0
$\begingroup$

I'm rendering 10 frames of a scene both on a Dell XPS 15 Notebook and my Desktop PC.

Desktop: i5-2500K + GTX 1070 8GB

Notebook: i7-8750H + GTX 1050-Ti mobile

Screenshot of the scene

Desktop:

CPU (i5-2500K): 2:34 min

GPU (GTX 1070): 0:37 min

Notebook:

CPU (i7-8750H): 1:24 min

GPU (GTX 1050-Ti): 0:51 min

Though as expected the 1070 is faster, I would expect a much bigger difference, as the 1070 shows around 3x the performance of the 1050-Ti Mobile in various benchmarks.

Is it possible, that the slower CPU in the PC bottlenecks the render speed, although it's not enabled as a CUDA device?

$\endgroup$
  • $\begingroup$ do you have both the CPU and the GPU checked under CUDA in the cycles render device in user preferences? If so uncheck CPU and render again. $\endgroup$ – David Oct 5 '19 at 14:11
  • $\begingroup$ No, GPU only, but I have tried hybrid rendering and it is indeed even slower on my Desktop, so I'm not using it. Good suggestion though, thanks. $\endgroup$ – couka Oct 5 '19 at 14:20
  • $\begingroup$ Might be that the scene isn't complex enough to see larger differences and most of the time is spend loading textures, kernel and building the BVH. $\endgroup$ – Robert Gützkow Oct 5 '19 at 14:32
  • $\begingroup$ In that case, would it make sense to increase the number of frames rendered to see a bigger difference? Or is loading all this stuff per-frame anyways? $\endgroup$ – couka Oct 5 '19 at 14:35
  • $\begingroup$ Yes, rendering more frames would reduce the impact of the initial loading on the whole render time (assuming that the time includes the loading). $\endgroup$ – Robert Gützkow Oct 5 '19 at 15:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.