{"id":961,"date":"2019-01-15T22:56:36","date_gmt":"2019-01-15T20:56:36","guid":{"rendered":"http:\/\/ideee.3deee.ch\/?p=961"},"modified":"2019-01-18T18:33:56","modified_gmt":"2019-01-18T16:33:56","slug":"tensorflow-gpu-installation-on-windows","status":"publish","type":"post","link":"https:\/\/ideee.3deee.ch\/?p=961","title":{"rendered":"Tensorflow GPU Installation on Windows"},"content":{"rendered":"\n<p><a href=\"https:\/\/www.tensorflow.org\">Tensorflow <\/a>is a library that is often used with Python to solve deep learning problems. This could be object recognition in images, style transfer of images as you might know it from apps like DeepArt and Prisma or language tools that can recognize the mood of a tweet. I use Tensorflow a lot with <a href=\"http:\/\/www.keras.io\">Keras <\/a>as API to try different things.<\/p>\n\n\n\n<p>If you just want to test something quickly, you don&#8217;t have to install it, you can also use the online tool <a href=\"https:\/\/colab.research.google.com\/\">Google CoLab<\/a>, there you also have the possibility to use a GPU and even for free. More about this in this <a href=\"http:\/\/ideee.3deee.ch\/?p=957\">blogpost &#8220;Use Google CoLab with GPU or TPU&#8221;.<\/a><\/p>\n\n\n\n<p>On the one hand you can run Tensorflow on the CPU, which works relatively well with the usual examples like MNIST. Especially, because these are relatively simple problems consisting of an overseeable data set of ~10000 data with small 8\u00d78 pixel images.<\/p>\n\n\n\n<p> <\/p>\n\n\n\n\r<div class=\"GARD gard_advanced_mode\" id=\"google-ads-Ws1l\"><script>\radUnit = document.getElementById(\"google-ads-Ws1l\");\rgoogle_ad_client = \"ca-pub-2912967267075159\";\radUnit = document.getElementById(\"google-ads-Ws1l\");\radWidth = adUnit.offsetWidth;\rif ( adWidth >= 999999 ) {\r\/* GETTING THE FIRST IF OUT OF THE WAY *\/ \r} else if ( adWidth >= 468 ) {\rif (document.querySelectorAll(\".ad_unit\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"9924945823\";\rgoogle_ad_width = 468;\rgoogle_ad_height = 60;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".ad_unit\").length;\rtag = \"ad_unit_468x60_\"+adcount;\radUnit.className = adUnit.className + \" ad_unit \" + tag;\r}\r} else if ( adWidth >= 300 ) {\rif (document.querySelectorAll(\".large_skyscraper\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"8448212620\";\rgoogle_ad_width = 300;\rgoogle_ad_height = 600;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".large_skyscraper\").length;\rtag = \"large_skyscraper_300x600_\"+adcount;\radUnit.className = adUnit.className + \" large_skyscraper \" + tag;\r}\r} else if ( adWidth >= 160 ) {\rif (document.querySelectorAll(\".ad_unit\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"7793371424\";\rgoogle_ad_width = 160;\rgoogle_ad_height = 600;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".ad_unit\").length;\rtag = \"ad_unit_160x600_\"+adcount;\radUnit.className = adUnit.className + \" ad_unit \" + tag;\r}\r} else {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r}<\/script>\r<script src=\"\/\/pagead2.googlesyndication.com\/pagead\/show_ads.js\"><\/script><\/div>\n\n\n\n<p> <\/p>\n\n\n\n<h2>Installation of Tensorflow for the CPU<\/h2>\n\n\n\n<p> For the CPU you can easily install Tensorflow with &#8220;pip install tensorflow&#8221; and try all tutorials if you have enough time.<\/p>\n\n\n\n<h2>Advantage of GPU support<\/h2>\n\n\n\n<p>The advantage of a GPU supported variant of Tensorflow is that the model is calculated much faster. A NVIDIA graphics card is required!<\/p>\n\n\n\n<p>Today I did a short test of a classification of a complex model. On the CPU I needed about 50 seconds despite Intel i7 with 8 cores.<br>\nThe same model needed only 16 seconds on the GPU!<\/p>\n\n\n\n<h2>Installation of Tensorflow for the GPU<\/h2>\n\n\n\n<p><br>Check that you install the correct version of Cuda and cuDNN for your Tensorflow version. Otherwise Tensorflow will not start.<\/p>\n\n\n\n<p>Tensorflow GPU Version 1.10.00 requires CudNN 7.4.2 for Cuda 9.0 and Cuda 9.0 &#8211;&gt;&nbsp;<a rel=\"noreferrer noopener\" target=\"_blank\" href=\"https:\/\/stackoverflow.com\/questions\/50622525\/which-tensorflow-and-cuda-version-combinations-are-compatible\">https:\/\/stackoverflow.com\/questions\/50622525\/which-tensorflow-and-cuda-version-combinations-are-compatible<\/a><\/p>\n\n\n\n<h2>1. Download &amp; Installation of CudNN<\/h2>\n\n\n\n<ul><li>You need a login to download it from &#8211;&gt;&nbsp;<a rel=\"noreferrer noopener\" target=\"_blank\" href=\"https:\/\/developer.nvidia.com\/rdp\/cudnn-download\">https:\/\/developer.nvidia.com\/rdp\/cudnn-download<\/a><\/li><li>Choose CudNN 7.4.2 for Cuda 9.0 and download it<\/li><li>-xtract it to f.e. &#8220;C:\\cudnn-9.0-v7.4.2.24&#8221;<\/li><li>add the following pathes to the System Environment Paths (&#8220;System &#8211;&gt; Advanced system settings &#8211;&gt; Environment Variables &#8211;&gt; System variables &#8211;&gt; Path<\/li><li>&#8211;&gt; &#8220;C:\\cudnn-9.0-v7.4.2.24&#8221;<\/li><li>&#8211;&gt; &#8220;C:\\cudnn-9.0-v7.4.2.24\\bin&#8221;<\/li><li><\/li><\/ul>\n\n\n\n<figure class=\"wp-block-image\"><img loading=\"lazy\" width=\"1024\" height=\"553\" src=\"http:\/\/ideee.3deee.ch\/wp-content\/uploads\/2019\/01\/CuDNN_Paths-1024x553.png\" alt=\"\" class=\"wp-image-962\" srcset=\"https:\/\/ideee.3deee.ch\/wp-content\/uploads\/2019\/01\/CuDNN_Paths-1024x553.png 1024w, https:\/\/ideee.3deee.ch\/wp-content\/uploads\/2019\/01\/CuDNN_Paths-300x162.png 300w, https:\/\/ideee.3deee.ch\/wp-content\/uploads\/2019\/01\/CuDNN_Paths-768x415.png 768w, https:\/\/ideee.3deee.ch\/wp-content\/uploads\/2019\/01\/CuDNN_Paths.png 1266w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p> <\/p>\n\n\n\n\r<div class=\"GARD gard_advanced_mode\" id=\"google-ads-Q25s\"><script>\radUnit = document.getElementById(\"google-ads-Q25s\");\rgoogle_ad_client = \"ca-pub-2912967267075159\";\radUnit = document.getElementById(\"google-ads-Q25s\");\radWidth = adUnit.offsetWidth;\rif ( adWidth >= 999999 ) {\r\/* GETTING THE FIRST IF OUT OF THE WAY *\/ \r} else if ( adWidth >= 468 ) {\rif (document.querySelectorAll(\".ad_unit\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"9924945823\";\rgoogle_ad_width = 468;\rgoogle_ad_height = 60;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".ad_unit\").length;\rtag = \"ad_unit_468x60_\"+adcount;\radUnit.className = adUnit.className + \" ad_unit \" + tag;\r}\r} else if ( adWidth >= 300 ) {\rif (document.querySelectorAll(\".large_skyscraper\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"8448212620\";\rgoogle_ad_width = 300;\rgoogle_ad_height = 600;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".large_skyscraper\").length;\rtag = \"large_skyscraper_300x600_\"+adcount;\radUnit.className = adUnit.className + \" large_skyscraper \" + tag;\r}\r} else if ( adWidth >= 160 ) {\rif (document.querySelectorAll(\".ad_unit\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"7793371424\";\rgoogle_ad_width = 160;\rgoogle_ad_height = 600;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".ad_unit\").length;\rtag = \"ad_unit_160x600_\"+adcount;\radUnit.className = adUnit.className + \" ad_unit \" + tag;\r}\r} else {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r}<\/script>\r<script src=\"\/\/pagead2.googlesyndication.com\/pagead\/show_ads.js\"><\/script><\/div>\n\n\n\n<p> <\/p>\n\n\n\n<h2>2. Download &amp; Installation of Cuda<\/h2>\n\n\n\n<ul><li>Download it from&nbsp;<a rel=\"noreferrer noopener\" target=\"_blank\" href=\"https:\/\/developer.nvidia.com\/cuda-90-download-archive\">https:\/\/developer.nvidia.com\/cuda-90-download-archive<\/a><\/li><li>Run the exe<\/li><li>optionally download all the patches and run them<\/li><\/ul>\n\n\n\n<p> <\/p>\n\n\n\n\r<div class=\"GARD gard_advanced_mode\" id=\"google-ads-xcFr\"><script>\radUnit = document.getElementById(\"google-ads-xcFr\");\rgoogle_ad_client = \"ca-pub-2912967267075159\";\radUnit = document.getElementById(\"google-ads-xcFr\");\radWidth = adUnit.offsetWidth;\rif ( adWidth >= 999999 ) {\r\/* GETTING THE FIRST IF OUT OF THE WAY *\/ \r} else if ( adWidth >= 468 ) {\rif (document.querySelectorAll(\".ad_unit\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"9924945823\";\rgoogle_ad_width = 468;\rgoogle_ad_height = 60;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".ad_unit\").length;\rtag = \"ad_unit_468x60_\"+adcount;\radUnit.className = adUnit.className + \" ad_unit \" + tag;\r}\r} else if ( adWidth >= 300 ) {\rif (document.querySelectorAll(\".large_skyscraper\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"8448212620\";\rgoogle_ad_width = 300;\rgoogle_ad_height = 600;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".large_skyscraper\").length;\rtag = \"large_skyscraper_300x600_\"+adcount;\radUnit.className = adUnit.className + \" large_skyscraper \" + tag;\r}\r} else if ( adWidth >= 160 ) {\rif (document.querySelectorAll(\".ad_unit\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"7793371424\";\rgoogle_ad_width = 160;\rgoogle_ad_height = 600;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".ad_unit\").length;\rtag = \"ad_unit_160x600_\"+adcount;\radUnit.className = adUnit.className + \" ad_unit \" + tag;\r}\r} else {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r}<\/script>\r<script src=\"\/\/pagead2.googlesyndication.com\/pagead\/show_ads.js\"><\/script><\/div>\n\n\n\n<p><\/p>\n\n\n\n<h2>3. Install Tensorflow<\/h2>\n\n\n\n<ul><li><a href=\"https:\/\/www.tensorflow.org\/install\/\ufeff\">tensorflow documentation<\/a><\/li><li>open a command line tool<\/li><li>optionally, activate your environment<\/li><li>install tensorflow for gpu over pip with the following command:<\/li><\/ul>\n\n\n\n<pre class=\"wp-block-code\"><code>pip install tensorflow-gpu<\/code><\/pre>\n\n\n\n<p><\/p>\n\n\n\n\r<div class=\"GARD gard_advanced_mode\" id=\"google-ads-6bFJ\"><script>\radUnit = document.getElementById(\"google-ads-6bFJ\");\rgoogle_ad_client = \"ca-pub-2912967267075159\";\radUnit = document.getElementById(\"google-ads-6bFJ\");\radWidth = adUnit.offsetWidth;\rif ( adWidth >= 999999 ) {\r\/* GETTING THE FIRST IF OUT OF THE WAY *\/ \r} else if ( adWidth >= 468 ) {\rif (document.querySelectorAll(\".ad_unit\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"9924945823\";\rgoogle_ad_width = 468;\rgoogle_ad_height = 60;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".ad_unit\").length;\rtag = \"ad_unit_468x60_\"+adcount;\radUnit.className = adUnit.className + \" ad_unit \" + tag;\r}\r} else if ( adWidth >= 300 ) {\rif (document.querySelectorAll(\".large_skyscraper\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"8448212620\";\rgoogle_ad_width = 300;\rgoogle_ad_height = 600;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".large_skyscraper\").length;\rtag = \"large_skyscraper_300x600_\"+adcount;\radUnit.className = adUnit.className + \" large_skyscraper \" + tag;\r}\r} else if ( adWidth >= 160 ) {\rif (document.querySelectorAll(\".ad_unit\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"7793371424\";\rgoogle_ad_width = 160;\rgoogle_ad_height = 600;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".ad_unit\").length;\rtag = \"ad_unit_160x600_\"+adcount;\radUnit.className = adUnit.className + \" ad_unit \" + tag;\r}\r} else {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r}<\/script>\r<script src=\"\/\/pagead2.googlesyndication.com\/pagead\/show_ads.js\"><\/script><\/div>\n\n\n\n<h2>4. Test your installation<\/h2>\n\n\n\n<p>I used the following scripts to test the installatioin in python.<\/p>\n\n\n\n<p>Start a python command line and paste the following code, which is from&nbsp;<a rel=\"noreferrer noopener\" target=\"_blank\" href=\"https:\/\/stackoverflow.com\/questions\/38559755\/how-to-get-current-available-gpus-in-tensorflow\">https:\/\/stackoverflow.com\/questions\/38559755\/how-to-get-current-available-gpus-in-tensorflow<\/a><\/p>\n\n\n\n<pre class=\"wp-block-preformatted\"><strong>from<\/strong> tensorflow.python.client <strong>import<\/strong> device_lib &nbsp; <br><strong>def<\/strong> get_available_gpus():     <br>    local_device_protos = device_lib.list_local_devices()     <br><strong>&nbsp;   return<\/strong> [x.name <strong>for<\/strong> x <strong>in<\/strong> local_device_protos <strong>if<\/strong> x.device_type == 'GPU'] get_available_gpus()<\/pre>\n\n\n\n<p>it should return something like this &#8230; it maybe varies depending on your GPU <\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>2019-01-15 12:07:32.972950: I tensorflow\/core\/platform\/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX AVX2\n2019-01-15 12:07:33.371039: I tensorflow\/core\/common_runtime\/gpu\/gpu_device.cc:1405] Found device 0 with properties:\nname: GeForce GTX 1080 major: 6 minor: 1 memoryClockRate(GHz): 1.835\npciBusID: 0000:6e:00.0\ntotalMemory: 8.00GiB freeMemory: 6.59GiB\n2019-01-15 12:07:33.378653: I tensorflow\/core\/common_runtime\/gpu\/gpu_device.cc:1484] Adding visible gpu devices: 0\n2019-01-15 12:07:33.848854: I tensorflow\/core\/common_runtime\/gpu\/gpu_device.cc:965] Device interconnect StreamExecutor with strength 1 edge matrix:\n2019-01-15 12:07:33.852797: I tensorflow\/core\/common_runtime\/gpu\/gpu_device.cc:971]      0\n2019-01-15 12:07:33.855312: I tensorflow\/core\/common_runtime\/gpu\/gpu_device.cc:984] 0:   N\n2019-01-15 12:07:33.857936: I tensorflow\/core\/common_runtime\/gpu\/gpu_device.cc:1097] Created TensorFlow device (\/device:GPU:0 with 6353 MB memory) -> physical GPU (device: 0, name: GeForce GTX 1080, pci bus id: 0000:6e:00.0, compute capability: 6.1)\n['\/device:GPU:0']<\/code><\/pre>\n\n\n\n<p> <\/p>\n\n\n\n\r<div class=\"GARD gard_advanced_mode\" id=\"google-ads-YeqQ\"><script>\radUnit = document.getElementById(\"google-ads-YeqQ\");\rgoogle_ad_client = \"ca-pub-2912967267075159\";\radUnit = document.getElementById(\"google-ads-YeqQ\");\radWidth = adUnit.offsetWidth;\rif ( adWidth >= 999999 ) {\r\/* GETTING THE FIRST IF OUT OF THE WAY *\/ \r} else if ( adWidth >= 468 ) {\rif (document.querySelectorAll(\".ad_unit\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"9924945823\";\rgoogle_ad_width = 468;\rgoogle_ad_height = 60;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".ad_unit\").length;\rtag = \"ad_unit_468x60_\"+adcount;\radUnit.className = adUnit.className + \" ad_unit \" + tag;\r}\r} else if ( adWidth >= 300 ) {\rif (document.querySelectorAll(\".large_skyscraper\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"8448212620\";\rgoogle_ad_width = 300;\rgoogle_ad_height = 600;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".large_skyscraper\").length;\rtag = \"large_skyscraper_300x600_\"+adcount;\radUnit.className = adUnit.className + \" large_skyscraper \" + tag;\r}\r} else if ( adWidth >= 160 ) {\rif (document.querySelectorAll(\".ad_unit\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"7793371424\";\rgoogle_ad_width = 160;\rgoogle_ad_height = 600;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".ad_unit\").length;\rtag = \"ad_unit_160x600_\"+adcount;\radUnit.className = adUnit.className + \" ad_unit \" + tag;\r}\r} else {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r}<\/script>\r<script src=\"\/\/pagead2.googlesyndication.com\/pagead\/show_ads.js\"><\/script><\/div>\n\n\n\n<p> <\/p>\n\n\n\n<p>\n\nAnd from&nbsp;<a rel=\"noreferrer noopener\" target=\"_blank\" href=\"https:\/\/www.tensorflow.org\/install\/pip\">https:\/\/www.tensorflow.org\/install\/pip<\/a>&nbsp;I used the following code snippet to test it\n\n<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>import tensorflow as tf\ntf.enable_eager_execution()\nprint(tf.reduce_sum(tf.random_normal([1000, 1000])))<\/code><\/pre>\n\n\n\n<p>\n\nalternatively you also could use this in a cmd window:\n\n<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>python -c \"import tensorflow as tf; tf.enable_eager_execution(); print(tf.reduce_sum(tf.random_normal([1000, 1000])))\"<\/code><\/pre>\n\n\n\n<p>\n\nBoth should output something like this:\n\n<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>2019-01-15 12:09:09.964358: I tensorflow\/core\/common_runtime\/gpu\/gpu_device.cc:1484] Adding visible gpu devices: 0\n2019-01-15 12:09:09.967450: I tensorflow\/core\/common_runtime\/gpu\/gpu_device.cc:965] Device interconnect StreamExecutor with strength 1 edge matrix:\n2019-01-15 12:09:09.972484: I tensorflow\/core\/common_runtime\/gpu\/gpu_device.cc:971]      0\n2019-01-15 12:09:09.975102: I tensorflow\/core\/common_runtime\/gpu\/gpu_device.cc:984] 0:   N\n2019-01-15 12:09:09.977746: I tensorflow\/core\/common_runtime\/gpu\/gpu_device.cc:1097] Created TensorFlow device (\/job:localhost\/replica:0\/task:0\/device:GPU:0 with 6353 MB memory) -> physical GPU (device: 0, name: GeForce GTX 1080, pci bus id: 0000:6e:00.0, compute capability: 6.1)\ntf.Tensor(-1654.6039, shape=(), dtype=float32)<\/code><\/pre>\n\n\n\n<p> <\/p>\n\n\n\n\r<div class=\"GARD gard_advanced_mode\" id=\"google-ads-6cFZ\"><script>\radUnit = document.getElementById(\"google-ads-6cFZ\");\rgoogle_ad_client = \"ca-pub-2912967267075159\";\radUnit = document.getElementById(\"google-ads-6cFZ\");\radWidth = adUnit.offsetWidth;\rif ( adWidth >= 999999 ) {\r\/* GETTING THE FIRST IF OUT OF THE WAY *\/ \r} else if ( adWidth >= 468 ) {\rif (document.querySelectorAll(\".ad_unit\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"9924945823\";\rgoogle_ad_width = 468;\rgoogle_ad_height = 60;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".ad_unit\").length;\rtag = \"ad_unit_468x60_\"+adcount;\radUnit.className = adUnit.className + \" ad_unit \" + tag;\r}\r} else if ( adWidth >= 300 ) {\rif (document.querySelectorAll(\".large_skyscraper\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"8448212620\";\rgoogle_ad_width = 300;\rgoogle_ad_height = 600;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".large_skyscraper\").length;\rtag = \"large_skyscraper_300x600_\"+adcount;\radUnit.className = adUnit.className + \" large_skyscraper \" + tag;\r}\r} else if ( adWidth >= 160 ) {\rif (document.querySelectorAll(\".ad_unit\").length > 3 ) {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r} else {\rgoogle_ad_slot = \"7793371424\";\rgoogle_ad_width = 160;\rgoogle_ad_height = 600;\radUnit.style.cssFloat  = \"\";\radUnit.style.styleFloat  = \"\";\radUnit.style.margin = \"\";\radUnit.style.textAlign = \"\";\radcount = document.querySelectorAll(\".ad_unit\").length;\rtag = \"ad_unit_160x600_\"+adcount;\radUnit.className = adUnit.className + \" ad_unit \" + tag;\r}\r} else {\rgoogle_ad_slot = \"0\";\radUnit.style.display = \"none\";\r}<\/script>\r<script src=\"\/\/pagead2.googlesyndication.com\/pagead\/show_ads.js\"><\/script><\/div>\n\n\n\n<p> <\/p>\n\n\n\n<figure class=\"wp-block-image\"><img loading=\"lazy\" width=\"979\" height=\"721\" src=\"http:\/\/ideee.3deee.ch\/wp-content\/uploads\/2019\/01\/TestTF_gpu_Installation.png\" alt=\"\" class=\"wp-image-963\" srcset=\"https:\/\/ideee.3deee.ch\/wp-content\/uploads\/2019\/01\/TestTF_gpu_Installation.png 979w, https:\/\/ideee.3deee.ch\/wp-content\/uploads\/2019\/01\/TestTF_gpu_Installation-300x221.png 300w, https:\/\/ideee.3deee.ch\/wp-content\/uploads\/2019\/01\/TestTF_gpu_Installation-768x566.png 768w\" sizes=\"(max-width: 979px) 100vw, 979px\" \/><\/figure>\n","protected":false},"excerpt":{"rendered":"<p>Tensorflow is a library that is often used with Python to solve deep learning problems. This could be object recognition in images, style transfer of images as you might know it from apps like DeepArt and Prisma or language tools that can recognize the mood of a tweet. I use&#8230;<\/p>\n","protected":false},"author":2,"featured_media":977,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_links_to":"","_links_to_target":""},"categories":[45,46],"tags":[135,130,131,132,134],"amp_enabled":true,"_links":{"self":[{"href":"https:\/\/ideee.3deee.ch\/index.php?rest_route=\/wp\/v2\/posts\/961"}],"collection":[{"href":"https:\/\/ideee.3deee.ch\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/ideee.3deee.ch\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/ideee.3deee.ch\/index.php?rest_route=\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/ideee.3deee.ch\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=961"}],"version-history":[{"count":2,"href":"https:\/\/ideee.3deee.ch\/index.php?rest_route=\/wp\/v2\/posts\/961\/revisions"}],"predecessor-version":[{"id":979,"href":"https:\/\/ideee.3deee.ch\/index.php?rest_route=\/wp\/v2\/posts\/961\/revisions\/979"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/ideee.3deee.ch\/index.php?rest_route=\/wp\/v2\/media\/977"}],"wp:attachment":[{"href":"https:\/\/ideee.3deee.ch\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=961"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/ideee.3deee.ch\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=961"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/ideee.3deee.ch\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=961"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}