{
  "id": 299116,
  "title": "Why can't I use GPU?",
  "url": "/competitions/tensorflow-great-barrier-reef/discussion/299116",
  "author_name": "",
  "post_date": "2022-01-06T08:13:13.291649900Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I got true when I run tf.test.is_gpu_available() but I can't use GPU when train model using TF object detection api. The message shows I use CPU instead of GPU.</p>\n<p>W0106 07:56:15.657460 139853556217664 cross_device_ops.py:1387] There are non-GPU devices in tf.distribute.Strategy, not using nccl allreduce.<br>\nI0106 07:56:15.659935 139853556217664 mirrored_strategy.py:369] Using MirroredStrategy with devices ('/job:localhost/replica:0/task:0/device:CPU:0',)</p>",
  "messages": [
    {
      "id": "1640145",
      "postDate": "01/06/2022 08:13:13",
      "content": "<p>I got true when I run tf.test.is_gpu_available() but I can't use GPU when train model using TF object detection api. The message shows I use CPU instead of GPU.</p>\n<p>W0106 07:56:15.657460 139853556217664 cross_device_ops.py:1387] There are non-GPU devices in tf.distribute.Strategy, not using nccl allreduce.<br>\nI0106 07:56:15.659935 139853556217664 mirrored_strategy.py:369] Using MirroredStrategy with devices ('/job:localhost/replica:0/task:0/device:CPU:0',)</p>",
      "rawMarkdown": "I got true when I run tf.test.is_gpu_available() but I can't use GPU when train model using TF object detection api. The message shows I use CPU instead of GPU.\n\nW0106 07:56:15.657460 139853556217664 cross_device_ops.py:1387] There are non-GPU devices in tf.distribute.Strategy, not using nccl allreduce.\nI0106 07:56:15.659935 139853556217664 mirrored_strategy.py:369] Using MirroredStrategy with devices ('/job:localhost/replica:0/task:0/device:CPU:0',)",
      "votes": null
    },
    {
      "id": "1640362",
      "postDate": "01/06/2022 12:25:08",
      "content": "<p>I can not paste image here </p>\n<p>You have to turn on the GPU at the accelerator panel. Remember the GPU quota is limited.<br>\nSolution:<br>\n<a href=\"https://ibb.co/XJQVWFN\" target=\"_blank\">https://ibb.co/XJQVWFN</a></p>",
      "rawMarkdown": "I can not paste image here \n\nYou have to turn on the GPU at the accelerator panel. Remember the GPU quota is limited.\nSolution:\nhttps://ibb.co/XJQVWFN",
      "votes": null
    },
    {
      "id": "1640808",
      "postDate": "01/06/2022 20:45:30",
      "content": "<p>I am sure I have turn on the GPU button otherwise the tf.test.is_gpu_available() won't response a true. Also I am sure the GPU only be used less 6 hours.</p>",
      "rawMarkdown": "I am sure I have turn on the GPU button otherwise the tf.test.is_gpu_available() won't response a true. Also I am sure the GPU only be used less 6 hours.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1640362,
      "author_name": "marcinstasko",
      "author_url": "",
      "post_date": "01/06/2022 12:25:08",
      "content": "<p>I can not paste image here </p>\n<p>You have to turn on the GPU at the accelerator panel. Remember the GPU quota is limited.<br>\nSolution:<br>\n<a href=\"https://ibb.co/XJQVWFN\" target=\"_blank\">https://ibb.co/XJQVWFN</a></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1640808,
      "author_name": "planet2020",
      "author_url": "",
      "post_date": "01/06/2022 20:45:30",
      "content": "<p>I am sure I have turn on the GPU button otherwise the tf.test.is_gpu_available() won't response a true. Also I am sure the GPU only be used less 6 hours.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1640145": "I got true when I run tf.test.is_gpu_available() but I can't use GPU when train model using TF object detection api. The message shows I use CPU instead of GPU.\n\nW0106 07:56:15.657460 139853556217664 cross_device_ops.py:1387] There are non-GPU devices in tf.distribute.Strategy, not using nccl allreduce.\nI0106 07:56:15.659935 139853556217664 mirrored_strategy.py:369] Using MirroredStrategy with devices ('/job:localhost/replica:0/task:0/device:CPU:0',)",
    "1640362": "I can not paste image here \n\nYou have to turn on the GPU at the accelerator panel. Remember the GPU quota is limited.\nSolution:\nhttps://ibb.co/XJQVWFN",
    "1640808": "I am sure I have turn on the GPU button otherwise the tf.test.is_gpu_available() won't response a true. Also I am sure the GPU only be used less 6 hours."
  },
  "source": "meta"
}