{
  "id": 52263,
  "title": "Can we run multiple kernels at the same time?",
  "url": "/competitions/talkingdata-adtracking-fraud-detection/discussion/52263",
  "author_name": "",
  "post_date": "2018-03-18T07:44:02.437078200Z",
  "votes": 8,
  "comment_count": 8,
  "views": 0,
  "content": "<p>Can multiple kernels be ran simultaneously? Is there a limit to how many? If yes, would they cut into each other's time/memory?  </p>\n\n<p>I tried doing that and my running kernel got an error.  But it's been getting sporadic errors while running before ('An error has occurred. Please try rerunning the script.') so, I'm not sure if it's the other kernel that caused it this time.</p>",
  "messages": [
    {
      "id": "297875",
      "postDate": "03/18/2018 07:44:02",
      "content": "<p>Can multiple kernels be ran simultaneously? Is there a limit to how many? If yes, would they cut into each other's time/memory?  </p>\n\n<p>I tried doing that and my running kernel got an error.  But it's been getting sporadic errors while running before ('An error has occurred. Please try rerunning the script.') so, I'm not sure if it's the other kernel that caused it this time.</p>",
      "rawMarkdown": "Can multiple kernels be ran simultaneously? Is there a limit to how many? If yes, would they cut into each other's time/memory?  \n\nI tried doing that and my running kernel got an error.  But it's been getting sporadic errors while running before ('An error has occurred. Please try rerunning the script.') so, I'm not sure if it's the other kernel that caused it this time.",
      "votes": null
    },
    {
      "id": "298010",
      "postDate": "03/18/2018 14:35:01",
      "content": "<p>We can certainly run multiple kernels. I don't know if there is a limit, but I've never encountered one. I don't know if they cut into each other's resources, but my impression is that they don't, except for the overall demand on the system (which is probably almost irrelevant for an individual).</p>\n\n<p>In my experience, the error you got is usually a problem with the user interface and not the running kernel, which is usually still running in the background, but you can't access it until it finishes and you can look at the results. For me this error typically happens if the kernel loses contact with the browser, for example if I close my laptop.</p>",
      "rawMarkdown": "We can certainly run multiple kernels. I don't know if there is a limit, but I've never encountered one. I don't know if they cut into each other's resources, but my impression is that they don't, except for the overall demand on the system (which is probably almost irrelevant for an individual).\n\nIn my experience, the error you got is usually a problem with the user interface and not the running kernel, which is usually still running in the background, but you can't access it until it finishes and you can look at the results. For me this error typically happens if the kernel loses contact with the browser, for example if I close my laptop.",
      "votes": null
    },
    {
      "id": "298044",
      "postDate": "03/18/2018 15:22:19",
      "content": "<p>Thanks.  That's good news.</p>\n\n<p>As for that error I have a kernel that keeps stopping at different points in time.  Sometimes after 20 min, sometimes after 1 and half hours.  It's not a code error (doesn't show anything wrong in output).  I often don't do anything to browser either.  It's a bit frustrating, as some models I can't test on my laptop at all and rely on kernel runs...  </p>\n\n<p>Admittedly I am pushing the kernel to the max...</p>",
      "rawMarkdown": "Thanks.  That's good news.\n\nAs for that error I have a kernel that keeps stopping at different points in time.  Sometimes after 20 min, sometimes after 1 and half hours.  It's not a code error (doesn't show anything wrong in output).  I often don't do anything to browser either.  It's a bit frustrating, as some models I can't test on my laptop at all and rely on kernel runs...  \n\nAdmittedly I am pushing the kernel to the max...",
      "votes": null
    },
    {
      "id": "298169",
      "postDate": "03/19/2018 01:00:14",
      "content": "<p>Yes you can. In mercari competition I run 20 kernel at the same time. In the end of the competition I  created around 1000 kernel so i assume there was no limit, and yes kernel performance are highly affected by its neighbour. </p>",
      "rawMarkdown": "Yes you can. In mercari competition I run 20 kernel at the same time. In the end of the competition I  created around 1000 kernel so i assume there was no limit, and yes kernel performance are highly affected by its neighbour.",
      "votes": null
    },
    {
      "id": "298176",
      "postDate": "03/19/2018 01:35:19",
      "content": "<p>Wow.   thanks!  How much is the performance affected?  Is it just slower, or is ram/disk space lessened?</p>",
      "rawMarkdown": "Wow.   thanks!  How much is the performance affected?  Is it just slower, or is ram/disk space lessened?",
      "votes": null
    },
    {
      "id": "298350",
      "postDate": "03/19/2018 12:29:32",
      "content": "<p>I experienced a similar error. But, as Andy said, it seems to be the problem with my user interface, not the kernel itself -- if you check the 'Version' tab you will see that the kernel is still running even after the error has occurred and if you just let the kernel finish its work it may still produce the output file(s). </p>",
      "rawMarkdown": "I experienced a similar error. But, as Andy said, it seems to be the problem with my user interface, not the kernel itself -- if you check the 'Version' tab you will see that the kernel is still running even after the error has occurred and if you just let the kernel finish its work it may still produce the output file(s).",
      "votes": null
    },
    {
      "id": "298578",
      "postDate": "03/19/2018 18:55:03",
      "content": "<p>Thanks, I'm going to try that</p>",
      "rawMarkdown": "Thanks, I'm going to try that",
      "votes": null
    },
    {
      "id": "374756",
      "postDate": "08/23/2018 18:36:17",
      "content": "<p>Hi Yulia, Muhammad, how is the performance effected? While one kernel is running, my other kernel on other dataset is getting restarted automatically ?</p>\n\n<p>Also can we use GPU on multiple Kernels and run?</p>",
      "rawMarkdown": "Hi Yulia, Muhammad, how is the performance effected? While one kernel is running, my other kernel on other dataset is getting restarted automatically ?\n\nAlso can we use GPU on multiple Kernels and run?",
      "votes": null
    },
    {
      "id": "2685568",
      "postDate": "03/07/2024 09:04:30",
      "content": "<p>I think we can run 2 Kaggle kernels (with P100) at the same time now. Could someone confirm?</p>",
      "rawMarkdown": "I think we can run 2 Kaggle kernels (with P100) at the same time now. Could someone confirm?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2685568,
      "author_name": "pratt3000",
      "author_url": "",
      "post_date": "03/07/2024 09:04:30",
      "content": "<p>I think we can run 2 Kaggle kernels (with P100) at the same time now. Could someone confirm?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 298010,
      "author_name": "aharless",
      "author_url": "",
      "post_date": "03/18/2018 14:35:01",
      "content": "<p>We can certainly run multiple kernels. I don't know if there is a limit, but I've never encountered one. I don't know if they cut into each other's resources, but my impression is that they don't, except for the overall demand on the system (which is probably almost irrelevant for an individual).</p>\n\n<p>In my experience, the error you got is usually a problem with the user interface and not the running kernel, which is usually still running in the background, but you can't access it until it finishes and you can look at the results. For me this error typically happens if the kernel loses contact with the browser, for example if I close my laptop.</p>",
      "votes": null,
      "replies": [
        {
          "id": 298044,
          "author_name": "yuliagm",
          "author_url": "",
          "post_date": "03/18/2018 15:22:19",
          "content": "<p>Thanks.  That's good news.</p>\n\n<p>As for that error I have a kernel that keeps stopping at different points in time.  Sometimes after 20 min, sometimes after 1 and half hours.  It's not a code error (doesn't show anything wrong in output).  I often don't do anything to browser either.  It's a bit frustrating, as some models I can't test on my laptop at all and rely on kernel runs...  </p>\n\n<p>Admittedly I am pushing the kernel to the max...</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 298350,
          "author_name": "graf10a",
          "author_url": "",
          "post_date": "03/19/2018 12:29:32",
          "content": "<p>I experienced a similar error. But, as Andy said, it seems to be the problem with my user interface, not the kernel itself -- if you check the 'Version' tab you will see that the kernel is still running even after the error has occurred and if you just let the kernel finish its work it may still produce the output file(s). </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 298578,
          "author_name": "yuliagm",
          "author_url": "",
          "post_date": "03/19/2018 18:55:03",
          "content": "<p>Thanks, I'm going to try that</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 298169,
      "author_name": "muhammadalfiansyah",
      "author_url": "",
      "post_date": "03/19/2018 01:00:14",
      "content": "<p>Yes you can. In mercari competition I run 20 kernel at the same time. In the end of the competition I  created around 1000 kernel so i assume there was no limit, and yes kernel performance are highly affected by its neighbour. </p>",
      "votes": null,
      "replies": [
        {
          "id": 298176,
          "author_name": "yuliagm",
          "author_url": "",
          "post_date": "03/19/2018 01:35:19",
          "content": "<p>Wow.   thanks!  How much is the performance affected?  Is it just slower, or is ram/disk space lessened?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 374756,
          "author_name": "ritesaluja",
          "author_url": "",
          "post_date": "08/23/2018 18:36:17",
          "content": "<p>Hi Yulia, Muhammad, how is the performance effected? While one kernel is running, my other kernel on other dataset is getting restarted automatically ?</p>\n\n<p>Also can we use GPU on multiple Kernels and run?</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "297875": "Can multiple kernels be ran simultaneously? Is there a limit to how many? If yes, would they cut into each other's time/memory?  \n\nI tried doing that and my running kernel got an error.  But it's been getting sporadic errors while running before ('An error has occurred. Please try rerunning the script.') so, I'm not sure if it's the other kernel that caused it this time.",
    "298010": "We can certainly run multiple kernels. I don't know if there is a limit, but I've never encountered one. I don't know if they cut into each other's resources, but my impression is that they don't, except for the overall demand on the system (which is probably almost irrelevant for an individual).\n\nIn my experience, the error you got is usually a problem with the user interface and not the running kernel, which is usually still running in the background, but you can't access it until it finishes and you can look at the results. For me this error typically happens if the kernel loses contact with the browser, for example if I close my laptop.",
    "298044": "Thanks.  That's good news.\n\nAs for that error I have a kernel that keeps stopping at different points in time.  Sometimes after 20 min, sometimes after 1 and half hours.  It's not a code error (doesn't show anything wrong in output).  I often don't do anything to browser either.  It's a bit frustrating, as some models I can't test on my laptop at all and rely on kernel runs...  \n\nAdmittedly I am pushing the kernel to the max...",
    "298169": "Yes you can. In mercari competition I run 20 kernel at the same time. In the end of the competition I  created around 1000 kernel so i assume there was no limit, and yes kernel performance are highly affected by its neighbour.",
    "298176": "Wow.   thanks!  How much is the performance affected?  Is it just slower, or is ram/disk space lessened?",
    "298350": "I experienced a similar error. But, as Andy said, it seems to be the problem with my user interface, not the kernel itself -- if you check the 'Version' tab you will see that the kernel is still running even after the error has occurred and if you just let the kernel finish its work it may still produce the output file(s).",
    "298578": "Thanks, I'm going to try that",
    "374756": "Hi Yulia, Muhammad, how is the performance effected? While one kernel is running, my other kernel on other dataset is getting restarted automatically ?\n\nAlso can we use GPU on multiple Kernels and run?",
    "2685568": "I think we can run 2 Kaggle kernels (with P100) at the same time now. Could someone confirm?"
  },
  "source": "meta"
}