繁体 English 中英

自动点唱机 AI 和 GPU 问题 [关闭]

[英]jukebox ai and GPU problems [closed]

原文 2021-07-30 07:44:14 0 0 linux/ ssh/ cuda/ gpu

我正在尝试在我的 uni 的 linux 服务器 (ssh) 上使用 open.ai 项目中的 jukebox-ai。 我按照网站上的所有说明进行操作，从使用 conda 创建环境到安装请求的所有软件包，但收到以下消息：

`Using cuda False
THCudaCheck FAIL file=/opt/conda/conda-bld/pytorch_1579040055865/work/aten/src/THC/THCGeneral.cpp line=50 error=38 : no CUDA-capable device is detected
Caught error during NCCL init (attempt 0 of 5): cuda runtime error (38) : no CUDA-capable device is detected at /opt/conda/conda-bld/pytorch_1579040055865/work/aten/src/THC/THCGeneral.cpp:50
Caught error during NCCL init (attempt 1 of 5): trying to initialize the default process group twice!
Caught error during NCCL init (attempt 2 of 5): trying to initialize the default process group twice!
Caught error during NCCL init (attempt 3 of 5): trying to initialize the default process group twice!
Caught error during NCCL init (attempt 4 of 5): trying to initialize the default process group twice!
Traceback (most recent call last):
  File "jukebox/sample.py", line 282, in <module>
    fire.Fire(run)
  File "/compLing/stud/hitheses/fr/matrad/jukebox/lib/python3.7/site-packages/fire/core.py", line 127, in Fire
    component_trace = _Fire(component, args, context, name)
  File "/compLing/stud/hitheses/fr/matrad/jukebox/lib/python3.7/site-packages/fire/core.py", line 366, in _Fire
    component, remaining_args)
  File "/compLing/stud/hitheses/fr/matrad/jukebox/lib/python3.7/site-packages/fire/core.py", line 542, in _CallCallable
    result = fn(*varargs, **kwargs)
  File "jukebox/sample.py", line 274, in run
    rank, local_rank, device = setup_dist_from_mpi(port=port)
  File "/export/data/stud/hitheses/fr/matrad/jukebox/utils/dist_utils.py", line 46, in setup_dist_from_mpi
    return _setup_dist_from_mpi(master_addr, backend, port, n_attempts, verbose)
  File "/export/data/stud/hitheses/fr/matrad/jukebox/utils/dist_utils.py", line 101, in _setup_dist_from_mpi
    raise RuntimeError("Failed to initialize NCCL")
RuntimeError: Failed to initialize NCCL`.

我也通过 tensorflow 检查过，但同样的故事：

  `2021-07-30 09:35:14.484425: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcudart.so.11.0
2021-07-30 09:35:15.764669: I tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2021-07-30 09:35:15.778151: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcuda.so.1
2021-07-30 09:35:15.783451: E tensorflow/stream_executor/cuda/cuda_driver.cc:328] failed call to cuInit: CUDA_ERROR_NO_DEVICE: no CUDA-capable device is detected
2021-07-30 09:35:15.783498: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:169] retrieving CUDA diagnostic information for host: heilk
2021-07-30 09:35:15.783511: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:176] hostname: hulk
2021-07-30 09:35:15.783628: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:200] libcuda reported version is: 460.32.3
2021-07-30 09:35:15.783660: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:204] kernel reported version is: 460.32.3
2021-07-30 09:35:15.783670: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:310] kernel version seems to match DSO: 460.32.3`

我想 Python 无法识别服务器上的 GPU（服务器应该有 4 个），我不知道如何在环境中激活它们。