Anyone else having these issues

[Bulwerk]

We are running on window 8.1 using C4D R16.050

Some times during rendering in the picture viewer Octane will glitch out and save a frame that has not yet finished being sampled and then continue to save that same incomplete frame for the rest of the duration of the sequence.
We have not been able to pin down a specific type of scene or common thread yet. We do a variety of different work from complex scenes to really simple ones and have had this issue with both.

We also have an issue when using the render que and network rendering. We set our scene up exactly as we want it, enabling network rendering and choosing the appropriate slave, then add it to the Que. Once the job launches the slave fails to contribute to the scene. We en up having to open the render setting and enable it as the render is already running.

Thanks for anyone with tips or advice!

[aoktar]

Which version are you using? You shouldn't see this with latest versions.

[Bulwerk]

Thanks for responding Aoktar,

We are using 2.23.2 RC4

I will try the latest and see if helps

[aoktar]

If you can show me some more concrete info or scenes i would like to contribute for a solution. But it's hard to know what happens now without any data. Maybe some bug/user mistake or hardware error... Try to see log outputs for some Cuda errors or interesting outputs from script->console or on-screen info in render view.

[Bulwerk]

Just got the error again. I am attaching the render log. Still operating on 2.23.2 RC4
Can't share the scene file because confidentiality.

[aoktar]

If you looks to log output you'll see the problem. There is some Cuda errors. These are not a plugin issue. You are exceeding limits of renderer most likely. You should invistigate your scene to find what causes this.

-> failed to deallocate device memory
CUDA error 702 on device 1: The device kernel took too long to execute. This can only occur if timeouts are enabled. The context cannot be used anymore and must be destroyed. All existing device memory allocations from this context are invalid and must be reconstructed.
-> could not get memory info
CUDA error 702 on device 1: The device kernel took too long to execute. This can only occur if timeouts are enabled. The context cannot be used anymore and must be destroyed. All existing device memory allocations from this context are invalid and must be reconstructed.

[Bulwerk]

Hmm interesting. Thanks for helping me with this Aoktar.

The scene is heavy. And we are using 4X TitanZ's. Weird how it is fine sometimes with scene and other times it simply stops.

[aoktar]

Hmm interesting. Thanks for helping me with this Aoktar.

The scene is heavy. And we are using 4X TitanZ's. Weird how it is fine sometimes with scene and other times it simply stops.

Wow quick reply. Yes as i said. Check your free vram while rendering. If you are working on some other software which can spend some extra vram and can cause out of vram. Current render frame will be corrupted when you are out of vram.