You can join Blender Cloud for €9.90/month and get access to all of our training & film content instantly!
Login Join Blender CloudWorkflow 05: Rendering the Project
3rd April 2017
info License: CC-BYAuthor
The great debate of CPU vs GPU rendering. I break it down into very simple considerations when deciding on which to use for your project. Spoiler alert: They're both awesome.
4 Comments
Join to comment publicly.
Greg Hodgins
18th February 2018 - 22:43
Rendering:
AMD, Vega has changed this, if you can get one. Through there HBCC, the GPU can expand its Vram to use system ram. HBCC will allow up to 40GB+ per card, if you have the system ram available. That is a game changer for me with Blender, even though I have a 16 core Threadripper. Two Vega 64s are considerably faster. It is beneficial to have both though when using different software. Fusion 9 render software, predominately, has better result but is 99% CPU dependent.
I am limited though, only having 64GB of ram. When ram prices go down, or I can afford it, I will get another 64GB ram. I am eager to use the Radeon Pro Render plugin for Blender. Their nodes are different though and I do not know enough yet to easily cross reference them.
Midge Sinnaeve
3rd March 2018 - 15:06
@Greg Hodgins: Out of core is also being worked on for CUDA rendering, I believe. With the daily builds I've been able to render some insane displacement on GPU. :D
Greg Hodgins
8th March 2018 - 19:21
@Midge Sinnaeve: I originally wanted Titans but cheaped out. I need two more GPUs but not at price right now.
Midge Sinnaeve
13th March 2018 - 14:04
@Greg Hodgins: Yeah, it's pretty crazy at the moment.