index out of bounds

[pkqpotter123]

I've encountered an array index out-of-bounds issue. Do you happen to know a solution? Thanks a lot. anaconda3/envs/torch2/bin/python RegFormer/demo.py anaconda3/envs/torch2/lib/python3.8/site-packages/torch/functional.py:445: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at /opt/conda/conda-bld/pytorch_1634272068694/work/aten/src/ATen/native/TensorShape.cpp:2157.) return _VF.meshgrid(tensors, kwargs) # type: ignore[attr-defined] /opt/conda/conda-bld/pytorch_1634272068694/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:111: operator(): block: [17,0,0], thread: [0,0,0] Assertion idx_dim >= 0 && idx_dim < index_size && "index out of bounds" failed. /opt/conda/conda-bld/pytorch_1634272068694/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:111: operator(): block: [17,0,0], thread: [1,0,0] Assertion idx_dim >= 0 && idx_dim < index_size && "index out of bounds" failed. /opt/conda/conda-bld/pytorch_1634272068694/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:111: operator(): block: [17,0,0], thread: [2,0,0] Assertion idx_dim >= 0 && idx_dim < index_size && "index out of bounds" failed. /opt/conda/conda-bld/pytorch_1634272068694/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:111: operator(): block: [17,0,0], thread: [3,0,0] Assertion idx_dim >= 0 && idx_dim < index_size && "index out of bounds" failed. /opt/conda/conda-bld/pytorch_1634272068694/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:111: operator(): block: [39,0,0], thread: [32,0,0] Assertion idx_dim >= 0 && idx_dim < index_size && "index out of bounds" failed. /opt/conda/conda-bld/pytorch_1634272068694/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:111: operator(): block: [25,0,0], thread: [25,0,0] Assertion idx_dim >= 0 && idx_dim < index_size && "index out of bounds" failed. Traceback (most recent call last): File "torchRegFormer/demo.py", line 34, in l0_q, l0_t, l1_q, l1_t, l2_q, l2_t, l3_q, l3_t, pc1_ouput, q_gt, t_gt, w_x, w_q = model([pos2], [pos1], T_gt, None, None) File "/data/zzf/anaconda3/envs/torch2/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl return forward_call(*input, *kwargs) File "torchRegFormer/regformer_model.py", line 522, in forward l1_cost_volume = self.cost_volume3(l1_xyz_warp_proj_f1, l1_xyz_proj_f2, l1_points_warp_cross_proj_f1, l1_points_warp_cross_proj_f2) #FE1 File "/data/zzf/anaconda3/envs/torch2/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl return forward_call(input, kwargs) File "torchRegFormer/conv_util.py", line 551, in forward qi_xyz_grouped = torch.gather(xyz2_bn3, 1, neighbor_idx.unsqueeze(-1).repeat(1, 1, 3)) RuntimeError: CUDA error: device-side assert triggered

Process finished with exit code 1

Originally posted by @pkqpotter123 in https://github.com/IRMVLab/RegFormer/issues/4#issuecomment-1846791500

[whqweqwe]

Have you solved this problem?

[pkqpotter123]

I'm sorry, the issue has not been resolved yet. If there are any updates, feel free to communicate at any time.

[liujiuming123]

requirements.txt

[liujiuming123]

We release the docker environment list here. Hopefully it can address your problem.

[whqweqwe]

@pkqpotter123 How did you solve this problem: return F.conv2d(input, weight, bias, self.stride, RuntimeError: cuDNN error: CUDNN_STATUS_INTERNAL_ERROR

[pkqpotter123]

@pkqpotter123 你是如何解决这个问题的： return F.conv2d(input,weight,bias,self.stride, 运行时错误：cuDNN 错误：CUDNN_STATUS_INTERNAL_ERROR

I added the following line in the script: os.environ['CUDA_LAUNCH_BLOCKING'] = '1'。 The error CUDNN_STATUS_INTERNAL_ERROR is a general and non-specific error. The purpose of the code os.environ['CUDA_LAUNCH_BLOCKING'] = '1' is to make CUDA errors more detailed. After adding this line, the error became 'index out of bounds'

[pkqpotter123]

We release the docker environment list here. Hopefully it can address your problem.

Thank you very much for your help.

[whqweqwe]

@pkqpotter123 Yes, but I used the author's requirements. txt and it still exists：perator(): block: [17,0,0], thread: [0,0,0] Assertion idx_dim >= 0 && idx_dim < index_size && "index out of bounds" failed. Have you solved this problem？

[liujiuming123]

We found that the index setting inside our CUDA operator was wrong when we debugged the codes, and we have corrected it now. You can try to compile them again. Hopefully, this can help you well.