Rendering animation - scene not reloading

[Rik56]

I am trying to render an animation, however it seems that the scene is not reloading between the frames, so I end up with a static image. Is there something I need to check?

I made a very simple scene with just one genesis 2 female character rotating her abdomen, here are the animation settings. Let me know if you need to see more.

http://imgur.com/a/44UAD

Daz 4.9.2.70 Pro, Win 10 64 bit, nvidia drivers 368.81, Octane version 3.01.1b, plugin version 3.1.1.23

[face_off]

Can you post the settings from the Preferences tab of the plugin pls? What does the Octane Log window say at the bottom of the System window? Have you tried with the latest version of the plugin?

Paul

[Rik56]

Ok, I updated the plugin and I also unchecked the "Disabled DAZ viewport" field and it works now. However, since it is a rather large scene, I have to use out-of-core rendering. But after the first frame is rendered, the plugin crashes and I get errors mentioned below. Is it possible that the out-of-core rendering option is ignored when rendering animation?

18:59:31.089: Failed to allocate page-locked memory
18:59:31.090: CUDA error 2 on device 1: The API was unable to allocate enough memory to perform the requested operation.
18:59:31.090: -> failed to deallocate device memory
18:59:31.090: CUDA error 2 on device 1: The API was unable to allocate enough memory to perform the requested operation.
18:59:31.090: -> could not get memory info
18:59:31.092: CUDA error 2 on device 2: The API was unable to allocate enough memory to perform the requested operation.
18:59:31.092: -> failed to deallocate device array
18:59:31.093: CUDA error 2 on device 2: The API was unable to allocate enough memory to perform the requested operation.
18:59:31.093: -> could not get memory info
18:59:31.094: CUDA error 2 on device 2: The API was unable to allocate enough memory to perform the requested operation.
18:59:31.094: -> failed to deallocate device memory
18:59:31.095: CUDA error 2 on device 2: The API was unable to allocate enough memory to perform the requested operation.
18:59:31.095: -> could not get memory info
18:59:31.095: device 2: failed to upload data texture 28 of context 0
18:59:31.098: CUDA error 2 on device 0: The API was unable to allocate enough memory to perform the requested operation.
18:59:31.098: device 1: failed to upload data texture 28 of context 0
18:59:31.098: Render engine failure occured! (WARNING)
18:59:31.099: -> failed to deallocate device array
18:59:31.099: CUDA error 2 on device 0: The API was unable to allocate enough memory to perform the requested operation.
18:59:31.100: -> could not get memory info
18:59:31.100: CUDA error 2 on device 0: The API was unable to allocate enough memory to perform the requested operation.
18:59:31.100: -> failed to deallocate device memory
18:59:31.101: CUDA error 2 on device 0: The API was unable to allocate enough memory to perform the requested operation.
18:59:31.102: -> could not get memory info
18:59:31.102: device 0: failed to upload data texture 28 of context 0
18:59:34.097: Render engine failure occured! (WARNING)
18:59:37.095: Render engine failure occured! (WARNING)

[face_off]

Is it possible that the out-of-core rendering option is ignored when rendering animation?

I doubt that would be happening. It sounds like you don't have enough VRAM. What are the specs of your card pls?

Paul

[Rik56]

I know I don't have enough Vram, this is why I am using out of core memory. This works for single frame, but then it crashes.

[face_off]

What are the specs of your card pls?

Paul

[Rik56]

2x980 ti 6gb + 980 4gb

[face_off]

2x980 ti 6gb + 980 4gb

May I suggest increasing the Out-Of-Core Memory used (ie. move the Out-of-Core Memory slider on the System tab to the right. As a test, you could also disable rendering with the 980, which would increase your VRAM by 2GB,

Paul

[Rik56]

Alright, there has been some progress. I pushed the out of core memory to 12gb, I was able to render 9 frames but after that the same error. I have a feeling that the amount of frames I can render is proportional to the out of core memory I assign, maybe there is something which is not properly cleared from the memory between frames and eventually it overflows?

Another thing, it won't let me allocate more than about 6gb memory until I open the viewport, but that is minor issue. This is how the system window looks like when the scene loads, there should be plenty memory to go around:

http://imgur.com/a/cJvzj

I will try to push ooc memory as high as I can and see how many I can render.

Here are the error messages I get:

17:06:28.761: +++++ Plugin running.
18:29:17.617: CUDA error 2 on device 1: The API was unable to allocate enough memory to perform the requested operation.
18:29:17.618: -> failed to deallocate device array
18:29:17.619: CUDA error 2 on device 1: The API was unable to allocate enough memory to perform the requested operation.
18:29:17.619: -> could not get memory info
18:29:17.620: CUDA error 2 on device 0: The API was unable to allocate enough memory to perform the requested operation.
18:29:17.622: -> could not get memory info
18:29:17.623: CUDA error 2 on device 1: The API was unable to allocate enough memory to perform the requested operation.
18:29:17.624: Failed to allocate page-locked memory
18:29:17.625: CUDA error 2 on device 0: The API was unable to allocate enough memory to perform the requested operation.
18:29:17.626: -> failed to copy memory to device.
18:29:17.628: CUDA error 2 on device 2: The API was unable to allocate enough memory to perform the requested operation.
18:29:17.628: -> failed to deallocate device memory
18:29:17.629: CUDA error 2 on device 1: The API was unable to allocate enough memory to perform the requested operation.
18:29:17.629: -> could not get memory info
18:29:17.630: device 1: failed to upload data texture 28 of context 0
18:29:17.636: -> failed to deallocate device memory
18:29:17.636: CUDA error 2 on device 2: The API was unable to allocate enough memory to perform the requested operation.
18:29:17.638: -> could not get memory info
18:29:17.638: device 2: failed to upload data texture 28 of context 0
18:29:17.639: Render engine failure occured! (WARNING)
18:29:17.639: device 0: failed to upload data texture 28 of context 0
18:29:17.640: CUDA error 2 on device 0: The API was unable to allocate enough memory to perform the requested operation.
18:29:17.640: -> failed to deallocate device memory
18:29:17.641: CUDA error 2 on device 0: The API was unable to allocate enough memory to perform the requested operation.
18:29:17.641: -> could not get memory info
18:29:17.642: CUDA error 2 on device 0: The API was unable to allocate enough memory to perform the requested operation.
18:29:17.642: -> failed to copy memory to device.
18:29:17.643: device 0: failed to upload data texture 22 of context 0
18:29:17.643: CUDA error 2 on device 0: The API was unable to allocate enough memory to perform the requested operation.
18:29:17.644: -> failed to deallocate device memory
18:29:17.644: CUDA error 2 on device 0: The API was unable to allocate enough memory to perform the requested operation.
18:29:17.645: -> could not get memory info
18:29:20.635: Render engine failure occured! (WARNING)
18:29:23.634: Render engine failure occured! (WARNING)

[abstrax]

I think what is too low is the amount of headroom you reserve when using out-of-core texture. Maybe try a larger value of say 1024MB to make sure that there is enough room to re-allocate geometry buffers, etc..

And maybe try reducing the amount of stuff you are uploading to the GPU in the first place. It usually helps with render speed to.