{
  "id": 603776,
  "title": "Notebook Timeout Error",
  "url": "/competitions/rsna-intracranial-aneurysm-detection/discussion/603776",
  "author_name": "Isaac Menard",
  "post_date": "2025-09-04T10:14:17.344000",
  "votes": 0,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hi,</p>\n<p>I'm encountering a <strong>timeout</strong> issue with my Kaggle notebook, and I'd appreciate some help figuring out why.</p>\n<p>My notebook is set up to run on a GPU P100. I've conducted a benchmark where I processed 2,640 image slices across 10 tests, and it took 2.20 minutes to complete.</p>\n<p>I'm trying to predict the total execution time for a larger dataset of 2,500 tests. Based on the training data (4,348 tests) which contains a total of 1,001,306 slices, I calculated an average of 1001306/4348≈230 slices per test. For the 2,500 tests I need to run, this gives me a total of approximately 2500∗230=575000 slices.</p>\n<p>Using my benchmark data, I predicted the total runtime as follows:</p>\n<p>Runtime=(575,000&nbsp;slices)∗(2.20&nbsp;minutes/2640&nbsp;slices)≈479.17&nbsp;minutes≈7.99&nbsp;hours</p>\n<p>My notebook, however, <strong>timed out after 12 hours</strong>.</p>\n<p>I'm wondering if there's an issue with my calculation or if something else is at play. Could the GPU P100 not be utilized for the entire duration of the testing phase?</p>\n<p>Any insights would be greatly appreciated. Thanks!</p>",
  "messages": [
    {
      "id": 3291100,
      "postDate": "2025-09-19T01:52:55.073Z",
      "content": "<p>Notebooks (not hidden submissions) timeout within 12 hours for GPU and CPU and 9 hours for TPU. You can split the cases processed over multiple notebook save and runs.</p>",
      "rawMarkdown": "Notebooks (not hidden submissions) timeout within 12 hours for GPU and CPU and 9 hours for TPU. You can split the cases processed over multiple notebook save and runs.",
      "votes": 1
    },
    {
      "id": 3284381,
      "postDate": "2025-09-08T07:46:45.643Z",
      "content": "<p>My guess is there's some overhead in RSNA server, that you need to account for.</p>",
      "rawMarkdown": "My guess is there's some overhead in RSNA server, that you need to account for.\n ",
      "replies": [
        {
          "id": 3284566,
          "postDate": "2025-09-08T11:36:50.930Z",
          "content": "<p>If you need to, my solution was to had a timeout function that send the prediction after 11.5h to avoid a timeout</p>",
          "rawMarkdown": "If you need to, my solution was to had a timeout function that send the prediction after 11.5h to avoid a timeout"
        }
      ]
    },
    {
      "id": 3281450,
      "postDate": "2025-09-04T10:14:17.343Z",
      "content": "<p>Hi,</p>\n<p>I'm encountering a <strong>timeout</strong> issue with my Kaggle notebook, and I'd appreciate some help figuring out why.</p>\n<p>My notebook is set up to run on a GPU P100. I've conducted a benchmark where I processed 2,640 image slices across 10 tests, and it took 2.20 minutes to complete.</p>\n<p>I'm trying to predict the total execution time for a larger dataset of 2,500 tests. Based on the training data (4,348 tests) which contains a total of 1,001,306 slices, I calculated an average of 1001306/4348≈230 slices per test. For the 2,500 tests I need to run, this gives me a total of approximately 2500∗230=575000 slices.</p>\n<p>Using my benchmark data, I predicted the total runtime as follows:</p>\n<p>Runtime=(575,000&nbsp;slices)∗(2.20&nbsp;minutes/2640&nbsp;slices)≈479.17&nbsp;minutes≈7.99&nbsp;hours</p>\n<p>My notebook, however, <strong>timed out after 12 hours</strong>.</p>\n<p>I'm wondering if there's an issue with my calculation or if something else is at play. Could the GPU P100 not be utilized for the entire duration of the testing phase?</p>\n<p>Any insights would be greatly appreciated. Thanks!</p>",
      "rawMarkdown": "Hi,\n\nI'm encountering a **timeout** issue with my Kaggle notebook, and I'd appreciate some help figuring out why.\n\nMy notebook is set up to run on a GPU P100. I've conducted a benchmark where I processed 2,640 image slices across 10 tests, and it took 2.20 minutes to complete.\n\nI'm trying to predict the total execution time for a larger dataset of 2,500 tests. Based on the training data (4,348 tests) which contains a total of 1,001,306 slices, I calculated an average of 1001306/4348≈230 slices per test. For the 2,500 tests I need to run, this gives me a total of approximately 2500∗230=575000 slices.\n\nUsing my benchmark data, I predicted the total runtime as follows:\n\nRuntime=(575,000 slices)∗(2.20 minutes/2640 slices)≈479.17 minutes≈7.99 hours\n\nMy notebook, however, **timed out after 12 hours**.\n\nI'm wondering if there's an issue with my calculation or if something else is at play. Could the GPU P100 not be utilized for the entire duration of the testing phase?\n\nAny insights would be greatly appreciated. Thanks!"
    }
  ],
  "comments": [
    {
      "id": 3291100,
      "author_name": "coderRKJ",
      "author_url": "",
      "post_date": "2025-09-19T01:52:55.073000",
      "content": "<p>Notebooks (not hidden submissions) timeout within 12 hours for GPU and CPU and 9 hours for TPU. You can split the cases processed over multiple notebook save and runs.</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 3284381,
      "author_name": "bykim0125",
      "author_url": "",
      "post_date": "2025-09-08T07:46:45.643000",
      "content": "<p>My guess is there's some overhead in RSNA server, that you need to account for.</p>",
      "votes": 0,
      "replies": [
        {
          "id": 3284566,
          "author_name": "Isaac Menard",
          "author_url": "",
          "post_date": "2025-09-08T11:36:50.930000",
          "content": "<p>If you need to, my solution was to had a timeout function that send the prediction after 11.5h to avoid a timeout</p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "3291100": "Notebooks (not hidden submissions) timeout within 12 hours for GPU and CPU and 9 hours for TPU. You can split the cases processed over multiple notebook save and runs.",
    "3284381": "My guess is there's some overhead in RSNA server, that you need to account for.\n ",
    "3281450": "Hi,\n\nI'm encountering a **timeout** issue with my Kaggle notebook, and I'd appreciate some help figuring out why.\n\nMy notebook is set up to run on a GPU P100. I've conducted a benchmark where I processed 2,640 image slices across 10 tests, and it took 2.20 minutes to complete.\n\nI'm trying to predict the total execution time for a larger dataset of 2,500 tests. Based on the training data (4,348 tests) which contains a total of 1,001,306 slices, I calculated an average of 1001306/4348≈230 slices per test. For the 2,500 tests I need to run, this gives me a total of approximately 2500∗230=575000 slices.\n\nUsing my benchmark data, I predicted the total runtime as follows:\n\nRuntime=(575,000 slices)∗(2.20 minutes/2640 slices)≈479.17 minutes≈7.99 hours\n\nMy notebook, however, **timed out after 12 hours**.\n\nI'm wondering if there's an issue with my calculation or if something else is at play. Could the GPU P100 not be utilized for the entire duration of the testing phase?\n\nAny insights would be greatly appreciated. Thanks!"
  }
}