samsue wrote:Hallo tcheng00, is everything okay with you?
I've struggled too with these Out-Of-Core related problems, most of the problems are caused by the insufficient memory of the pc.
I recommend you take a look at these topics.
viewtopic.php?f=27&t=74536viewtopic.php?f=27&t=73903&start=10were tense months, unfortunately there is not much material, explaining the limitations, known problems and best practices.
So far, I'm not having too many problems, I get the occasional denoising doesn't have enough memory error, then I check only 1 gpu to do the denoising and the second gpu to do the tone adjustment. Everything seems to be working find again.
Also, to note, the default setting seems not so good for general applications. Please change this ASAP, so you don't get a lot of new user frustrations. Hopefully, there is a dynamic way to automate the technical settings based on hardware and memory requirement so we don't need to deal with these kinda settings in the future.
I recommend changing GPU headroom to a larger amount to get stability (this changes based on memory/scene size, hopefully this will be dynamically automated in the future). I use the max of 1 GB, but I think I can go lower, just don't know the right combination yet. If you have dual GPUs, use one for denoising and the other for Tone. Do not use NVlink for preview while moving the camera around, it gets out of sync and errors out. I think data transfer over nvlink is not faster enough and gets out of sync on a 2070 Super (this is just my guess). You can use NVlink for final render or preview if you don't change the perspective view.
Also, the GI clamp is so high that it produces fireflies. I've set mine to 10 to maintain a bit of dynamic range without creating hot pixels.
Keep in mind, the optimization for lights does not seem to work for Emission lights that well. So if you can avoid using mesh lights, the Vectron geo lights are better and faster.
I'm not exactly an expert in Octane, but that is what I found so far. Please feel free to correct me. I'm still learning