OctaneRender 2023.1 for Houdini production (2023.1.0.0)

[ricky_otoy]

Hello - I dont seem to be able to get complex ngons rendering. Previously i'd click the 'Tessellate NGons' button under 'Tessellation' on the Geo Octane parmeters but it's not having any effect?

Hi stevepickup, please pm me the particular scene file. Thanks!

[ricky_otoy]

Hello - I dont seem to be able to get complex ngons rendering. Previously i'd click the 'Tessellate NGons' button under 'Tessellation' on the Geo Octane parmeters but it's not having any effect?

Hi stevepickup, please pm me the particular scene file. Thanks!

[stevepickup]

Thank you - just sent over now.

Additional info: If I disable the polyextrude it works okay

[Renbry]

Trying to render Batch frames (5 frames per batch) but getting numerous, interesting Cuda crashes across Windows + Linux machines.

Going back to 1 frame batches to try and level this out :S

2023-11-13 11:54:48: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 0: an illegal memory access was encountered
2023-11-13 11:54:48: 0: STDOUT: *** OCTANE API ERROR: -> failed to download symbol(stats data)
2023-11-13 11:54:48: 0: STDOUT: *** OCTANE API ERROR: device 0: path tracing kernel failed

2023-11-13 11:55:13: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 0: an illegal memory access was encountered
2023-11-13 11:55:13: 0: STDOUT: *** OCTANE API ERROR: -> failed to launch kernel(kernel49_simple)
2023-11-13 11:55:13: 0: STDOUT: *** OCTANE API ERROR: device 0: path tracing kernel failed

2023-11-13 11:58:49: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 3: an illegal memory access was encountered
2023-11-13 11:58:49: 0: STDOUT: *** OCTANE API ERROR: -> failed to synchronize event
2023-11-13 11:58:49: 0: STDOUT: *** OCTANE API ERROR: device 3: path tracing kernel failed

2023-11-13 12:25:40: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 3: an illegal memory access was encountered
2023-11-13 12:25:40: 0: STDOUT: *** OCTANE API ERROR: -> failed to launch kernel(kernel37)
2023-11-13 12:25:40: 0: STDOUT: *** OCTANE API ERROR: device 3: path tracing kernel failed
2023-11-13 12:25:40: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 0: an illegal memory access was encountered
2023-11-13 12:25:40: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 1: an illegal memory access was encountered
2023-11-13 12:25:40: 0: STDOUT: *** OCTANE API ERROR: -> failed to launch kernel(kernel49_simple)
2023-11-13 12:25:40: 0: STDOUT: *** OCTANE API ERROR: -> failed to synchronize event
2023-11-13 12:25:40: 0: STDOUT: *** OCTANE API ERROR: device 0: path tracing kernel failed
2023-11-13 12:25:40: 0: STDOUT: *** OCTANE API ERROR: device 1: path tracing kernel failed
2023-11-13 12:25:40: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 2: an illegal memory access was encountered
2023-11-13 12:25:40: 0: STDOUT: *** OCTANE API ERROR: -> failed to synchronize event
2023-11-13 12:25:40: 0: STDOUT: *** OCTANE API ERROR: device 2: path tracing kernel failed
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 0: an illegal memory access was encountered
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 2: an illegal memory access was encountered
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 1: an illegal memory access was encountered
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 3: an illegal memory access was encountered
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: -> failed to copy memory to device.
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: -> failed to copy memory to device.
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: -> failed to copy memory to device.
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: -> failed to copy memory to device.
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: device 0: failed to upload data texture 31 '0:__data11__'
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: device 2: failed to upload data texture 31 '0:__data11__'
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: device 1: failed to upload data texture 31 '0:__data11__'
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: device 3: failed to upload data texture 31 '0:__data11__'
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 0: an illegal memory access was encountered
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 2: an illegal memory access was encountered
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 1: an illegal memory access was encountered
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 3: an illegal memory access was encountered
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: -> failed to deallocate device memory
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: -> failed to deallocate device memory
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: -> failed to deallocate device memory
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: -> failed to deallocate device memory
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 0: an illegal memory access was encountered
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 2: an illegal memory access was encountered
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 1: an illegal memory access was encountered
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: CUDA error (700) on device 3: an illegal memory access was encountered
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: -> could not get memory info
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: -> could not get memory info
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: -> could not get memory info
2023-11-13 12:25:46: 0: STDOUT: *** OCTANE API ERROR: -> could not get memory info

[PDivision1]

Renbry
Can you send me the houdini scene file that is causing these crashes so that we could investigate it?

[Renbry]

Hi,

Apologies - I won't have time to make these scenes self-contained; they're for a holographic concert I'm in the process of trying to deliver (XXTBs of data coming into the scene)

But i did wonder : the only thing fruity that I'm trying to use in 2023 is the analytic lights. Are they dangerous/untested/unreliable enough that I should avoid them ? I'm not used to SO many Cuda crashes over and over and over. It's been a big surprise to come back to Octane since June (Octane 2022) and fight tooth and nail to get farm rendering anywhere near reliable.

Once scene hummed along nicely under Windows but now another scene is misbehaving under Linux + Windows.

I'll keep fisting through this...

[PDivision1]

Renbry
Are your GPU drivers up to date?
Analytic lights shouldn't cause troubles but you can try replacing them with regular ones to see if it's still crashing.

[Renbry]

Across 16 machines they're all on recent drivers and on Linux they're the newest available.

I'm also using the Bake Camera if that rings any bells. This is a holographic concert where I'm baking down lighting to a volumetric recording of the performer.

I'll have to pause everything or wait for this track to finish to go ripping and shredding. I can't turn off Analytic lights without a very different brightness (based on Beta testing them) so I'm stuck until this track is through.

Planning to nuke out the old Ubuntu Linux for Rocky 9 which may give me some help. More soon!

[PDivision1]

Renbry
Could you please check if clean Nvidia driver installation fixes the CUDA errors?

[PDivision1]

One more thing. I think it's always been broken. But there is no way to export undenoised and denoised beauty together. It seems to work in IPR that you have both, but when I export to EXR they both look the same.
AND! Ai upsampler now working, but in IPR only. When I try to render to disk, it doesn't work.

- AI upsampler doesn't work when exporting into EXR. The reason is the model was trained on sRGB input and can't be used with linear sRGB image.
- As for the undenoised and denoised beauty are identical - I could not reproduce this. Can you please provide an example scene?