search

paul007pl / MVP_Benchmark

MVP Benchmark for Multi-View Partial Point Cloud Completion and Registration
https://mvp-dataset.github.io/
Apache License 2.0
116 stars 10 forks source link

doesn't support CUDA >=11.0 #4

Open amiltonwong opened 2 years ago

amiltonwong commented 2 years ago

Hi, @paul007pl ,

According to the setup.sh, CUDA 10.1 is used. However, my GPU (RTX 30xx series) only support CUDA >=11.0. And the error output after inputing command python train.py -c ./cfgs/pcn.yaml is displayed as follows, it related to the CUDA version problem.

root@milton-LabPC:/data/code13/MVP_Benchmark/completion# python train.py -c ./cfgs/pcn.yaml
INFO:root:Munch({'batch_size': 32, 'workers': 0, 'nepoch': 100, 'model_name': 'pcn', 'load_model': None, 'start_epoch': 0, 'num_points': 2048, 'work_dir': 'log/', 'flag': 'debug', 'loss': 'cd', 'manual_seed': None, 'use_mean_feature': False, 'step_interval_to_print': 500, 'epoch_interval_to_save': 1, 'epoch_interval_to_val': 1, 'varying_constant': '0.01, 0.1, 0.5, 1', 'varying_constant_epochs': '5, 15, 30', 'lr': 0.0001, 'lr_decay': True, 'lr_decay_interval': 40, 'lr_decay_rate': 0.7, 'lr_step_decay_epochs': None, 'lr_step_decay_rates': None, 'lr_clip': 1e-06, 'optimizer': 'Adam', 'weight_decay': 0, 'betas': '0.9, 0.999', 'save_vis': True, 'eval_emd': False})
(62400, 2048, 3)
(2400, 2048, 3) (62400,)
(41600, 2048, 3)
(1600, 2048, 3) (41600,)
INFO:root:Length of train dataset:62400
INFO:root:Length of test dataset:41600
INFO:root:Random Seed: 6693
Jitting Chamfer 3D
Traceback (most recent call last):
  File "train.py", line 213, in <module>
    train()
  File "train.py", line 48, in train
    model_module = importlib.import_module('.%s' % args.model_name, 'models')
  File "/root/anaconda3/envs/pytorch1.5_4d_pls/lib/python3.7/importlib/__init__.py", line 127, in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
  File "<frozen importlib._bootstrap>", line 1006, in _gcd_import
  File "<frozen importlib._bootstrap>", line 983, in _find_and_load
  File "<frozen importlib._bootstrap>", line 967, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 677, in _load_unlocked
  File "<frozen importlib._bootstrap_external>", line 728, in exec_module
  File "<frozen importlib._bootstrap>", line 219, in _call_with_frames_removed
  File "/media/root/mdata/data/code13/MVP_Benchmark/completion/models/pcn.py", line 10, in <module>
    from model_utils import gen_grid_up, calc_emd, calc_cd
  File "/media/root/mdata/data/code13/MVP_Benchmark/completion/model_utils.py", line 20, in <module>
    from metrics import cd, fscore, emd
  File "../utils/metrics/__init__.py", line 1, in <module>
    from .CD import (cd, fscore)
  File "../utils/metrics/CD/__init__.py", line 1, in <module>
    from .chamfer3D.dist_chamfer_3D import chamfer_3DDist as cd
  File "../utils/metrics/CD/chamfer3D/dist_chamfer_3D.py", line 15, in <module>
    "/".join(os.path.abspath(__file__).split('/')[:-1] + ["chamfer3D.cu"]),
  File "/root/anaconda3/envs/pytorch1.5_4d_pls/lib/python3.7/site-packages/torch/utils/cpp_extension.py", line 898, in load
    is_python_module)
  File "/root/anaconda3/envs/pytorch1.5_4d_pls/lib/python3.7/site-packages/torch/utils/cpp_extension.py", line 1086, in _jit_compile
    with_cuda=with_cuda)
  File "/root/anaconda3/envs/pytorch1.5_4d_pls/lib/python3.7/site-packages/torch/utils/cpp_extension.py", line 1179, in _write_ninja_file_and_build_library
    with_cuda=with_cuda)
  File "/root/anaconda3/envs/pytorch1.5_4d_pls/lib/python3.7/site-packages/torch/utils/cpp_extension.py", line 1469, in _write_ninja_file_to_build_library
    cuda_flags = common_cflags + COMMON_NVCC_FLAGS + _get_cuda_arch_flags()
  File "/root/anaconda3/envs/pytorch1.5_4d_pls/lib/python3.7/site-packages/torch/utils/cpp_extension.py", line 1316, in _get_cuda_arch_flags
    raise ValueError("Unknown CUDA arch ({}) or GPU not supported".format(arch))
ValueError: Unknown CUDA arch (8.6) or GPU not supported

Any suggestions to fix this issue?

Thanks~

MaxChanger commented 2 years ago

I think the fastest way is to update the version of pytorch, when CUDA>=11.0, pytorch>=1.7.x is the most suitable version (according to the official website lower version should not be supported to install), whether the accuracy receives the impact needs to be tested.

amiltonwong commented 2 years ago

Hi, @MaxChanger

Then I use pytorch 1.7.0 version. However, I got the following error:

(pytorch1.7.0) root@milton-LabPC:/media/root/mdata/data/code13/MVP_Benchmark/completion# python train.py -c ./cfgs/pcn.yaml
INFO:root:Munch({'batch_size': 32, 'workers': 0, 'nepoch': 100, 'model_name': 'pcn', 'load_model': None, 'start_epoch': 0, 'num_points': 2048, 'work_dir': 'log/', 'flag': 'debug', 'loss': 'cd', 'manual_seed': None, 'use_mean_feature': False, 'step_interval_to_print': 500, 'epoch_interval_to_save': 1, 'epoch_interval_to_val': 1, 'varying_constant': '0.01, 0.1, 0.5, 1', 'varying_constant_epochs': '5, 15, 30', 'lr': 0.0001, 'lr_decay': True, 'lr_decay_interval': 40, 'lr_decay_rate': 0.7, 'lr_step_decay_epochs': None, 'lr_step_decay_rates': None, 'lr_clip': 1e-06, 'optimizer': 'Adam', 'weight_decay': 0, 'betas': '0.9, 0.999', 'save_vis': True, 'eval_emd': False})
(62400, 2048, 3)
(2400, 2048, 3) (62400,)
(41600, 2048, 3)
(1600, 2048, 3) (41600,)
INFO:root:Length of train dataset:62400
INFO:root:Length of test dataset:41600
INFO:root:Random Seed: 3648
Jitting Chamfer 3D
Traceback (most recent call last):
  File "/root/anaconda3/envs/pytorch1.7.0/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1516, in _run_ninja_build
    subprocess.run(
  File "/root/anaconda3/envs/pytorch1.7.0/lib/python3.8/subprocess.py", line 516, in run
    raise CalledProcessError(retcode, process.args,
subprocess.CalledProcessError: Command '['ninja', '-v']' returned non-zero exit status 1.

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "train.py", line 213, in <module>
    train()
  File "train.py", line 48, in train
    model_module = importlib.import_module('.%s' % args.model_name, 'models')
  File "/root/anaconda3/envs/pytorch1.7.0/lib/python3.8/importlib/__init__.py", line 127, in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
  File "<frozen importlib._bootstrap>", line 1014, in _gcd_import
  File "<frozen importlib._bootstrap>", line 991, in _find_and_load
  File "<frozen importlib._bootstrap>", line 975, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 671, in _load_unlocked
  File "<frozen importlib._bootstrap_external>", line 783, in exec_module
  File "<frozen importlib._bootstrap>", line 219, in _call_with_frames_removed
  File "/media/root/mdata/data/code13/MVP_Benchmark/completion/models/pcn.py", line 10, in <module>
    from model_utils import gen_grid_up, calc_emd, calc_cd
  File "/media/root/mdata/data/code13/MVP_Benchmark/completion/model_utils.py", line 20, in <module>
    from metrics import cd, fscore, emd
  File "../utils/metrics/__init__.py", line 1, in <module>
    from .CD import (cd, fscore)
  File "../utils/metrics/CD/__init__.py", line 1, in <module>
    from .chamfer3D.dist_chamfer_3D import chamfer_3DDist as cd
  File "../utils/metrics/CD/chamfer3D/dist_chamfer_3D.py", line 12, in <module>
    chamfer_3D = load(name="chamfer_3D",
  File "/root/anaconda3/envs/pytorch1.7.0/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 969, in load
    return _jit_compile(
  File "/root/anaconda3/envs/pytorch1.7.0/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1176, in _jit_compile
    _write_ninja_file_and_build_library(
  File "/root/anaconda3/envs/pytorch1.7.0/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1280, in _write_ninja_file_and_build_library
    _run_ninja_build(
  File "/root/anaconda3/envs/pytorch1.7.0/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1538, in _run_ninja_build
    raise RuntimeError(message) from e
RuntimeError: Error building extension 'chamfer_3D': [1/2] /usr/local/cuda-11.0/bin/nvcc -DTORCH_EXTENSION_NAME=chamfer_3D -DTORCH_API_INCLUDE_EXTENSION_H -isystem /root/anaconda3/envs/pytorch1.7.0/lib/python3.8/site-packages/torch/include -isystem /root/anaconda3/envs/pytorch1.7.0/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -isystem /root/anaconda3/envs/pytorch1.7.0/lib/python3.8/site-packages/torch/include/TH -isystem /root/anaconda3/envs/pytorch1.7.0/lib/python3.8/site-packages/torch/include/THC -isystem /usr/local/cuda-11.0/include -isystem /root/anaconda3/envs/pytorch1.7.0/include/python3.8 -D_GLIBCXX_USE_CXX11_ABI=0 -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr -gencode=arch=compute_86,code=sm_86 --compiler-options '-fPIC' -std=c++14 -c /media/root/mdata/data/code13/MVP_Benchmark/utils/metrics/CD/chamfer3D/chamfer3D.cu -o chamfer3D.cuda.o 
FAILED: chamfer3D.cuda.o 
/usr/local/cuda-11.0/bin/nvcc -DTORCH_EXTENSION_NAME=chamfer_3D -DTORCH_API_INCLUDE_EXTENSION_H -isystem /root/anaconda3/envs/pytorch1.7.0/lib/python3.8/site-packages/torch/include -isystem /root/anaconda3/envs/pytorch1.7.0/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -isystem /root/anaconda3/envs/pytorch1.7.0/lib/python3.8/site-packages/torch/include/TH -isystem /root/anaconda3/envs/pytorch1.7.0/lib/python3.8/site-packages/torch/include/THC -isystem /usr/local/cuda-11.0/include -isystem /root/anaconda3/envs/pytorch1.7.0/include/python3.8 -D_GLIBCXX_USE_CXX11_ABI=0 -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr -gencode=arch=compute_86,code=sm_86 --compiler-options '-fPIC' -std=c++14 -c /media/root/mdata/data/code13/MVP_Benchmark/utils/metrics/CD/chamfer3D/chamfer3D.cu -o chamfer3D.cuda.o 
nvcc fatal   : Unsupported gpu architecture 'compute_86'
ninja: build stopped: subcommand failed.

(pytorch1.7.0) root@milton-LabPC:/media/root/mdata/data/code13/MVP_Benchmark/completion

Any hints to fix this issue? Thanks~

MaxChanger commented 2 years ago

Hello, I'm not sure if you used 3080 or 3090. I search and found some feasible solutions, DeepSpeed/issues/607, pytorch/issues/45021, pytorch/issues/45028. In short, maybe the current CUDA computing power is higher than the CUDA computing power supported by the current PyTorch version. Hope these can help you.

EMCP commented 2 years ago

you need to be on the latest 1.9+ pytorch build.. anything older for me doesn't run on my 3090 RTX