title
Installing the GPU version of TensorFlow for making use of your CUDA GPU

description
Welcome to part nine of the Deep Learning with Neural Networks and TensorFlow tutorials. If you are going to realistically continue with deep learning, you're going to need to start using a GPU. While there exists demo data that, like the MNIST sample we used, you can successfully work with, it is not going to prepare you for many of the hardships that large datasets come with, and you wont be able to try out many of the more interesting examples of what neural networks are capable of. Thus, in this tutorial, we're going to be covering the GPU version of TensorFlow. In order to use the GPU version of TensorFlow, you will need an NVIDIA GPU with a compute capability greater than 3.0. I had been using a couple GTX 980s, which had been relatively decent, but I was not able to create models to the size that I wanted so I have bought a GTX Titan X instead, which is much more enjoyable to work with, so pay close attention to VRAM on the card. 4 GB is what I had on my 980s, which gives you 8 GB total, but 4+4 in SLI is not the same as 8GB on one card, for example. You can use two cards at once, but this is not ideal. Buy the single best GPU that you can. https://pythonprogramming.net https://twitter.com/sentdex https://www.facebook.com/pythonprogramming.net/ https://plus.google.com/+sentdex

detail
{'title': 'Installing the GPU version of TensorFlow for making use of your CUDA GPU', 'heatmap': [{'end': 127.041, 'start': 94.542, 'weight': 0.712}, {'end': 468.695, 'start': 457.527, 'weight': 0.735}, {'end': 815.4, 'start': 792.29, 'weight': 0.99}], 'summary': 'Discusses the importance of using the gpu version of tensorflow for deep learning, recommends a cuda-enabled gpu with a compute capability greater than 3, explains the process of dual booting windows and linux for nvidia gpu systems, and covers the installation of cuda toolkit 7.5, cudnn 4.0, and gpu version of tensorflow with specific instructions for ubuntu 16.04.', 'chapters': [{'end': 171.758, 'segs': [{'end': 171.758, 'src': 'heatmap', 'start': 94.542, 'weight': 0, 'content': [{'end': 102.851, 'text': "Anyway, you'd want to pick up an NVIDIA GPU with a compute capability greater than 3, go to the text-based version if you need a list,", 'start': 94.542, 'duration': 8.309}, {'end': 108.277, 'text': "and then we'll come down here and you can use CUDA-enabled GeForce products.", 'start': 102.851, 'duration': 5.426}, {'end': 110.28, 'text': 'These are like your desktop GPUs.', 'start': 108.558, 'duration': 1.722}, {'end': 118.53, 'text': "And basically as long as you've got something better or something equal to or better than the GTX 650, then you're totally fine.", 'start': 111.121, 'duration': 7.409}, {'end': 127.041, 'text': "So to give you an idea, like a GeForce, let's say GTX 650, you could probably pick one of these up on eBay for like $70 or something.", 'start': 118.85, 'duration': 8.191}, {'end': 135.905, 'text': "I highly, highly, highly recommend that even if you don't currently have one, go get one.", 'start': 129.502, 'duration': 6.403}, {'end': 139.185, 'text': 'Because not necessarily a 650.', 'start': 136.545, 'duration': 2.64}, {'end': 146.187, 'text': 'Ideally, you want to get basically the best GPU, the best single GPU that you can.', 'start': 139.185, 'duration': 7.002}, {'end': 151.228, 'text': 'Whenever I first started working with this stuff, I was on a couple of GTX 980s.', 'start': 147.147, 'duration': 4.081}, {'end': 161.533, 'text': "I was running them in SLI, and I that's when I found that SLI is not as ideal as it's marketed to be, so.", 'start': 153.008, 'duration': 8.525}, {'end': 166.575, 'text': 'so anyways, I have since upgraded myself to a Titan X, which is Absolutely glorious.', 'start': 161.533, 'duration': 5.042}, {'end': 171.758, 'text': 'I would highly recommend a Titan X, even though Nvidia did not send me one for free.', 'start': 167.375, 'duration': 4.383}], 'summary': 'Get an nvidia gpu with compute capability >3, gtx 650 or better, for $70 on ebay. titan x is recommended.', 'duration': 77.216, 'max_score': 94.542, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/io6Ajf5XkaM/pics/io6Ajf5XkaM94542.jpg'}], 'start': 2.162, 'title': 'Using gpu version of tensorflow', 'summary': 'Discusses the importance of using the gpu version of tensorflow for deep learning, emphasizing the need for a cuda-enabled gpu with a compute capability greater than 3 and recommends investing in a high-performance gpu like the titan x.', 'chapters': [{'end': 171.758, 'start': 2.162, 'title': 'Using gpu version of tensorflow', 'summary': 'Discusses the importance of using the gpu version of tensorflow for deep learning, emphasizing the need for a cuda-enabled gpu with a compute capability greater than 3, and recommends investing in a high-performance gpu like the titan x.', 'duration': 169.596, 'highlights': ['The importance of using the GPU version of TensorFlow for deep learning is emphasized, with the requirement of a CUDA-enabled GPU with a compute capability greater than 3. CUDA-enabled GPU with a compute capability greater than 3', 'Recommendation to invest in a high-performance GPU like the Titan X for optimal performance in deep learning tasks. Recommendation for high-performance GPU like the Titan X', 'Suggestion to pick up an NVIDIA GPU with a compute capability greater than 3, with the example of a GeForce GTX 650 available for around $70. Example of a GeForce GTX 650 available for around $70']}], 'duration': 169.596, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/io6Ajf5XkaM/pics/io6Ajf5XkaM2162.jpg', 'highlights': ['Recommendation for high-performance GPU like the Titan X', 'CUDA-enabled GPU with a compute capability greater than 3', 'Example of a GeForce GTX 650 available for around $70']}, {'end': 295.982, 'segs': [{'end': 203.497, 'src': 'embed', 'start': 171.758, 'weight': 1, 'content': [{'end': 175.98, 'text': 'That was very rude of them anyways, Moving along.', 'start': 171.758, 'duration': 4.222}, {'end': 181.763, 'text': 'so make sure you get an Nvidia GPU compute capability greater than three From there.', 'start': 175.98, 'duration': 5.783}, {'end': 183.765, 'text': "what you're gonna want to do is if you are on Windows.", 'start': 181.763, 'duration': 2.002}, {'end': 187.487, 'text': "obviously, if you're already on Linux Or if you're already on Mac OS, you don't need this step.", 'start': 183.765, 'duration': 3.722}, {'end': 196.332, 'text': "but if you're on Windows, what you need to do is Dual boot into Linux, or I suppose you could build a new computer and just put Linux on it.", 'start': 187.487, 'duration': 8.845}, {'end': 203.497, 'text': "But if you're, if you want to continue using your main computer, Then you'll want to dual boot Linux and Windows.", 'start': 196.332, 'duration': 7.165}], 'summary': 'Recommend getting nvidia gpu compute capability greater than 3 for windows users; consider dual-booting into linux for main computer.', 'duration': 31.739, 'max_score': 171.758, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/io6Ajf5XkaM/pics/io6Ajf5XkaM171758.jpg'}, {'end': 252.317, 'src': 'embed', 'start': 228.864, 'weight': 0, 'content': [{'end': 237.446, 'text': "And then you decide, OK, how much size do I want? And however much you shrink by is how large of a partition you're giving to Ubuntu.", 'start': 228.864, 'duration': 8.582}, {'end': 239.887, 'text': 'So do that.', 'start': 238.247, 'duration': 1.64}, {'end': 244.168, 'text': 'And when you hit shrink, this is a plausibly very long process.', 'start': 240.147, 'duration': 4.021}, {'end': 245.91, 'text': 'It could take hours.', 'start': 244.828, 'duration': 1.082}, {'end': 252.317, 'text': "Because what's happening, especially if you have an older hard drive, you've got files that are stored all over the place on that hard drive.", 'start': 246.57, 'duration': 5.747}], 'summary': 'Shrinking the partition for ubuntu can take hours due to file storage on older hard drives.', 'duration': 23.453, 'max_score': 228.864, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/io6Ajf5XkaM/pics/io6Ajf5XkaM228864.jpg'}, {'end': 302.984, 'src': 'embed', 'start': 274.334, 'weight': 2, 'content': [{'end': 276.355, 'text': 'So back up whatever you care about.', 'start': 274.334, 'duration': 2.021}, {'end': 280.296, 'text': "Also, don't stop this process just because it's taking a few hours.", 'start': 276.815, 'duration': 3.481}, {'end': 280.896, 'text': 'Get over it.', 'start': 280.396, 'duration': 0.5}, {'end': 283.117, 'text': 'Maybe do it overnight while you sleep or something.', 'start': 281.076, 'duration': 2.041}, {'end': 286.378, 'text': "If you have like a solid state drive, it's actually pretty quick.", 'start': 283.837, 'duration': 2.541}, {'end': 288.239, 'text': 'Like you did this within like a few minutes.', 'start': 286.418, 'duration': 1.821}, {'end': 295.982, 'text': "But I've heard horror stories of people using this on like regular hard drives and it's not not great.", 'start': 288.879, 'duration': 7.103}, {'end': 297.703, 'text': 'Anyway, moving along.', 'start': 296.742, 'duration': 0.961}, {'end': 302.984, 'text': "once you've done that, you'll want to download an iso of ubuntu.", 'start': 298.963, 'duration': 4.021}], 'summary': 'Back up important data, overnight for a few hours, quick on solid state drive, download ubuntu iso.', 'duration': 28.65, 'max_score': 274.334, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/io6Ajf5XkaM/pics/io6Ajf5XkaM274334.jpg'}], 'start': 171.758, 'title': 'Dual booting for nvidia gpu', 'summary': 'Explains the process of dual booting windows and linux to allocate space for ubuntu on an nvidia gpu system, emphasizing the need for a compute capability greater than three and the potential risks and time involved in the process.', 'chapters': [{'end': 295.982, 'start': 171.758, 'title': 'Dual booting for nvidia gpu', 'summary': 'Explains the process of dual booting windows and linux to allocate space for ubuntu on an nvidia gpu system, emphasizing the need for a compute capability greater than three and the potential risks and time involved in the process.', 'duration': 124.224, 'highlights': ['Allocate space for Ubuntu by shrinking the C drive in Windows, which can be a time-consuming process depending on the type of hard drive, potentially taking hours and posing a risk of corruption. Shrinking the C drive to create unallocated space for Ubuntu, with the process potentially taking hours and carrying a risk of data corruption.', 'Emphasize the need for an Nvidia GPU with a compute capability greater than three for the dual booting process. Highlighting the necessity of an Nvidia GPU with a compute capability greater than three for the dual booting process.', 'Suggests backing up important data and allowing sufficient time for the shrinking process, particularly if using a regular hard drive. Encouraging the backup of important data and patience during the shrinking process, especially for users with regular hard drives.']}], 'duration': 124.224, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/io6Ajf5XkaM/pics/io6Ajf5XkaM171758.jpg', 'highlights': ['Allocate space for Ubuntu by shrinking the C drive in Windows, which can be time-consuming and pose a risk of corruption.', 'Emphasize the need for an Nvidia GPU with a compute capability greater than three for the dual booting process.', 'Suggest backing up important data and allowing sufficient time for the shrinking process, especially for users with regular hard drives.']}, {'end': 414.165, 'segs': [{'end': 349.512, 'src': 'embed', 'start': 318.911, 'weight': 0, 'content': [{'end': 322.815, 'text': "okay, we're going to be developing in it, so you'd probably want a Ubuntu desktop.", 'start': 318.911, 'duration': 3.904}, {'end': 324.757, 'text': "You don't necessarily have to have a Ubuntu desktop.", 'start': 322.835, 'duration': 1.922}, {'end': 325.758, 'text': 'You could go for the server.', 'start': 324.777, 'duration': 0.981}, {'end': 328.08, 'text': "But the desktop's what I'm choosing.", 'start': 326.739, 'duration': 1.341}, {'end': 334.547, 'text': 'And rather than downloading it from their little site, I would choose alternative downloads and torrents and download the torrent.', 'start': 328.741, 'duration': 5.806}, {'end': 335.648, 'text': "It'll be a lot faster.", 'start': 334.687, 'duration': 0.961}, {'end': 338.911, 'text': 'And use, like, uTorrent or something like that to download it.', 'start': 335.868, 'duration': 3.043}, {'end': 343.006, 'text': 'Once you have a boon to.', 'start': 341.525, 'duration': 1.481}, {'end': 349.512, 'text': 'rather, if you follow along with, like the virtual box, We were able to just like specify where the ISO was, and it was super simple.', 'start': 343.006, 'duration': 6.506}], 'summary': 'Recommend using ubuntu desktop for development, download it via torrents for faster speed, and easily set it up with virtual box.', 'duration': 30.601, 'max_score': 318.911, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/io6Ajf5XkaM/pics/io6Ajf5XkaM318911.jpg'}, {'end': 379.472, 'src': 'embed', 'start': 355.096, 'weight': 1, 'content': [{'end': 362.762, 'text': "I use this universal USB installer And obviously don't click this download button.", 'start': 355.096, 'duration': 7.666}, {'end': 369.189, 'text': 'you would come down here, probably, and choose this one, and Anyway you can use a different one if you want.', 'start': 362.762, 'duration': 6.427}, {'end': 370.589, 'text': "It doesn't really matter.", 'start': 369.209, 'duration': 1.38}, {'end': 374.931, 'text': "You just need something that's going to create a boot media out of your USB drive.", 'start': 370.629, 'duration': 4.302}, {'end': 379.472, 'text': "Once you've done that and put the ISO on there, what you're going to want to do is restart your computer.", 'start': 375.491, 'duration': 3.981}], 'summary': 'Use a universal usb installer to create boot media for iso on usb drive.', 'duration': 24.376, 'max_score': 355.096, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/io6Ajf5XkaM/pics/io6Ajf5XkaM355096.jpg'}, {'end': 420.509, 'src': 'embed', 'start': 394.775, 'weight': 2, 'content': [{'end': 399.197, 'text': 'Make sure you choose to install Ubuntu alongside of Windows.', 'start': 394.775, 'duration': 4.422}, {'end': 401.539, 'text': "Do not replace Windows unless that's what you want to do.", 'start': 399.257, 'duration': 2.282}, {'end': 411.224, 'text': "Once you have Ubuntu and you're all installed, this is kind of where if you're on macOS or otherwise on Ubuntu already, you'd be picking back up.", 'start': 402.779, 'duration': 8.445}, {'end': 414.165, 'text': 'And basically, we need to do these three major steps.', 'start': 411.684, 'duration': 2.481}, {'end': 420.509, 'text': 'We need to install the CUDA toolkit, cuDNN, and then finally the GPU version of TensorFlow.', 'start': 414.405, 'duration': 6.104}], 'summary': 'Install ubuntu alongside windows, then install cuda toolkit, cudnn, and gpu version of tensorflow.', 'duration': 25.734, 'max_score': 394.775, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/io6Ajf5XkaM/pics/io6Ajf5XkaM394775.jpg'}], 'start': 296.742, 'title': 'Installing ubuntu for development', 'summary': 'Explains the process of downloading and installing ubuntu for development, emphasizes alternative downloads and torrents for faster downloads, and the steps to create a bootable usb drive and install ubuntu alongside windows.', 'chapters': [{'end': 414.165, 'start': 296.742, 'title': 'Installing ubuntu for development', 'summary': 'Explains the process of downloading and installing ubuntu for development, emphasizing the use of alternative downloads and torrents for faster downloads, and the steps to create a bootable usb drive and install ubuntu alongside windows.', 'duration': 117.423, 'highlights': ['The chapter emphasizes the use of alternative downloads and torrents for downloading Ubuntu, highlighting the faster download speed compared to the standard download from the website.', 'It explains the process of creating a bootable USB drive, highlighting the recommendation to use a universal USB installer and the importance of choosing the correct option to create a boot media.', 'The chapter outlines the steps to install Ubuntu alongside Windows, emphasizing the caution to not replace Windows unless intentional and the subsequent major steps to follow after installation.']}], 'duration': 117.423, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/io6Ajf5XkaM/pics/io6Ajf5XkaM296742.jpg', 'highlights': ['The chapter emphasizes the use of alternative downloads and torrents for downloading Ubuntu, highlighting the faster download speed compared to the standard download from the website.', 'It explains the process of creating a bootable USB drive, highlighting the recommendation to use a universal USB installer and the importance of choosing the correct option to create a boot media.', 'The chapter outlines the steps to install Ubuntu alongside Windows, emphasizing the caution to not replace Windows unless intentional and the subsequent major steps to follow after installation.']}, {'end': 951.805, 'segs': [{'end': 492.611, 'src': 'heatmap', 'start': 457.527, 'weight': 0, 'content': [{'end': 460.129, 'text': 'Go ahead and download whatever the latest versions are.', 'start': 457.527, 'duration': 2.602}, {'end': 461.65, 'text': "You probably won't be in trouble.", 'start': 460.189, 'duration': 1.461}, {'end': 468.695, 'text': 'But you can also go to tensorflow.org and just see what they say is supported by TensorFlow.', 'start': 461.95, 'duration': 6.745}, {'end': 472.758, 'text': "And right now, that's 7.5 for Cuda Toolkit and then 4.0 CUDNN.", 'start': 469.015, 'duration': 3.743}, {'end': 476.442, 'text': 'So, um, so yeah.', 'start': 474.799, 'duration': 1.643}, {'end': 484.354, 'text': 'Also, another lovely thing is the, for whatever reason, the Kuda Toolkit only supports up to Ubuntu 15.04.', 'start': 477.083, 'duration': 7.271}, {'end': 492.611, 'text': 'Um, You can still use it on 1604, and I would highly recommend you use Ubuntu 1604, not 1504.', 'start': 484.354, 'duration': 8.257}], 'summary': 'Latest supported versions: tensorflow 7.5 for cuda toolkit and 4.0 for cudnn. prefer ubuntu 16.04 over 15.04.', 'duration': 78.206, 'max_score': 457.527, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/io6Ajf5XkaM/pics/io6Ajf5XkaM457527.jpg'}, {'end': 661.778, 'src': 'embed', 'start': 635.192, 'weight': 2, 'content': [{'end': 642.054, 'text': 'so this is your graphics driver, basically, and then this is the CUDA toolkit for 7.5.', 'start': 635.192, 'duration': 6.862}, {'end': 644.597, 'text': 'So make sure you allow those both to be executable.', 'start': 642.054, 'duration': 2.543}, {'end': 649.042, 'text': "Once you've done both of those things, go ahead and install the graphics driver.", 'start': 645.158, 'duration': 3.884}, {'end': 655.99, 'text': "It'll probably error that you're not on a 32-bit system or something, and that's totally fine, unless you're actually on a 32-bit system.", 'start': 649.102, 'duration': 6.888}, {'end': 661.778, 'text': 'Anyway, install the driver, and then install Coodletoolkit 7.5.', 'start': 656.771, 'duration': 5.007}], 'summary': 'Install graphics driver and cuda toolkit 7.5 for compatibility with system.', 'duration': 26.586, 'max_score': 635.192, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/io6Ajf5XkaM/pics/io6Ajf5XkaM635192.jpg'}, {'end': 710.292, 'src': 'embed', 'start': 681.733, 'weight': 3, 'content': [{'end': 683.435, 'text': 'So just use dash dash override.', 'start': 681.733, 'duration': 1.702}, {'end': 688.318, 'text': "And that's literally all you have to do to install this on 16.04.", 'start': 683.835, 'duration': 4.483}, {'end': 694.342, 'text': "Other than I suppose that after you run this command you'll go through a license agreement,", 'start': 688.318, 'duration': 6.024}, {'end': 700.245, 'text': "you'll press and hold the spacebar for like an hour and then it has some default paths and stuff.", 'start': 694.342, 'duration': 5.903}, {'end': 705.329, 'text': "but I think even before that at some point it's going to ask you do you want to install the graphics driver??", 'start': 700.245, 'duration': 5.084}, {'end': 708.551, 'text': "No, you don't want to install their graphics driver.", 'start': 706.309, 'duration': 2.242}, {'end': 710.292, 'text': "Other than that, everything's default.", 'start': 708.851, 'duration': 1.441}], 'summary': 'Installing on 16.04 using dash dash override, default paths, and rejecting graphics driver.', 'duration': 28.559, 'max_score': 681.733, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/io6Ajf5XkaM/pics/io6Ajf5XkaM681733.jpg'}, {'end': 820.023, 'src': 'heatmap', 'start': 792.29, 'weight': 0.99, 'content': [{'end': 794.611, 'text': 'And then do Python 3 in your shell.', 'start': 792.29, 'duration': 2.321}, {'end': 796.271, 'text': 'Try to import TensorFlow.', 'start': 795.031, 'duration': 1.24}, {'end': 798.292, 'text': 'And hopefully you see something like this.', 'start': 796.731, 'duration': 1.561}, {'end': 801.673, 'text': 'And you just see a bunch of Successfullys right here.', 'start': 799.252, 'duration': 2.421}, {'end': 804.973, 'text': "OK So if you've done that, congratulations.", 'start': 802.613, 'duration': 2.36}, {'end': 807.715, 'text': "You got the GPU version of TensorFlow working, and you're done.", 'start': 805.153, 'duration': 2.562}, {'end': 815.4, 'text': "If you're having some sort of error with libqdnn or really any of these, but usually it's libqdnn, honestly.", 'start': 808.795, 'duration': 6.605}, {'end': 820.023, 'text': "You probably didn't copy and paste that stuff or something like that.", 'start': 816.52, 'duration': 3.503}], 'summary': 'Successfully import tensorflow in python 3 shell for gpu version.', 'duration': 27.733, 'max_score': 792.29, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/io6Ajf5XkaM/pics/io6Ajf5XkaM792290.jpg'}], 'start': 414.405, 'title': 'Installing gpu tensorflow', 'summary': 'Covers the installation of cuda toolkit 7.5, cudnn 4.0, and gpu version of tensorflow, with specific instructions for ubuntu 16.04, emphasizing version compatibility, error resolution, and the flexibility to use the latest versions, including cuda toolkit 8 release candidate availability.', 'chapters': [{'end': 461.65, 'start': 414.405, 'title': 'Installing cuda toolkit, cudnn, and gpu tensorflow', 'summary': 'Discusses the installation of cuda toolkit 7.5, cudnn 4.0, and gpu version of tensorflow, with a mention of cuda toolkit 8 release candidate availability, emphasizing the flexibility to use the latest versions.', 'duration': 47.245, 'highlights': ['The chapter emphasizes the installation of CUDA Toolkit 7.5, cuDNN 4.0, and GPU version of TensorFlow.', 'It mentions the availability of CUDA Toolkit 8 release candidate and advises flexibility to use the latest versions.', 'The speaker suggests that many viewers might be using a later version than the mentioned 7.5 and 4.0.']}, {'end': 951.805, 'start': 461.95, 'title': 'Installing gpu version of tensorflow on ubuntu 16.04', 'summary': 'Provides a detailed guide for installing the gpu version of tensorflow on ubuntu 16.04, including specific version requirements for cuda toolkit and cudnn, and step-by-step instructions for driver installation, toolkit installation, and troubleshooting, with an emphasis on version compatibility and error resolution.', 'duration': 489.855, 'highlights': ['Ensure compatibility with specific versions of Cuda Toolkit and CUDNN The current supported versions for TensorFlow are 7.5 for Cuda Toolkit and 4.0 for CUDNN, emphasizing the importance of version compatibility.', 'Installation process simplified for Ubuntu 16.04 The guide recommends using Ubuntu 16.04 and highlights only one major difference in the installation process compared to Ubuntu 15.04, emphasizing the simplicity for Ubuntu 16.04 installation.', 'Detailed instructions for driver installation and toolkit installation Step-by-step instructions are provided for installing the NVIDIA graphics driver and Cuda Toolkit 7.5, including specific commands and parameters, ensuring a clear and precise installation process.']}], 'duration': 537.4, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/io6Ajf5XkaM/pics/io6Ajf5XkaM414405.jpg', 'highlights': ['The chapter emphasizes the installation of CUDA Toolkit 7.5, cuDNN 4.0, and GPU version of TensorFlow.', 'Ensure compatibility with specific versions of Cuda Toolkit and CUDNN The current supported versions for TensorFlow are 7.5 for Cuda Toolkit and 4.0 for CUDNN, emphasizing the importance of version compatibility.', 'Detailed instructions for driver installation and toolkit installation Step-by-step instructions are provided for installing the NVIDIA graphics driver and Cuda Toolkit 7.5, including specific commands and parameters, ensuring a clear and precise installation process.', 'The guide recommends using Ubuntu 16.04 and highlights only one major difference in the installation process compared to Ubuntu 15.04, emphasizing the simplicity for Ubuntu 16.04 installation.', 'It mentions the availability of CUDA Toolkit 8 release candidate and advises flexibility to use the latest versions.', 'The speaker suggests that many viewers might be using a later version than the mentioned 7.5 and 4.0.']}], 'highlights': ['Recommendation for high-performance GPU like the Titan X', 'CUDA-enabled GPU with a compute capability greater than 3', 'Example of a GeForce GTX 650 available for around $70', 'Emphasize the need for an Nvidia GPU with a compute capability greater than three for the dual booting process', 'Allocate space for Ubuntu by shrinking the C drive in Windows, which can be time-consuming and pose a risk of corruption', 'The chapter emphasizes the use of alternative downloads and torrents for downloading Ubuntu, highlighting the faster download speed compared to the standard download from the website', 'The chapter outlines the steps to install Ubuntu alongside Windows, emphasizing the caution to not replace Windows unless intentional and the subsequent major steps to follow after installation', 'The chapter emphasizes the installation of CUDA Toolkit 7.5, cuDNN 4.0, and GPU version of TensorFlow', 'Ensure compatibility with specific versions of Cuda Toolkit and CUDNN The current supported versions for TensorFlow are 7.5 for Cuda Toolkit and 4.0 for CUDNN, emphasizing the importance of version compatibility', 'Detailed instructions for driver installation and toolkit installation Step-by-step instructions are provided for installing the NVIDIA graphics driver and Cuda Toolkit 7.5, including specific commands and parameters, ensuring a clear and precise installation process', 'The guide recommends using Ubuntu 16.04 and highlights only one major difference in the installation process compared to Ubuntu 15.04, emphasizing the simplicity for Ubuntu 16.04 installation', 'It mentions the availability of CUDA Toolkit 8 release candidate and advises flexibility to use the latest versions', 'The speaker suggests that many viewers might be using a later version than the mentioned 7.5 and 4.0']}