I've enabled GPU rendering (CUDA) for my GTX 1060 and it's awesome. 3D viewport views render much faster than before. Monitoring my GPU shows it at 60-100% usage while Cycles samples.

However, when rendering still images or animations to file, my GPU usage is flat or at zero as CPU usage spikes.

I'm running blender 2.79b on a win 10 system with one GPU.

Yes, I have the latest driver from Nvdia. (399.07)

Yes, I have gone to the Nvdia control panel and set the OpenGL rendering GPU to my card.

*Update. I seem to have located a bottleneck in GPU usage. Looking at MS Process Explorer, it seems that blender is only using, or only has use for part of the GPU..? I'm uncertain if the types of scenes I create is a factor in creating this bottleneck, or if the problem is a solvable blender config issue.

the little engine that could

  • $\begingroup$ Have you also set 'Device' to 'GPU Compute' in the Properties > Render tab > Render panel? (The option is left open, whether to use CPU or GPU, when using the Cycles renderer) $\endgroup$
    – Robin Betts
    Sep 4, 2018 at 7:28
  • $\begingroup$ Yes I have set the Device to GPU Compute and my Feature Set is "Supported." $\endgroup$ Sep 4, 2018 at 8:10
  • $\begingroup$ What is the size of the tiles in the render settings under Performance section? $\endgroup$ Sep 4, 2018 at 11:43
  • $\begingroup$ I've tested values from 64x64 up to 512x512. $\endgroup$ Sep 4, 2018 at 22:47

3 Answers 3


MS Process Explorer does not accurately report GPU usage.


I had this problem back in the day, with 2.79 and my 1080. What I believe the issue was, was a combination of how new gpu rendering for Blender was and the fact that the scene(s) I was rendering were bigger than the GPU memory I had available. The manual even alludes to the problem:

Why does a scene that renders on the CPU not render on the GPU?

There maybe be multiple causes, but the most common is that there is not enough memory on your graphics card. We can currently only render scenes that fit in graphics card memory, and this is usually smaller than that of the CPU. Note that, for example, 8k, 4k, 2k and 1k image textures take up respectively 256MB, 64MB, 16MB and 4MB of memory.

We do intend to add a system to support scenes bigger than GPU memory, but this will not be added soon

So what I think happens is that the viewport is using scaled-down preview textures and models and can squeak by the card's mem limit, but when it comes time to do the full render with full textures it sails past the card's storage capacity. If you are on windows you might see an error in the Commandline (very top -> Window -> "Toggle System Console").

Also, like, GPU rendering was kind of untrustworthy at the time, so be ready to deal with that.


Have you enable the CUDA options on "Cycles Compute Device" setting under the System tab on the Preference setting? If you didn't that may caused your problem enter image description here

  • $\begingroup$ Yes, Bogiva. As stated in the first sentence of my question. $\endgroup$ Sep 4, 2018 at 9:18

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .