{
  "id": 4634,
  "title": "Running pylearn2 on EC2 ??",
  "url": "/competitions/challenges-in-representation-learning-the-black-box-learning-challenge/discussion/4634",
  "author_name": "",
  "post_date": "2013-05-20T13:09:53.530Z",
  "votes": null,
  "comment_count": 2,
  "views": 2435,
  "content": "<p>Anyone have any luck getting pyLearn2/Theano/etc up and running on an EC2 ubuntu gpu instance?</p>",
  "messages": [
    {
      "id": "24518",
      "postDate": "05/20/2013 13:09:53",
      "content": "<p>Anyone have any luck getting pyLearn2/Theano/etc up and running on an EC2 ubuntu gpu instance?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "24702",
      "postDate": "05/23/2013 04:29:28",
      "content": "<p>I wasted a few hours today trying to install NVIDIA drivers on Ubuntu 13.04. There isn't much documentation on this. I had more success with 12.04, though. Here's a hacky script that I put together:</p>\r\n<pre>#!/bin/bash\r\nsudo apt-get update\r\nsudo apt-get -y install make python-numpy python-scipy python-dev python-pip python-nose g&#43;&#43; libopenblas-dev gitsudo pip install Theano\r\nsudo pip install --upgrade --no-deps git&#43;git://github.com/Theano/Theano.git\r\nsudo apt-get -y install freeglut3\r\nsudo ln -s /usr/lib/x86_64-linux-gnu/libglut.so.3 /usr/lib/libglut.so\r\nwget http://developer.download.nvidia.com/compute/cuda/5_0/rel-update-1/installers/cuda_5.0.35_linux_64_ubuntu11.10-1\r\n.run\r\nchmod &#43;x ./cuda_5.0.35_linux_64_ubuntu11.10-1.run\r\nsudo ./cuda_5.0.35_linux_64_ubuntu11.10-1.run\r\n\r\necho &quot;enable multiverse by editing /etc/apt/sources.list and then press enter&quot;\r\nread\r\nsudo apt-get update\r\nsudo apt-get install nvidia-cuda-toolkit g&#43;&#43;-4.4 gcc-4.4\r\n\r\n</pre>\r\n<p>This should get Theano and CUDA installed on Ubuntu 12.04.</p>\r\n<p></p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "24705",
      "postDate": "05/23/2013 05:11:30",
      "content": "<p>A few people have tried doing theano on amazon GPU instances (I remember from the theano-users mailing list). Quick googling shows&nbsp;http://blog.sbos.in/post/37828521741/setting-up-gpu-cluster-instance-on-amazon-ec2-for-a</p>\r\n<p></p>\r\n<p>It's unclear whether the added virtualization makes the speed-ups still good enough (and last I checked the GPU instances on amazon were quite expensive).</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 24702,
      "author_name": "anlthms",
      "author_url": "",
      "post_date": "05/23/2013 04:29:28",
      "content": "<p>I wasted a few hours today trying to install NVIDIA drivers on Ubuntu 13.04. There isn't much documentation on this. I had more success with 12.04, though. Here's a hacky script that I put together:</p>\r\n<pre>#!/bin/bash\r\nsudo apt-get update\r\nsudo apt-get -y install make python-numpy python-scipy python-dev python-pip python-nose g&#43;&#43; libopenblas-dev gitsudo pip install Theano\r\nsudo pip install --upgrade --no-deps git&#43;git://github.com/Theano/Theano.git\r\nsudo apt-get -y install freeglut3\r\nsudo ln -s /usr/lib/x86_64-linux-gnu/libglut.so.3 /usr/lib/libglut.so\r\nwget http://developer.download.nvidia.com/compute/cuda/5_0/rel-update-1/installers/cuda_5.0.35_linux_64_ubuntu11.10-1\r\n.run\r\nchmod &#43;x ./cuda_5.0.35_linux_64_ubuntu11.10-1.run\r\nsudo ./cuda_5.0.35_linux_64_ubuntu11.10-1.run\r\n\r\necho &quot;enable multiverse by editing /etc/apt/sources.list and then press enter&quot;\r\nread\r\nsudo apt-get update\r\nsudo apt-get install nvidia-cuda-toolkit g&#43;&#43;-4.4 gcc-4.4\r\n\r\n</pre>\r\n<p>This should get Theano and CUDA installed on Ubuntu 12.04.</p>\r\n<p></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 24705,
      "author_name": "dumitru0",
      "author_url": "",
      "post_date": "05/23/2013 05:11:30",
      "content": "<p>A few people have tried doing theano on amazon GPU instances (I remember from the theano-users mailing list). Quick googling shows&nbsp;http://blog.sbos.in/post/37828521741/setting-up-gpu-cluster-instance-on-amazon-ec2-for-a</p>\r\n<p></p>\r\n<p>It's unclear whether the added virtualization makes the speed-ups still good enough (and last I checked the GPU instances on amazon were quite expensive).</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "24518": "",
    "24702": "",
    "24705": ""
  },
  "source": "meta"
}