CUDA Error: 'unspecified launch failure'

[ANaughtySnail]

Dear author, thank you for your contribution to lidar scene render. We are very interested in your code, but we encountered the following error when trying to run the example script.

[INFO] Loaded number of Gaussians: 1080000.
[INFO] Converted number of triangles: 2160000
[INFO] Acceleration structure built.
0%| | 0/240 [00:00<?, ?it/s]Optix Log[2][ERROR]: 'Validation mode caught builtin exception 18446744073709551612
Validation mode detected traversal stack overflow.
Launch index (295,234,0)

Error recording event to prevent concurrent launches on the same OptixPipeline (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error recording resource event on user stream (CUDA error string: unspecified launch failure, CUDA error code: 719)'
/root/outside/EnvGS/submodule/diff-surfel-tracing/trace_surfels.cpp:295 Optix Error: 'Optix Log[2][ERROR]: 'Failed to synchronize with given stream (CUDA error string: unspecified launch failure, CUDA error code: 719)
Validation mode found given stream in erroneous state'
Error during validation mode run'
/root/outside/EnvGS/submodule/diff-surfel-tracing/trace_surfels.cpp:299 CUDA Error: 'unspecified launch failure'
0%| | 0/240 [00:00<?, ?it/s]
Traceback (most recent call last):
File "example/render.py", line 159, in
main()
File "example/render.py", line 127, in main
rgb, dpt, acc, norm, dist, aux, mid, wet = tracer(
File "/usr/local/lib/python3.8/dist-packages/torch/nn/modules/module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
File "/usr/local/lib/python3.8/dist-packages/diff_surfel_tracing/init.py", line 275, in forward
return trace_surfels(self.optix_context,
File "/usr/local/lib/python3.8/dist-packages/diff_surfel_tracing/init.py", line 33, in trace_surfels
return _TraceSurfels.apply(optix_context,
File "/usr/local/lib/python3.8/dist-packages/torch/autograd/function.py", line 506, in apply
return super().apply(*args, **kwargs) # type: ignore[misc]
File "/usr/local/lib/python3.8/dist-packages/diff_surfel_tracing/init.py", line 120, in forward
return out_rgb, out_dpt, out_acc, out_norm, out_dist[..., :1].clone(), out_aux, mid_val, acc_wet
RuntimeError: CUDA error: unspecified launch failure
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.

Optix Log[2][PIPELINE]: 'Error releasing namedConstant's internal resources (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error synching on OptixPipeline event (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error destroying OptixPipeline event (CUDA error string: unspecified launch failure, CUDA error code: 719)'
/root/outside/EnvGS/submodule/diff-surfel-tracing/optix_tracer/optix_wrapper.cpp:209 Optix Error: 'Invalid value'
Optix Log[2][PIPELINE]: 'Error releasing namedConstant's internal resources (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error synching on OptixPipeline event (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error destroying OptixPipeline event (CUDA error string: unspecified launch failure, CUDA error code: 719)'

I am using versions optix-7.7 and cuda-11.8. We look forward to your reply!

[xbillowy]

Hi,

I have tested with:

• NVIDIA driver 530.30.02, CUDA 11.8;
• python==3.10, python==3.11
• torch==2.3.1+cu118
• GCC 9.5.0 (since you've successfully compiled the tracer, maybe this is not a problem, higher versions are also ok)

I guess this may relate to the version of the NVIDIA driver, NVIDIA-OptiX-SDK-7.7.0 requires 530.41 or newer as mentioned in #4.

You can also check if the diff-surfel-tracing is successfully built, there should be a forward.ptx and backward.ptx in the installation directory, namely /usr/local/lib/python3.8/dist-packages/diff_surfel_tracing/ in your case.

[Yang-Xijie]

Hi, I met the same issue with you. When running python example/render.py, the following error appeared:

[INFO] Loaded number of Gaussians: 1080000.
[INFO] Converted number of triangles: 2160000
[INFO] Acceleration structure built.
  0%|                                                                                                  | 0/240 [00:00<?, ?it/s]Optix Log[2][ERROR]: 'Validation mode caught builtin exception 18446744073709551612
Validation mode detected traversal stack overflow.
        Launch index (330,157,0)

Error recording event to prevent concurrent launches on the same OptixPipeline (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error recording resource event on user stream (CUDA error string: unspecified launch failure, CUDA error code: 719)'
trace_surfels.cpp:295 Optix Error: 'Optix Log[2][ERROR]: 'Failed to synchronize with given stream (CUDA error string: unspecified launch failure, CUDA error code: 719)
Validation mode found given stream in erroneous state'
Error during validation mode run'
trace_surfels.cpp:299 CUDA Error: 'unspecified launch failure'
  0%|                                                                                                  | 0/240 [00:00<?, ?it/s]
Traceback (most recent call last):
  File "render.py", line 159, in <module>
    main()
  File "render.py", line 127, in main
    rgb, dpt, acc, norm, dist, aux, mid, wet = tracer(
  File "/cpfs01/user/yangxijie/mc3/envs/envgs/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
    return forward_call(*args, **kwargs)
  File "/cpfs01/user/yangxijie/mc3/envs/envgs/lib/python3.8/site-packages/diff_surfel_tracing/__init__.py", line 275, in forward
    return trace_surfels(self.optix_context,
  File "/cpfs01/user/yangxijie/mc3/envs/envgs/lib/python3.8/site-packages/diff_surfel_tracing/__init__.py", line 33, in trace_surfels
    return _TraceSurfels.apply(optix_context,
  File "/cpfs01/user/yangxijie/mc3/envs/envgs/lib/python3.8/site-packages/torch/autograd/function.py", line 506, in apply
    return super().apply(*args, **kwargs)  # type: ignore[misc]
  File "/cpfs01/user/yangxijie/mc3/envs/envgs/lib/python3.8/site-packages/diff_surfel_tracing/__init__.py", line 120, in forward
    return out_rgb, out_dpt, out_acc, out_norm, out_dist[..., :1].clone(), out_aux, mid_val, acc_wet
RuntimeError: CUDA error: unspecified launch failure
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.

Optix Log[2][PIPELINE]: 'Error releasing namedConstant's internal resources (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error synching on OptixPipeline event (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error destroying OptixPipeline event (CUDA error string: unspecified launch failure, CUDA error code: 719)'
optix_tracer/optix_wrapper.cpp:209 Optix Error: 'Invalid value'
Optix Log[2][PIPELINE]: 'Error releasing namedConstant's internal resources (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error synching on OptixPipeline event (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error destroying OptixPipeline event (CUDA error string: unspecified launch failure, CUDA error code: 719)'
optix_tracer/optix_wrapper.cpp:210 Optix Error: 'Invalid value'
optix_tracer/optix_wrapper.cpp:213 CUDA Error: 'unspecified launch failure'
optix_tracer/optix_wrapper.cpp:214 CUDA Error: 'unspecified launch failure'
optix_tracer/optix_wrapper.cpp:215 CUDA Error: 'unspecified launch failure'
optix_tracer/optix_wrapper.cpp:216 CUDA Error: 'unspecified launch failure'
optix_tracer/optix_wrapper.cpp:217 CUDA Error: 'unspecified launch failure'
optix_tracer/optix_wrapper.cpp:218 CUDA Error: 'unspecified launch failure'
optix_tracer/optix_wrapper.cpp:219 CUDA Error: 'unspecified launch failure'
Optix Log[2][DEVICECTX]: 'Error destroying resource event. (CUDA error string: unspecified launch failure, CUDA error code: 719)'
optix_tracer/optix_wrapper.cpp:220 Optix Error: 'Invalid device context'

I am using a server with NVIDIA A100 and Ubuntu system. Some system info:

$ uname -a
Linux dsw-146382-dcb9fb8d8-47p6r 5.10.134-008.12.kangaroo.al8.x86_64 #1 SMP Tue Jan 21 13:54:01 CST 2025 x86_64 x86_64 x86_64 GNU/Linux

$ nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Wed_Sep_21_10:33:58_PDT_2022
Cuda compilation tools, release 11.8, V11.8.89
Build cuda_11.8.r11.8/compiler.31833905_0

$ nvidia-smi
Mon Mar  3 10:20:28 2025       
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.54.03              Driver Version: 535.54.03    CUDA Version: 12.2     |
|-----------------------------------------+----------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |         Memory-Usage | GPU-Util  Compute M. |
|                                         |                      |               MIG M. |
|=========================================+======================+======================|
|   0  NVIDIA A800-SXM4-80GB          On  | 00000000:00:01.0 Off |                    0 |
| N/A   23C    P0              50W / 400W |      2MiB / 81920MiB |      0%      Default |
|                                         |                      |             Disabled |
+-----------------------------------------+----------------------+----------------------+
                                                                                         
+---------------------------------------------------------------------------------------+
| Processes:                                                                            |
|  GPU   GI   CI        PID   Type   Process name                            GPU Memory |
|        ID   ID                                                             Usage      |
|=======================================================================================|
|  No running processes found                                                           |
+---------------------------------------------------------------------------------------+

diff-surfel-tracing is successfully built, since I found ptx/backward.ptx and ptx/forward.ptx in the build folder and pip's site-packages folder.

Have you solved this issue?

[Yang-Xijie]

I also tried with python 3.11 and pytorch 2.3.1. This time even the ptx/backward.ptx and ptx/forward.ptx were not generated... Though pip said "Successfully installed diff_surfel_tracing-0.0.1"...

[Yang-Xijie]

For Python 3.10 and pytorch 2.3.1, the ptx/backward.ptx and ptx/forward.ptx were generated, but when running render.py, the following error appeared:

[INFO] Loaded number of Gaussians: 1080000.
[INFO] Converted number of triangles: 2160000
[INFO] Acceleration structure built.
  0%|                                                                                | 0/240 [00:00<?, ?it/s]Optix Log[2][ERROR]: 'Validation mode caught builtin exception 18446744073709551612
Validation mode detected traversal stack overflow.
        Launch index (295,234,0)

Error recording event to prevent concurrent launches on the same OptixPipeline (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error recording resource event on user stream (CUDA error string: unspecified launch failure, CUDA error code: 719)'
trace_surfels.cpp:295 Optix Error: 'Optix Log[2][ERROR]: 'Failed to synchronize with given stream (CUDA error string: unspecified launch failure, CUDA error code: 719)
Validation mode found given stream in erroneous state'
Error during validation mode run'
trace_surfels.cpp:299 CUDA Error: 'unspecified launch failure'
  0%|                                                                                | 0/240 [00:00<?, ?it/s]
Traceback (most recent call last):
  File "/cpfs01/user/yangxijie/_projects/250303-2dgs-tracer/EnvGS1/submodule/diff-surfel-tracing/example/render.py", line 159, in <module>
    main()
  File "/cpfs01/user/yangxijie/_projects/250303-2dgs-tracer/EnvGS1/submodule/diff-surfel-tracing/example/render.py", line 127, in main
    rgb, dpt, acc, norm, dist, aux, mid, wet = tracer(
  File "/cpfs01/user/yangxijie/mc3/envs/envgs10/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/cpfs01/user/yangxijie/mc3/envs/envgs10/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
    return forward_call(*args, **kwargs)
  File "/cpfs01/user/yangxijie/mc3/envs/envgs10/lib/python3.10/site-packages/diff_surfel_tracing/__init__.py", line 275, in forward
    return trace_surfels(self.optix_context,
  File "/cpfs01/user/yangxijie/mc3/envs/envgs10/lib/python3.10/site-packages/diff_surfel_tracing/__init__.py", line 33, in trace_surfels
    return _TraceSurfels.apply(optix_context,
  File "/cpfs01/user/yangxijie/mc3/envs/envgs10/lib/python3.10/site-packages/torch/autograd/function.py", line 598, in apply
    return super().apply(*args, **kwargs)  # type: ignore[misc]
  File "/cpfs01/user/yangxijie/mc3/envs/envgs10/lib/python3.10/site-packages/diff_surfel_tracing/__init__.py", line 120, in forward
    return out_rgb, out_dpt, out_acc, out_norm, out_dist[..., :1].clone(), out_aux, mid_val, acc_wet
RuntimeError: CUDA error: unspecified launch failure
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.

Optix Log[2][PIPELINE]: 'Error releasing namedConstant's internal resources (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error synching on OptixPipeline event (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error destroying OptixPipeline event (CUDA error string: unspecified launch failure, CUDA error code: 719)'
optix_tracer/optix_wrapper.cpp:209 Optix Error: 'Segmentation fault

[xbillowy]

Hi, I met the same issue with you. When running python example/render.py, the following error appeared:

[INFO] Loaded number of Gaussians: 1080000.
[INFO] Converted number of triangles: 2160000
[INFO] Acceleration structure built.
  0%|                                                                                                  | 0/240 [00:00<?, ?it/s]Optix Log[2][ERROR]: 'Validation mode caught builtin exception 18446744073709551612
Validation mode detected traversal stack overflow.
        Launch index (330,157,0)

Error recording event to prevent concurrent launches on the same OptixPipeline (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error recording resource event on user stream (CUDA error string: unspecified launch failure, CUDA error code: 719)'
trace_surfels.cpp:295 Optix Error: 'Optix Log[2][ERROR]: 'Failed to synchronize with given stream (CUDA error string: unspecified launch failure, CUDA error code: 719)
Validation mode found given stream in erroneous state'
Error during validation mode run'
trace_surfels.cpp:299 CUDA Error: 'unspecified launch failure'
  0%|                                                                                                  | 0/240 [00:00<?, ?it/s]
Traceback (most recent call last):
  File "render.py", line 159, in <module>
    main()
  File "render.py", line 127, in main
    rgb, dpt, acc, norm, dist, aux, mid, wet = tracer(
  File "/cpfs01/user/yangxijie/mc3/envs/envgs/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
    return forward_call(*args, **kwargs)
  File "/cpfs01/user/yangxijie/mc3/envs/envgs/lib/python3.8/site-packages/diff_surfel_tracing/__init__.py", line 275, in forward
    return trace_surfels(self.optix_context,
  File "/cpfs01/user/yangxijie/mc3/envs/envgs/lib/python3.8/site-packages/diff_surfel_tracing/__init__.py", line 33, in trace_surfels
    return _TraceSurfels.apply(optix_context,
  File "/cpfs01/user/yangxijie/mc3/envs/envgs/lib/python3.8/site-packages/torch/autograd/function.py", line 506, in apply
    return super().apply(*args, **kwargs)  # type: ignore[misc]
  File "/cpfs01/user/yangxijie/mc3/envs/envgs/lib/python3.8/site-packages/diff_surfel_tracing/__init__.py", line 120, in forward
    return out_rgb, out_dpt, out_acc, out_norm, out_dist[..., :1].clone(), out_aux, mid_val, acc_wet
RuntimeError: CUDA error: unspecified launch failure
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.

Optix Log[2][PIPELINE]: 'Error releasing namedConstant's internal resources (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error synching on OptixPipeline event (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error destroying OptixPipeline event (CUDA error string: unspecified launch failure, CUDA error code: 719)'
optix_tracer/optix_wrapper.cpp:209 Optix Error: 'Invalid value'
Optix Log[2][PIPELINE]: 'Error releasing namedConstant's internal resources (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error synching on OptixPipeline event (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error destroying OptixPipeline event (CUDA error string: unspecified launch failure, CUDA error code: 719)'
optix_tracer/optix_wrapper.cpp:210 Optix Error: 'Invalid value'
optix_tracer/optix_wrapper.cpp:213 CUDA Error: 'unspecified launch failure'
optix_tracer/optix_wrapper.cpp:214 CUDA Error: 'unspecified launch failure'
optix_tracer/optix_wrapper.cpp:215 CUDA Error: 'unspecified launch failure'
optix_tracer/optix_wrapper.cpp:216 CUDA Error: 'unspecified launch failure'
optix_tracer/optix_wrapper.cpp:217 CUDA Error: 'unspecified launch failure'
optix_tracer/optix_wrapper.cpp:218 CUDA Error: 'unspecified launch failure'
optix_tracer/optix_wrapper.cpp:219 CUDA Error: 'unspecified launch failure'
Optix Log[2][DEVICECTX]: 'Error destroying resource event. (CUDA error string: unspecified launch failure, CUDA error code: 719)'
optix_tracer/optix_wrapper.cpp:220 Optix Error: 'Invalid device context'

I am using a server with NVIDIA A100 and Ubuntu system. Some system info:

$ uname -a
Linux dsw-146382-dcb9fb8d8-47p6r 5.10.134-008.12.kangaroo.al8.x86_64 #1 SMP Tue Jan 21 13:54:01 CST 2025 x86_64 x86_64 x86_64 GNU/Linux

$ nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Wed_Sep_21_10:33:58_PDT_2022
Cuda compilation tools, release 11.8, V11.8.89
Build cuda_11.8.r11.8/compiler.31833905_0

$ nvidia-smi
Mon Mar 3 10:20:28 2025
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.54.03 Driver Version: 535.54.03 CUDA Version: 12.2 |
|-----------------------------------------+----------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+======================+======================|
| 0 NVIDIA A800-SXM4-80GB On | 00000000:00:01.0 Off | 0 |
| N/A 23C P0 50W / 400W | 2MiB / 81920MiB | 0% Default |
| | | Disabled |
+-----------------------------------------+----------------------+----------------------+

+---------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=======================================================================================|
| No running processes found |
+---------------------------------------------------------------------------------------+
diff-surfel-tracing is successfully built, since I found ptx/backward.ptx and ptx/forward.ptx in the build folder and pip's site-packages folder.

Have you solved this issue?

Hi,
I haven't tested on A100, since A100 has no RT cores for ray-tracing acceleration, I would suggest testing on 3090/4090/5090 for better performance.

[Burningdust21]

For Python 3.10 and pytorch 2.3.1, the ptx/backward.ptx and ptx/forward.ptx were generated, but when running render.py, the following error appeared:

[INFO] Loaded number of Gaussians: 1080000.
[INFO] Converted number of triangles: 2160000
[INFO] Acceleration structure built.
  0%|                                                                                | 0/240 [00:00<?, ?it/s]Optix Log[2][ERROR]: 'Validation mode caught builtin exception 18446744073709551612
Validation mode detected traversal stack overflow.
        Launch index (295,234,0)

Error recording event to prevent concurrent launches on the same OptixPipeline (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error recording resource event on user stream (CUDA error string: unspecified launch failure, CUDA error code: 719)'
trace_surfels.cpp:295 Optix Error: 'Optix Log[2][ERROR]: 'Failed to synchronize with given stream (CUDA error string: unspecified launch failure, CUDA error code: 719)
Validation mode found given stream in erroneous state'
Error during validation mode run'
trace_surfels.cpp:299 CUDA Error: 'unspecified launch failure'
  0%|                                                                                | 0/240 [00:00<?, ?it/s]
Traceback (most recent call last):
  File "/cpfs01/user/yangxijie/_projects/250303-2dgs-tracer/EnvGS1/submodule/diff-surfel-tracing/example/render.py", line 159, in <module>
    main()
  File "/cpfs01/user/yangxijie/_projects/250303-2dgs-tracer/EnvGS1/submodule/diff-surfel-tracing/example/render.py", line 127, in main
    rgb, dpt, acc, norm, dist, aux, mid, wet = tracer(
  File "/cpfs01/user/yangxijie/mc3/envs/envgs10/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/cpfs01/user/yangxijie/mc3/envs/envgs10/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
    return forward_call(*args, **kwargs)
  File "/cpfs01/user/yangxijie/mc3/envs/envgs10/lib/python3.10/site-packages/diff_surfel_tracing/__init__.py", line 275, in forward
    return trace_surfels(self.optix_context,
  File "/cpfs01/user/yangxijie/mc3/envs/envgs10/lib/python3.10/site-packages/diff_surfel_tracing/__init__.py", line 33, in trace_surfels
    return _TraceSurfels.apply(optix_context,
  File "/cpfs01/user/yangxijie/mc3/envs/envgs10/lib/python3.10/site-packages/torch/autograd/function.py", line 598, in apply
    return super().apply(*args, **kwargs)  # type: ignore[misc]
  File "/cpfs01/user/yangxijie/mc3/envs/envgs10/lib/python3.10/site-packages/diff_surfel_tracing/__init__.py", line 120, in forward
    return out_rgb, out_dpt, out_acc, out_norm, out_dist[..., :1].clone(), out_aux, mid_val, acc_wet
RuntimeError: CUDA error: unspecified launch failure
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.

Optix Log[2][PIPELINE]: 'Error releasing namedConstant's internal resources (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error synching on OptixPipeline event (CUDA error string: unspecified launch failure, CUDA error code: 719)
Error destroying OptixPipeline event (CUDA error string: unspecified launch failure, CUDA error code: 719)'
optix_tracer/optix_wrapper.cpp:209 Optix Error: 'Segmentation fault

Hi, thanks for your interest. There seems to be a problem with non RT-core GPUs, a workaround is to increase the optix stack in this line:

    OPTIX_CHECK(optixPipelineSetStackSize(*pipeline, direct_callable_stack_size_from_traversal,
                                          direct_callable_stack_size_from_state,
                                          continuation_stack_size,
                                          2));  // Change from 1 to 2

We'll implement a better fix once we identify the root cause.