How to edit and maximize performance.
![nate_c399e9f1](https://secure.gravatar.com/avatar/3229ba32b9386cb13596aef2f06d2254?&r=pg&s=100&d=https%3A%2F%2Fvanillicon.com%2F3229ba32b9386cb13596aef2f06d2254_100.png)
I've been using Daz for about 3 years now, but I haven't touched much of the settings in fear I would break something. In any case, I enabled GPU in render settings alonside CPU and I can get through some scenes pretty well. One thing I noticed though was while my CPU maxes out at 100%, my gpu only gets to about 6-12%. Is this normal? I tested this out on 3 separate computers so I suspect that's the default. If anything, I want to find a way to limit usage on CPU so I can work on other things while increasing GPU performance. Say for instance I want Daz to max out the CPU at 80% while working the GPU at 50% or more. Is this possible and is it a good idea? I run a Ryzen 2600X and dual GTX 980ti not in SLI, soon to upgrade to a RTX 2060.
Comments
If the CPU is at 100% it's likely that the GPU is not being used. However, if you are looking at the Windows Task Manager to assess things it won't, by default, be helpful - you need to tell it to show Compute_0 or CVUDA, if it's on offer, to see the Iray use. What driver do you have - right-click on the desktop>nVidia Control Panel? DS 4.12 needs at least 430.86.
No, once you have the correct drivers that should be it - though it does seem that the version of Iray in 4.12 uses more memory on non-RTX cards than previous versions so you may still have some issues.
As I understand it the RTX card will have less memory used, at least up front - I don't know how much memory the RTX operations use within the GPU.
From both my experiments and others statements, including people posting numbers in the benchmark threads, using both CPU and GPU may make a render take longer than running the render on GPU only.
After updating my Nvidia drivers to the latest version(436.48), I ran two test benches on Daz 4.12 on the same scene.
On the first render, I had my CPU(Ryzen 2700X Stock Clock) and dual GTX 980TIs enabled. The CPU still ran at 100% and the first GTX card jumped up and down from 6% to 15% while the second hovered around 4%, and the whole render took 2 minutes and 57 seconds to complete.
On the second test I disabled the CPU and just used the cards for rendering. This resulted in my CPU running at a constant 25% to 28% while the first GPU kept spiking from 22% to 38% and sometimes 55%, though the second card stayed from 18% to 25%. That render only took 2 minutes and 56 seconds.
Both times the video memory was using 4gb out of 6gb.
Are these good results or are there better performance tweaks I can make to improve my timing?