no nvdiffrast_plugin_gl.so error. Any suggestion?

[garrisonz]

Thanks for your great work!!

After run "pip install ." , successfully installed VHAP-0.0.1 nvdiffrast-0.3.1

But I cannot run track.py as no nvdiffrast_plugin_gl.so error. Any suggestion?

The log is following:

(VHAP) :~/w/3drecon/VHAP$ python vhap/track.py --data.root_folder "data/monocular/" --exp.output_folder output/monocular/zyp3_whiteBg_staticOffset --data.sequence zyp3
2024-10-06 22:06:56.796809: I tensorflow/core/util/port.cc:110] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.
2024-10-06 22:06:56.825984: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 AVX_VNNI FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
2024-10-06 22:06:57.213637: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
Ignoring unknown cluster teeth.
/home/zhangyupeng/miniconda3/envs/VHAP/lib/python3.10/site-packages/pytorch3d/ops/laplacian_matrices.py:51: UserWarning: torch.sparse.SparseTensor(indices, values, shape, *, device=) is deprecated.  Please use torch.sparse_coo_tensor(indices, values, shape, dtype=, device=). (Triggered internally at ../torch/csrc/utils/tensor_new.cpp:641.)
  A = torch.sparse.FloatTensor(idx, ones, (V, V))
/home/zhangyupeng/miniconda3/envs/VHAP/lib/python3.10/site-packages/torch/utils/cpp_extension.py:1965: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation.
If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
  warnings.warn(
Traceback (most recent call last):
  File "/home/zhangyupeng/w/3drecon/VHAP/vhap/track.py", line 20, in <module>
    tracker = GlobalTracker(cfg)
  File "/home/zhangyupeng/w/3drecon/VHAP/vhap/model/tracker.py", line 1325, in __init__
    super().__init__(cfg)
  File "/home/zhangyupeng/w/3drecon/VHAP/vhap/model/tracker.py", line 68, in __init__
    self.render = NVDiffRenderer(
  File "/home/zhangyupeng/w/3drecon/VHAP/vhap/util/render_nvdiffrast.py", line 72, in __init__
    self.glctx = dr.RasterizeGLContext() if use_opengl else dr.RasterizeCudaContext()
  File "/home/zhangyupeng/miniconda3/envs/VHAP/lib/python3.10/site-packages/nvdiffrast/torch/ops.py", line 221, in __init__
    self.cpp_wrapper = _get_plugin(gl=True).RasterizeGLStateWrapper(output_db, mode == 'automatic', cuda_device_idx)
  File "/home/zhangyupeng/miniconda3/envs/VHAP/lib/python3.10/site-packages/nvdiffrast/torch/ops.py", line 118, in _get_plugin
    torch.utils.cpp_extension.load(name=plugin_name, sources=source_paths, extra_cflags=opts, extra_cuda_cflags=opts+['-lineinfo'], extra_ldflags=ldflags, with_cuda=True, verbose=False)
  File "/home/zhangyupeng/miniconda3/envs/VHAP/lib/python3.10/site-packages/torch/utils/cpp_extension.py", line 1312, in load
    return _jit_compile(
  File "/home/zhangyupeng/miniconda3/envs/VHAP/lib/python3.10/site-packages/torch/utils/cpp_extension.py", line 1747, in _jit_compile
    return _import_module_from_library(name, build_directory, is_python_module)
  File "/home/zhangyupeng/miniconda3/envs/VHAP/lib/python3.10/site-packages/torch/utils/cpp_extension.py", line 2141, in _import_module_from_library
    module = importlib.util.module_from_spec(spec)
  File "<frozen importlib._bootstrap>", line 571, in module_from_spec
  File "<frozen importlib._bootstrap_external>", line 1176, in create_module
  File "<frozen importlib._bootstrap>", line 241, in _call_with_frames_removed
ImportError: /home/zhangyupeng/.cache/torch_extensions/py310_cu121/nvdiffrast_plugin_gl/nvdiffrast_plugin_gl.so: cannot open shared object file: No such file or directory

[ShenhanQian]

Hi, seems like nvdiffrast is not successfully installed/compiled. Please try reinstall it via:

pip uninstall nvdiffrast
rm -r ~/.cache/torch_extensions/*/nvdiffrast*
pip install nvdiffrast@git+https://github.com/ShenhanQian/nvdiffrast@backface-culling

[garrisonz]

@ShenhanQian

After run reinstall it:

(VHAP) :~/w/3drecon/VHAP$ pip install nvdiffrast@git+https://github.com/ShenhanQian/nvdiffrast@backface-culling
Looking in indexes: https://pypi.org/simple, https://pypi.ngc.nvidia.com
Collecting nvdiffrast@ git+https://github.com/ShenhanQian/nvdiffrast@backface-culling
  Cloning https://github.com/ShenhanQian/nvdiffrast (to revision backface-culling) to /tmp/pip-install-fjcqfrbd/nvdiffrast_742fb56d361d40d093cb41d5ca4d9fed
  Running command git clone --filter=blob:none --quiet https://github.com/ShenhanQian/nvdiffrast /tmp/pip-install-fjcqfrbd/nvdiffrast_742fb56d361d40d093cb41d5ca4d9fed
  Running command git checkout -b backface-culling --track origin/backface-culling
  Switched to a new branch 'backface-culling'
  Branch 'backface-culling' set up to track remote branch 'backface-culling' from 'origin'.
  Resolved https://github.com/ShenhanQian/nvdiffrast to commit 22718580f24a313c429ba2c304794c264351f108
  Preparing metadata (setup.py) ... done
Requirement already satisfied: numpy in /home/zhangyupeng/miniconda3/envs/VHAP/lib/python3.10/site-packages (from nvdiffrast@ git+https://github.com/ShenhanQian/nvdiffrast@backface-culling) (1.22.3)

The above error "nvdiffrast_plugin_gl.so: cannot open shared object file: No such file or directory" still there

[ShenhanQian]

Can you show the result of ls ~/.cache/torch_extensions/*/nvdiffrast*

[garrisonz]

(VHAP) :~/w/3drecon/VHAP$ ls ~/.cache/torch_extensions/*/nvdiffrast*
/home/zhangyupeng/.cache/torch_extensions/py310_cu121/nvdiffrast_plugin:
build.ninja

/home/zhangyupeng/.cache/torch_extensions/py310_cu121/nvdiffrast_plugin_gl:
build.ninja

For more information, there is only one file under nvdiffrast_plugin_gl folder :

(VHAP) :~/w/3drecon/VHAP$ ll /home/zhangyupeng/.cache/torch_extensions/py310_cu121/nvdiffrast_plugin_gl/
total 12
drwxrwxr-x 2 zhangyupeng zhangyupeng 4096 10月  6 22:34 ./
drwxrwxr-x 4 zhangyupeng zhangyupeng 4096 10月  6 22:31 ../
-rw-rw-r-- 1 zhangyupeng zhangyupeng 3610 10月  6 22:27 build.ninja
(VHAP) :~/w/3drecon/VHAP$

[ShenhanQian]

Looks like the compilation does run as expected. You may refer to nvdiffrast's repo

[Lxiangyue]

I met the cuda 304 error, and the [F glutil.cpp:338] eglInitialize() failed

I fixed it simply with replacing dr.RasterizeGLContext with dr.RasterizeCudaContext in ShenhanQian/nvdiffrec and VHAP. And re-install these two repo.

[ShenhanQian]

Xiangyue, thanks for sharing your solution!

Given an upgrade of nvdiffrast, VHAP now runs entirely based on CUDA. Please pull the latest commit and upgrade nvdiffrast as instructed here: https://github.com/ShenhanQian/VHAP/releases/tag/v0.0.2

[garrisonz]

Sovlved by

export LD_LIBRARY_PATH=/home/xxxxxx/miniconda3/envs/VHAP/lib/:$LD_LIBRARY_PATH