{
  "id": 302582,
  "title": "GPU On but Not Engaged",
  "url": "/competitions/g-research-crypto-forecasting/discussion/302582",
  "author_name": "",
  "post_date": "2022-01-23T08:05:47.447841Z",
  "votes": 7,
  "comment_count": 4,
  "views": 0,
  "content": "<p>My GPU is on and Internet settings on Kaggle is off. However, my CPU is engaged to about 99% but my GPU is 0.00%. I constantly run out of memory. How can I engage my GPU so that I will have enough memory to participate in this competition. I will appreciate your help.</p>\n<p>Thanks.</p>",
  "messages": [
    {
      "id": "1661084",
      "postDate": "01/23/2022 08:05:47",
      "content": "<p>My GPU is on and Internet settings on Kaggle is off. However, my CPU is engaged to about 99% but my GPU is 0.00%. I constantly run out of memory. How can I engage my GPU so that I will have enough memory to participate in this competition. I will appreciate your help.</p>\n<p>Thanks.</p>",
      "rawMarkdown": "My GPU is on and Internet settings on Kaggle is off. However, my CPU is engaged to about 99% but my GPU is 0.00%. I constantly run out of memory. How can I engage my GPU so that I will have enough memory to participate in this competition. I will appreciate your help.\n\nThanks.",
      "votes": null
    },
    {
      "id": "1661125",
      "postDate": "01/23/2022 08:37:40",
      "content": "<p>I don't know why you are having this problem. 😂</p>\n<p>Generally speaking, we only need to turn on the GPU, and then your code also has GPU-related code, then it can be used normally.</p>\n<p>Hope you can explain your question in more detail.</p>",
      "rawMarkdown": "I don't know why you are having this problem. 😂\n\nGenerally speaking, we only need to turn on the GPU, and then your code also has GPU-related code, then it can be used normally.\n\nHope you can explain your question in more detail.",
      "votes": null
    },
    {
      "id": "1661564",
      "postDate": "01/23/2022 14:31:29",
      "content": "<p>I only know of the TPU-related code. How should I write the GPU-related code? Thanks</p>",
      "rawMarkdown": "I only know of the TPU-related code. How should I write the GPU-related code? Thanks",
      "votes": null
    },
    {
      "id": "1661588",
      "postDate": "01/23/2022 14:48:33",
      "content": "<p>First of all, I'm not sure if you understand what I mean, what I actually want to say is some usage code of the GPU. 😅<br>\nFor example: In pytorch, \"device = torch.device('cuda:0'); model.to(device)\"; <br>\nIn XGBoost, \"tree_method = 'gpu_hist' \".</p>\n<p>Secondly, maybe my point of view is very simple, so the most important thing is to see your problem.</p>\n<p>Finally, I hope you can explain your problem clearly, because normally, the GPU in Kaggle can be used normally. 👀</p>",
      "rawMarkdown": "First of all, I'm not sure if you understand what I mean, what I actually want to say is some usage code of the GPU. 😅\nFor example: In pytorch, \"device = torch.device('cuda:0'); model.to(device)\"; \nIn XGBoost, \"tree_method = 'gpu_hist' \".\n\nSecondly, maybe my point of view is very simple, so the most important thing is to see your problem.\n\nFinally, I hope you can explain your problem clearly, because normally, the GPU in Kaggle can be used normally. 👀",
      "votes": null
    },
    {
      "id": "1662604",
      "postDate": "01/24/2022 12:20:22",
      "content": "<p>Thanks. </p>\n<p>I am trying to use deep learning for prediction. I use tensorflow. I am on feature engineering at this stage. I am trying to take care of the missing values in the Target columns. Just after doing this, I get a message which says my notebook has restarted because I have used up my memory. When I check the GPU usage, it says 0.00% but the CPU says 99.9%</p>",
      "rawMarkdown": "Thanks. \n\nI am trying to use deep learning for prediction. I use tensorflow. I am on feature engineering at this stage. I am trying to take care of the missing values in the Target columns. Just after doing this, I get a message which says my notebook has restarted because I have used up my memory. When I check the GPU usage, it says 0.00% but the CPU says 99.9%",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1661125,
      "author_name": "",
      "author_url": "",
      "post_date": "01/23/2022 08:37:40",
      "content": "<p>I don't know why you are having this problem. 😂</p>\n<p>Generally speaking, we only need to turn on the GPU, and then your code also has GPU-related code, then it can be used normally.</p>\n<p>Hope you can explain your question in more detail.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1661564,
          "author_name": "zammie",
          "author_url": "",
          "post_date": "01/23/2022 14:31:29",
          "content": "<p>I only know of the TPU-related code. How should I write the GPU-related code? Thanks</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1661588,
          "author_name": "",
          "author_url": "",
          "post_date": "01/23/2022 14:48:33",
          "content": "<p>First of all, I'm not sure if you understand what I mean, what I actually want to say is some usage code of the GPU. 😅<br>\nFor example: In pytorch, \"device = torch.device('cuda:0'); model.to(device)\"; <br>\nIn XGBoost, \"tree_method = 'gpu_hist' \".</p>\n<p>Secondly, maybe my point of view is very simple, so the most important thing is to see your problem.</p>\n<p>Finally, I hope you can explain your problem clearly, because normally, the GPU in Kaggle can be used normally. 👀</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1662604,
          "author_name": "zammie",
          "author_url": "",
          "post_date": "01/24/2022 12:20:22",
          "content": "<p>Thanks. </p>\n<p>I am trying to use deep learning for prediction. I use tensorflow. I am on feature engineering at this stage. I am trying to take care of the missing values in the Target columns. Just after doing this, I get a message which says my notebook has restarted because I have used up my memory. When I check the GPU usage, it says 0.00% but the CPU says 99.9%</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1661084": "My GPU is on and Internet settings on Kaggle is off. However, my CPU is engaged to about 99% but my GPU is 0.00%. I constantly run out of memory. How can I engage my GPU so that I will have enough memory to participate in this competition. I will appreciate your help.\n\nThanks.",
    "1661125": "I don't know why you are having this problem. 😂\n\nGenerally speaking, we only need to turn on the GPU, and then your code also has GPU-related code, then it can be used normally.\n\nHope you can explain your question in more detail.",
    "1661564": "I only know of the TPU-related code. How should I write the GPU-related code? Thanks",
    "1661588": "First of all, I'm not sure if you understand what I mean, what I actually want to say is some usage code of the GPU. 😅\nFor example: In pytorch, \"device = torch.device('cuda:0'); model.to(device)\"; \nIn XGBoost, \"tree_method = 'gpu_hist' \".\n\nSecondly, maybe my point of view is very simple, so the most important thing is to see your problem.\n\nFinally, I hope you can explain your problem clearly, because normally, the GPU in Kaggle can be used normally. 👀",
    "1662604": "Thanks. \n\nI am trying to use deep learning for prediction. I use tensorflow. I am on feature engineering at this stage. I am trying to take care of the missing values in the Target columns. Just after doing this, I get a message which says my notebook has restarted because I have used up my memory. When I check the GPU usage, it says 0.00% but the CPU says 99.9%"
  },
  "source": "meta"
}