![\"screenshot_kaggle\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2231509%2F77d5f9b3a0de672cb8687dffe1f07ec7%2FScreenshot_2019-12-28%20bengaliai%20-%20CNN%20Inference%20Kaggle.png?generation=1577556608012131&alt=media\")
My kernel is getting restarted again and again with my GPU is turned ON.\nAny solution ?
", "messages": [ { "id": "705192", "postDate": "12/28/2019 16:24:20", "content": "My kernel is getting restarted again and again with my GPU is turned ON.\nAny solution ?
", "rawMarkdown": "My kernel is getting restarted again and again with my GPU is turned ON.\nAny solution ?", "votes": null }, { "id": "705198", "postDate": "12/28/2019 16:29:08", "content": "P.S :- It's taking around 110+ seconds to read each parquet file with GPU=ON and 50+seconds with GPU=OFF....isn't this unusual ?
", "rawMarkdown": "P.S :- It's taking around 110+ seconds to read each parquet file with GPU=ON and 50+seconds with GPU=OFF....isn't this unusual ?", "votes": null }, { "id": "705293", "postDate": "12/28/2019 18:12:18", "content": "When turning on GPU in a Kaggle notebook, this error message explains that you obtain less CPU cores if you use a GPU, hence making the load of each parquet file longer !
\n\n
I agree that the conditions of this competition are rough for GPU use as a lot of time is used to do CPU-related computation. (e.g reading parquets)
", "rawMarkdown": "When turning on GPU in a Kaggle notebook, this error message explains that you obtain less CPU cores if you use a GPU, hence making the load of each parquet file longer ! \n\n\n\n I agree that the conditions of this competition are rough for GPU use as a lot of time is used to do CPU-related computation. (e.g reading parquets)", "votes": null }, { "id": "705300", "postDate": "12/28/2019 18:36:06", "content": "Thanks for the reply π \nLoad on RAM was exploding , thats why kernel was getting restarted.
", "rawMarkdown": "Thanks for the reply π \nLoad on RAM was exploding , thats why kernel was getting restarted.", "votes": null }, { "id": "705521", "postDate": "12/29/2019 01:48:26", "content": "Optimize your code :), I am using iPadPro while traveling now. There are frequent freezing kernel needed to restart, so I canβt proceed my work. Maybe I have to rest until I can turn on my laptop XD.
", "rawMarkdown": "Optimize your code :), I am using iPadPro while traveling now. There are frequent freezing kernel needed to restart, so I canβt proceed my work. Maybe I have to rest until I can turn on my laptop XD.", "votes": null }, { "id": "705714", "postDate": "12/29/2019 10:35:13", "content": "Yupp π , this competition requires good data management along with speed as it is a rerun competition
", "rawMarkdown": "Yupp π , this competition requires good data management along with speed as it is a rerun competition", "votes": null }, { "id": "706146", "postDate": "12/30/2019 01:13:12", "content": "The problem is, your RAM ran out. Then, the kernel would be killed and automatically restart again. I met this problem every time I load a bunch of data and my RAM was running high. Hope this helps!
", "rawMarkdown": "The problem is, your RAM ran out. Then, the kernel would be killed and automatically restart again. I met this problem every time I load a bunch of data and my RAM was running high. Hope this helps!", "votes": null }, { "id": "706248", "postDate": "12/30/2019 06:13:52", "content": "Yup , i figured that out already. Thanks !!\nAnd also you only get 13GBs of RAM with GPU ON while 16GBs with GPU OFF
", "rawMarkdown": "Yup , i figured that out already. Thanks !!\nAnd also you only get 13GBs of RAM with GPU ON while 16GBs with GPU OFF", "votes": null } ], "comments": [ { "id": 705198, "author_name": "namanj27", "author_url": "", "post_date": "12/28/2019 16:29:08", "content": "P.S :- It's taking around 110+ seconds to read each parquet file with GPU=ON and 50+seconds with GPU=OFF....isn't this unusual ?
", "votes": null, "replies": [ { "id": 705293, "author_name": "dimartinot", "author_url": "", "post_date": "12/28/2019 18:12:18", "content": "When turning on GPU in a Kaggle notebook, this error message explains that you obtain less CPU cores if you use a GPU, hence making the load of each parquet file longer !
\n\n
I agree that the conditions of this competition are rough for GPU use as a lot of time is used to do CPU-related computation. (e.g reading parquets)
", "votes": null, "replies": [] }, { "id": 705300, "author_name": "namanj27", "author_url": "", "post_date": "12/28/2019 18:36:06", "content": "Thanks for the reply π \nLoad on RAM was exploding , thats why kernel was getting restarted.
", "votes": null, "replies": [] } ] }, { "id": 705521, "author_name": "hanjoonchoe", "author_url": "", "post_date": "12/29/2019 01:48:26", "content": "Optimize your code :), I am using iPadPro while traveling now. There are frequent freezing kernel needed to restart, so I canβt proceed my work. Maybe I have to rest until I can turn on my laptop XD.
", "votes": null, "replies": [ { "id": 705714, "author_name": "namanj27", "author_url": "", "post_date": "12/29/2019 10:35:13", "content": "Yupp π , this competition requires good data management along with speed as it is a rerun competition
", "votes": null, "replies": [] } ] }, { "id": 706146, "author_name": "unkownhihi", "author_url": "", "post_date": "12/30/2019 01:13:12", "content": "The problem is, your RAM ran out. Then, the kernel would be killed and automatically restart again. I met this problem every time I load a bunch of data and my RAM was running high. Hope this helps!
", "votes": null, "replies": [ { "id": 706248, "author_name": "namanj27", "author_url": "", "post_date": "12/30/2019 06:13:52", "content": "Yup , i figured that out already. Thanks !!\nAnd also you only get 13GBs of RAM with GPU ON while 16GBs with GPU OFF
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "705192": "My kernel is getting restarted again and again with my GPU is turned ON.\nAny solution ?", "705198": "P.S :- It's taking around 110+ seconds to read each parquet file with GPU=ON and 50+seconds with GPU=OFF....isn't this unusual ?", "705293": "When turning on GPU in a Kaggle notebook, this error message explains that you obtain less CPU cores if you use a GPU, hence making the load of each parquet file longer ! \n\n\n\n I agree that the conditions of this competition are rough for GPU use as a lot of time is used to do CPU-related computation. (e.g reading parquets)", "705300": "Thanks for the reply π \nLoad on RAM was exploding , thats why kernel was getting restarted.", "705521": "Optimize your code :), I am using iPadPro while traveling now. There are frequent freezing kernel needed to restart, so I canβt proceed my work. Maybe I have to rest until I can turn on my laptop XD.", "705714": "Yupp π , this competition requires good data management along with speed as it is a rerun competition", "706146": "The problem is, your RAM ran out. Then, the kernel would be killed and automatically restart again. I met this problem every time I load a bunch of data and my RAM was running high. Hope this helps!", "706248": "Yup , i figured that out already. Thanks !!\nAnd also you only get 13GBs of RAM with GPU ON while 16GBs with GPU OFF" }, "source": "meta" }