com

Read this article at https://www.pugetsystems.com/guides/1864
Jared Lewis (Technology Consultant)

Consultant's Corner: Building a Workstation for Blender

Written on August 12, 2020 by Jared Lewis
Share:

A recent call from a customer came with a unique problem: She needed a system optimized to run Blender, a program that Puget Labs has not yet directly tested. Blender is an open-source application used for a host of content creation, from animation and visual effects, to virtual reality, rendering, and computer games. So how could I help this customer, and design a system for an application that we don’t have our own data for?

First, we turn to the developer and look at their recommendations. Most developers only publish minimum specs, which don’t often result in a good user experience. With Blender, we are fortunate that they not only publish Minimum recommendations, but also Recommended and Optimal in their requirements documentation.

Blender

Second, I use feedback from customers who I have built systems for in the past, who use the application and have reported good performance with their selected components. We also take into account other similar software applications that we have performed testing on. Using all of this information, we get a good picture of how to best build for the application.

Using the Blender requirements link, I walked the customer through the parts selection that would meet these optimal recommendations, and also her budget, offering great performance for the application. We decided on a Puget Spirit, with the following configuration:

  • An Intel Core i9 10900k 10-Core CPU
  • 32GB of system memory
  • The NVIDIA RTX 2080ti 11GB Graphics card
  • A fast SSD hard drive

These components put the system in the optimal category for almost every part, offering good performance for the cost. For optimal performance, Blender recommends an 8-core CPU, so we went with an Intel 10-core model to give her a little headroom. Next, we went with the recommended amount of system memory, which is 32GB. We selected the 2080ti GPU because it provides nearly 3x the amount of VRAM Blender recommends while still allowing for the customer's budget to be distributed among components that directly contribute to performance. An increase to a Quadro P5000 16GB or Quadro RTX 5000 16GB graphics card would exceed the 12GB video memory recommendation and offer increased performance if your budget allows it.

The end result? The customer reported happily that her new Puget Spirit was able to run Blender effectively, giving her a new tool to assist in her workflow, and allow her to focus on doing the animation and design she loves, without worrying about the system’s ability to keep up.

If you have questions about an application not listed on our website, give us a call or fill out the form below!

Tags: Blender, Intel, NVIDIA, Quadro, GeForce
Ampere

https://nvidianews.nvidia.c...
https://www.nvidia.com/en-u...

You should also test & benchmark Nvidia's next-gen GeForce GPUs when launched in September for Blender also. Unfortunately it will require
Blender to update to the latest CUDA Toolkit 11.0.x & Nvidia Optix release for proper support, the current releases does not yet support next-gen GeForce GPUs.

https://developer.nvidia.co...
https://developer.nvidia.co...

Posted on 2020-08-13 07:25:33