CUDA kernel failed when infer

[YuxiangChai]

I got this error when I tried to run infer.py:

CUDA kernel failed : no kernel image is available for execution on the device
void group_points_kernel_wrapper(int, int, int, int, int, const float*, const int*, float*) at L:38 in /mnt/data/mmlab_ee/yxchai/TriplaneGaussian/tgs/models/snowflake/pointnet2_ops_lib/pointnet2_ops/_ext-src/src/group_points_gpu.cu

Environment:

• RTX 3090
• CUDA 11.3
• Pytorch 1.12.1

This happens in line model(batch) in infer.py

[jmorgan-wellovate]

I got this error when I tried to run infer.py:

CUDA kernel failed : no kernel image is available for execution on the device
void group_points_kernel_wrapper(int, int, int, int, int, const float*, const int*, float*) at L:38 in /mnt/data/mmlab_ee/yxchai/TriplaneGaussian/tgs/models/snowflake/pointnet2_ops_lib/pointnet2_ops/_ext-src/src/group_points_gpu.cu

Environment:

• RTX 3090
• CUDA 11.3
• Pytorch 1.12.1

This happens in line model(batch) in infer.py

Are you using conda or pip install pytorch?

[zouzx]

It seems the problem of pointnet2_ops. Is the pointnet2_ops successfully installed? Maybe this solution from #1 (comment) can help? We have updated the setup.py for auto-detection of TORCH_CUDA_ARCH_LIST, or you can manually set your own TORCH_CUDA_ARCH_LIST at

TriplaneGaussian/tgs/models/snowflake/pointnet2_ops_lib/setup.py

Lines 20 to 21 in d0be649

	os.environ["TORCH_CUDA_ARCH_LIST"] = ".".join(map(str, torch.cuda.get_device_capability()))
	# os.environ["TORCH_CUDA_ARCH_LIST"] = "5.0;6.0;6.1;6.2;7.0;7.5;8.0;8.6"

.

[YuxiangChai]

It seems the problem of pointnet2_ops. Is the pointnet2_ops successfully installed? Maybe this solution from #1 (comment) can help? We have updated the setup.py for auto-detection of TORCH_CUDA_ARCH_LIST, or you can manually set your own TORCH_CUDA_ARCH_LIST at

TriplaneGaussian/tgs/models/snowflake/pointnet2_ops_lib/setup.py

Lines 20 to 21 in d0be649

	os.environ["TORCH_CUDA_ARCH_LIST"] = ".".join(map(str, torch.cuda.get_device_capability()))
	# os.environ["TORCH_CUDA_ARCH_LIST"] = "5.0;6.0;6.1;6.2;7.0;7.5;8.0;8.6"

.

Thanks! This works!