{
  "id": 315235,
  "title": "TPU not available",
  "url": "/competitions/tpu-getting-started/discussion/315235",
  "author_name": "Bunny Tang",
  "post_date": "2022-03-27T02:09:48.304000",
  "votes": 2,
  "comment_count": 0,
  "views": null,
  "content": "<p>Hi,</p>\n<p>When I start the notebook, I got<br>\nTPUs are popular right now. You are #8 in the queue. You can wait, try connecting again later, or use another accelerator.</p>\n<p>I run the first few cells, cell by cell, and got below errors message. If I continue to run the notebook with model training etc, i will receive a popup about \"your TPU is idle, tune TPU off to save quota\"</p>\n<p>The model learning is now extreme slow.  Any help!</p>\n<p><code>import tensorflow as tf</code><br>\n2022-03-27 01:59:33.125864: W tensorflow/stream_executor/platform/default/dso_loader.cc:60] Could not load dynamic library 'libcudart.so.11.0'; dlerror: libcudart.so.11.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /opt/conda/lib<br>\n2022-03-27 01:59:33.125995: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.</p>\n<p><code>print(\"Tensorflow version \" + tf.__version__)</code><br>\nTensorflow version 2.4.1</p>\n<p><code>print(\"Num TPUs Available: \", len(tf.config.experimental.list_physical_devices('TPU')))</code><br>\nNum TPUs Available:  0<br>\n2022-03-27 01:59:38.533960: I tensorflow/compiler/jit/xla_cpu_device.cc:41] Not creating XLA devices, tf_xla_enable_xla_devices not set<br>\n2022-03-27 01:59:38.536941: W tensorflow/stream_executor/platform/default/dso_loader.cc:60] Could not load dynamic library 'libcuda.so.1'; dlerror: libcuda.so.1: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /opt/conda/lib<br>\n2022-03-27 01:59:38.536985: W tensorflow/stream_executor/cuda/cuda_driver.cc:326] failed call to cuInit: UNKNOWN ERROR (303)<br>\n2022-03-27 01:59:38.537013: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (da95d6668f9e): /proc/driver/nvidia/version does not exist</p>\n<p><code>print(\"Num GPUs Available: \", len(tf.config.experimental.list_physical_devices('GPU')))</code><br>\nNum GPUs Available:  0</p>\n<p>Best regards<br>\ntanghung</p>",
  "messages": [
    {
      "id": 1736121,
      "postDate": "2022-03-27T02:09:48.303Z",
      "content": "<p>Hi,</p>\n<p>When I start the notebook, I got<br>\nTPUs are popular right now. You are #8 in the queue. You can wait, try connecting again later, or use another accelerator.</p>\n<p>I run the first few cells, cell by cell, and got below errors message. If I continue to run the notebook with model training etc, i will receive a popup about \"your TPU is idle, tune TPU off to save quota\"</p>\n<p>The model learning is now extreme slow.  Any help!</p>\n<p><code>import tensorflow as tf</code><br>\n2022-03-27 01:59:33.125864: W tensorflow/stream_executor/platform/default/dso_loader.cc:60] Could not load dynamic library 'libcudart.so.11.0'; dlerror: libcudart.so.11.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /opt/conda/lib<br>\n2022-03-27 01:59:33.125995: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.</p>\n<p><code>print(\"Tensorflow version \" + tf.__version__)</code><br>\nTensorflow version 2.4.1</p>\n<p><code>print(\"Num TPUs Available: \", len(tf.config.experimental.list_physical_devices('TPU')))</code><br>\nNum TPUs Available:  0<br>\n2022-03-27 01:59:38.533960: I tensorflow/compiler/jit/xla_cpu_device.cc:41] Not creating XLA devices, tf_xla_enable_xla_devices not set<br>\n2022-03-27 01:59:38.536941: W tensorflow/stream_executor/platform/default/dso_loader.cc:60] Could not load dynamic library 'libcuda.so.1'; dlerror: libcuda.so.1: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /opt/conda/lib<br>\n2022-03-27 01:59:38.536985: W tensorflow/stream_executor/cuda/cuda_driver.cc:326] failed call to cuInit: UNKNOWN ERROR (303)<br>\n2022-03-27 01:59:38.537013: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (da95d6668f9e): /proc/driver/nvidia/version does not exist</p>\n<p><code>print(\"Num GPUs Available: \", len(tf.config.experimental.list_physical_devices('GPU')))</code><br>\nNum GPUs Available:  0</p>\n<p>Best regards<br>\ntanghung</p>",
      "rawMarkdown": "Hi,\n\nWhen I start the notebook, I got\nTPUs are popular right now. You are #8 in the queue. You can wait, try connecting again later, or use another accelerator.\n\nI run the first few cells, cell by cell, and got below errors message. If I continue to run the notebook with model training etc, i will receive a popup about \"your TPU is idle, tune TPU off to save quota\"\n\nThe model learning is now extreme slow.  Any help!\n\n\n`import tensorflow as tf`\n2022-03-27 01:59:33.125864: W tensorflow/stream_executor/platform/default/dso_loader.cc:60] Could not load dynamic library 'libcudart.so.11.0'; dlerror: libcudart.so.11.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /opt/conda/lib\n2022-03-27 01:59:33.125995: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.\n\n`print(\"Tensorflow version \" + tf.__version__)`\nTensorflow version 2.4.1\n\n`print(\"Num TPUs Available: \", len(tf.config.experimental.list_physical_devices('TPU')))`\nNum TPUs Available:  0\n2022-03-27 01:59:38.533960: I tensorflow/compiler/jit/xla_cpu_device.cc:41] Not creating XLA devices, tf_xla_enable_xla_devices not set\n2022-03-27 01:59:38.536941: W tensorflow/stream_executor/platform/default/dso_loader.cc:60] Could not load dynamic library 'libcuda.so.1'; dlerror: libcuda.so.1: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /opt/conda/lib\n2022-03-27 01:59:38.536985: W tensorflow/stream_executor/cuda/cuda_driver.cc:326] failed call to cuInit: UNKNOWN ERROR (303)\n2022-03-27 01:59:38.537013: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (da95d6668f9e): /proc/driver/nvidia/version does not exist\n\n`print(\"Num GPUs Available: \", len(tf.config.experimental.list_physical_devices('GPU')))`\nNum GPUs Available:  0\n\nBest regards\ntanghung",
      "votes": 2
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1736121": "Hi,\n\nWhen I start the notebook, I got\nTPUs are popular right now. You are #8 in the queue. You can wait, try connecting again later, or use another accelerator.\n\nI run the first few cells, cell by cell, and got below errors message. If I continue to run the notebook with model training etc, i will receive a popup about \"your TPU is idle, tune TPU off to save quota\"\n\nThe model learning is now extreme slow.  Any help!\n\n\n`import tensorflow as tf`\n2022-03-27 01:59:33.125864: W tensorflow/stream_executor/platform/default/dso_loader.cc:60] Could not load dynamic library 'libcudart.so.11.0'; dlerror: libcudart.so.11.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /opt/conda/lib\n2022-03-27 01:59:33.125995: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.\n\n`print(\"Tensorflow version \" + tf.__version__)`\nTensorflow version 2.4.1\n\n`print(\"Num TPUs Available: \", len(tf.config.experimental.list_physical_devices('TPU')))`\nNum TPUs Available:  0\n2022-03-27 01:59:38.533960: I tensorflow/compiler/jit/xla_cpu_device.cc:41] Not creating XLA devices, tf_xla_enable_xla_devices not set\n2022-03-27 01:59:38.536941: W tensorflow/stream_executor/platform/default/dso_loader.cc:60] Could not load dynamic library 'libcuda.so.1'; dlerror: libcuda.so.1: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /opt/conda/lib\n2022-03-27 01:59:38.536985: W tensorflow/stream_executor/cuda/cuda_driver.cc:326] failed call to cuInit: UNKNOWN ERROR (303)\n2022-03-27 01:59:38.537013: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (da95d6668f9e): /proc/driver/nvidia/version does not exist\n\n`print(\"Num GPUs Available: \", len(tf.config.experimental.list_physical_devices('GPU')))`\nNum GPUs Available:  0\n\nBest regards\ntanghung"
  }
}