search

nv-tlabs / NKSR

[CVPR 2023 Highlight] Neural Kernel Surface Reconstruction
https://research.nvidia.com/labs/toronto-ai/NKSR
Other
747 stars 43 forks source link

RuntimeError: CUDA error: an illegal memory access was encountered #9

Closed xiemeilong closed 1 year ago

xiemeilong commented 1 year ago

I just replaced the ply file with my own ply file, then executed recons_colored_mesh.py, and encountered an error：

06-13 15:34:48 (__init__.py:230) [INFO] Input scale factor: 3.5629 
Traceback (most recent call last):
  File "/mnt/disk2/labs/NKSR/examples/recons_colored_mesh.py", line 29, in <module>
    field = nksr.reconstruct(input_xyz, input_normal, detail_level=1.0)
  File "/mnt/disk2/.conda/envs/nksr/lib/python3.10/site-packages/nksr/__init__.py", line 283, in reconstruct
    output_field.solve(
  File "/mnt/disk2/.conda/envs/nksr/lib/python3.10/site-packages/nksr/fields/kernel_field.py", line 291, in solve
    lhs_mat.add_block(
  File "/mnt/disk2/.conda/envs/nksr/lib/python3.10/site-packages/nksr/solver.py", line 34, in add_block
    self.inv_diag[pos_i] = 1.0 / a_x[a_i == a_j]
RuntimeError: CUDA error: an illegal memory access was encountered
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.

My gpu is 3060 12GB.

xiemeilong commented 1 year ago

It turns out that the numbers in my coordinate system were too large.