{"id":2731,"date":"2020-06-30T22:35:52","date_gmt":"2020-06-30T13:35:52","guid":{"rendered":"https:\/\/www.beeete2.com\/blog\/?p=2731"},"modified":"2020-06-30T22:39:03","modified_gmt":"2020-06-30T13:39:03","slug":"windows%e3%81%abtensorflow2-1-gpu%e7%92%b0%e5%a2%83%e3%82%92%e6%a7%8b%e7%af%89%e3%81%99%e3%82%8b","status":"publish","type":"post","link":"https:\/\/www.beeete2.com\/blog\/?p=2731","title":{"rendered":"Windows\u306bTensorFlow2.1 + GPU\u74b0\u5883\u3092\u69cb\u7bc9\u3059\u308b"},"content":{"rendered":"<h2>cuDNN\u306e\u521d\u671f\u5316\u306b\u5931\u6557<\/h2>\n<p>Keras\u3092\u4f7f\u3063\u3066\u3044\u3066\u30d0\u30c3\u30af\u30a8\u30f3\u30c9\u306bTensorFlow\u3092\u4f7f\u3063\u3066\u3044\u308b\u3068\u304d\u306bConv2D\u30ec\u30a4\u30e4\u30fc\u304c\u542b\u307e\u308c\u308b\u30e2\u30c7\u30eb\u306e\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3092\u3057\u305f\u3068\u3053\u308d\u4ee5\u4e0b\u306e\u3088\u3046\u306a\u30a8\u30e9\u30fc\u304c\u767a\u751f\u3057\u305f\u306e\u3067\u5bfe\u5fdc\u3057\u307e\u3057\u305f\u3002<br \/>\n\u7d50\u679c\u7684\u306b\u306fCUDA Toolkit\u3068cuDNN\u3092\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3057\u3066\u5bfe\u5fdc\u3057\u307e\u3057\u305f\u3002<\/p>\n<p>This is probably because cuDNN failed to initialize windows<\/p>\n<h2>\u74b0\u5883\u60c5\u5831<\/h2>\n<p>\u767a\u751f\u6642\u306e\u74b0\u5883\u3067\u3059\u3002<\/p>\n<table style=\"border-collapse: collapse; width: 100%; height: 102px;\">\n<tbody>\n<tr style=\"height: 17px;\">\n<td style=\"width: 50%; height: 17px;\">OS<\/td>\n<td style=\"width: 50%; height: 17px;\">Windows 10 Pro<\/td>\n<\/tr>\n<tr style=\"height: 17px;\">\n<td style=\"width: 50%; height: 17px;\">Python<\/td>\n<td style=\"width: 50%; height: 17px;\">3.7.7<\/td>\n<\/tr>\n<tr style=\"height: 17px;\">\n<td style=\"width: 50%; height: 17px;\">TensorFlow<\/td>\n<td style=\"width: 50%; height: 17px;\">2.1.0<\/td>\n<\/tr>\n<tr style=\"height: 17px;\">\n<td style=\"width: 50%; height: 17px;\">Keras<\/td>\n<td style=\"width: 50%; height: 17px;\">2.2.4-tf (TensorFlow\u7d44\u307f\u8fbc\u307f)<\/td>\n<\/tr>\n<tr style=\"height: 17px;\">\n<td style=\"width: 50%; height: 17px;\">Anaconda<\/td>\n<td style=\"width: 50%; height: 17px;\">YES<\/td>\n<\/tr>\n<tr style=\"height: 17px;\">\n<td style=\"width: 50%; height: 17px;\">GPU<\/td>\n<td style=\"width: 50%; height: 17px;\">NVIDIA GeForce 2070<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>\u307e\u3055\u304b\u30b2\u30fc\u30e0\u4ee5\u5916\u306e\u3053\u3068\u306bGeForce 2070\u3092\u4f7f\u3046\u3053\u3068\u306b\u306a\u308d\u3046\u3068\u306f\u601d\u3044\u3082\u3057\u307e\u305b\u3093\u3067\u3057\u305f\u30fb\u30fb\u30fb\u3002<br \/>\nGeForce RTX 2070\u3092\u8cfc\u5165\u3057\u3066\u6d6e\u304b\u308c\u3066\u3044\u308b\u30dd\u30b9\u30c8\u306f<a href=\"https:\/\/www.beeete2.com\/blog\/?p=2082\">ROG-STRIX-RTX2070-O8G-GAMING\u8cb7\u3044\u307e\u3057\u305f<\/a>\u306b\u306a\u308a\u307e\u3059\u3002<br \/>\n\t\t\t<a href=\"https:\/\/www.beeete2.com\/blog\/?p=2082\" class=\"st-cardlink\">\n\t\t\t<div class=\"kanren st-cardbox\" >\n\t\t\t\t\t\t\t\t<dl class=\"clearfix\">\n\t\t\t\t\t<dt class=\"st-card-img\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"150\" height=\"150\" src=\"https:\/\/www.beeete2.com\/blog\/wp-content\/uploads\/2019\/03\/computer_jisaku_bunkai-150x150.png\" class=\"attachment-st_thumb150 size-st_thumb150 wp-post-image\" alt=\"\" srcset=\"https:\/\/www.beeete2.com\/blog\/wp-content\/uploads\/2019\/03\/computer_jisaku_bunkai-150x150.png 150w, https:\/\/www.beeete2.com\/blog\/wp-content\/uploads\/2019\/03\/computer_jisaku_bunkai-100x100.png 100w\" sizes=\"(max-width: 150px) 100vw, 150px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t<\/dt>\n\t\t\t\t\t<dd>\n\t\t\t\t\t\t\t\t\t\t\t\t\t<h5 class=\"st-cardbox-t\">ROG-STRIX-RTX2070-O8G-GAMING\u8cb7\u3044\u307e\u3057\u305f<\/h5>\n\t\t\t\t\t\t\n\t\t\t\t\t\t\t\t\t\t\t\t\t<div class=\"st-card-excerpt smanone\">\n\t\t\t\t\t\t\t\t<p>\u3064\u3044\u306bNVIDIA GeForce RTX 2070\u30b7\u30ea\u30fc\u30ba\u306eROG-STRIX-RTX2070-O8G-GAMING\u3092\u8cfc\u5165\u3057\u307e\u3057\u305f\u3002 \u81ea\u5206\u306f\u30e9\u30a4\u30c8\u306a\u30b2\u30fc\u30e0\u3057\u304b\u3084\u3063\u3066\u3044\u306a\u3044\u306e\u30672070\u30b7\u30ea\u30fc\u30ba\u306f\u81ea\u5206\u306b &#8230; <\/p>\n\t\t\t\t\t\t\t<\/div>\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<p class=\"cardbox-more\">\u7d9a\u304d\u3092\u898b\u308b<\/p>\n\t\t\t\t\t\t\t\t\t\t\t<\/dd>\n\t\t\t\t<\/dl>\n\t\t\t<\/div>\n\t\t\t<\/a>\n\t\t\t<\/p>\n<p>&nbsp;<\/p>\n<h2>CUDA\u306e\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u3068\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb<\/h2>\n<p>TensorFlow\u306e\u30d0\u30fc\u30b8\u30e7\u30f3\u306b\u3088\u3063\u3066\u4f7f\u3046CUDA Toolkit\u306e\u30d0\u30fc\u30b8\u30e7\u30f3\u3092\u6307\u5b9a\u3059\u308b\u5fc5\u8981\u304c\u3042\u308a\u307e\u3059\u3002<br \/>\n\u4eca\u56de\u306fTensorFlow <code>2.1<\/code>\u00a0\u3092\u4f7f\u3046\u306e\u3067CUDA Toolkit 10.1 update2\u3092\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u3057\u307e\u3059\u3002<br \/>\n<a href=\"https:\/\/developer.nvidia.com\/cuda-downloads\" target=\"_blank\" rel=\"noopener noreferrer\">CUDA Toolkit Download<\/a><\/p>\n<p>\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u5f8c\u306f\u9069\u5f53\u306b\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3057\u307e\u3059\u3002<br \/>\n\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u5f8c\u3001\u30bf\u30fc\u30df\u30ca\u30eb\u3067<code>nvcc -V<\/code>\u3092\u5b9f\u884c\u3057\u3066\u4ee5\u4e0b\u306e\u3088\u3046\u306a\u8868\u793a\u306b\u306a\u308c\u3070\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u5b8c\u4e86\u3067\u3059\u3002<br \/>\n<!--StartFragment --><\/p>\n<div>PS C:\\&gt; nvcc -V<br \/>\nnvcc: NVIDIA (R) Cuda compiler driver<br \/>\nCopyright (c) 2005-2019 NVIDIA Corporation<br \/>\nBuilt on Sun_Jul_28_19:12:52_Pacific_Daylight_Time_2019<br \/>\nCuda compilation tools, release 10.1, V10.1.243<\/div>\n<p><!--EndFragment --><\/p>\n<h2>cuDNN\u306e\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u3068\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb<\/h2>\n<p>\u5148\u7a0b\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3057\u305fCUDA Toolkit\u306b\u5bfe\u5fdc\u3057\u305fcuDNN\u3092\u30c0\u30a6\u30f3\u30ed\u30fc\u30c8\u3057\u3066\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3057\u307e\u3059\u3002<br \/>\n\u4eca\u56de\u306fCuda Toolkit <code>10.1<\/code>\u3092\u4f7f\u7528\u3059\u308b\u306e\u3067cuDNN v7.6.5 (November 5th, 2019), for CUDA 10.1\u3092\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u3057\u307e\u3059\u3002<\/p>\n<p>\u6b8b\u5ff5\u306a\u304c\u3089\u30a4\u30f3\u30b9\u30c8\u30fc\u30e9\u30fc\u306f\u306a\u3044\u3088\u3046\u306a\u306e\u3067\u3001\u6614\u306a\u304c\u3089\u306e\u65b9\u6cd5\u3067PATH\u3092\u901a\u3059\u304bCUDA Toolkit\u306e\u30d7\u30ed\u30b0\u30e9\u30e0\u30d5\u30a9\u30eb\u30c0\u306b\u30d5\u30a1\u30a4\u30eb\u3092\u30b3\u30d4\u30fc\u3059\u308b\u5fc5\u8981\u304c\u3042\u308b\u307f\u305f\u3044\u3067\u3059\u3002<br \/>\n\u4eca\u56de\u306fCUDA Toolkit\u306e\u30d7\u30ed\u30b0\u30e9\u30e0\u30d5\u30a9\u30eb\u30c0\u306b\u30b3\u30d4\u30fc\u3057\u307e\u3059\u3002<br \/>\n\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u3057\u305fzip\u30d5\u30a1\u30a4\u30eb\u3092\u89e3\u51cd\u3057\u307e\u3059\u3002<br \/>\n\u3059\u308b\u3068{bin, lib, include}\u3068\u3044\u3046\u30d5\u30a9\u30eb\u30c0\u304c\u3042\u308b\u306e\u3067CUDA Toolkit\u306e\u30d5\u30a9\u30eb\u30c0\u306b\u30b3\u30d4\u30fc\u3057\u307e\u3059\u3002<br \/>\n\u81ea\u5206\u306e\u74b0\u5883\u3067\u306f\u4ee5\u4e0b\u306e\u30d5\u30a9\u30eb\u30c0\u3067\u3057\u305f\u3002<br \/>\nC:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v10.1<br \/>\n\u4e0a\u66f8\u304d\u3055\u308c\u308b\u3082\u306e\u306f\u306a\u304b\u3063\u305f\u306e\u3067\u5b89\u5fc3\u3067\u3059\u3002<\/p>\n<h2>\u52d5\u4f5c\u78ba\u8a8d<\/h2>\n<p>\u30a8\u30e9\u30fc\u304c\u767a\u751f\u3057\u305f\u7573\u307f\u8fbc\u307f\u306e\u30ec\u30a4\u30e4\u30fc\uff08Conv2D\uff09\u304c\u542b\u307e\u308c\u308b\u30e2\u30c7\u30eb\u3067\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3057\u3066\u307f\u3066\u30a8\u30e9\u30fc\u304c\u767a\u751f\u3057\u306a\u304f\u306a\u308c\u3070OK\u3067\u3059\u3002<\/p>\n<div class=\"st-mybox  has-title \" style=\"background:#E1F5FE;border-color:#B3E5FC;border-width:2px;border-radius:5px;margin: 25px 0;\"><p class=\"st-mybox-title\" style=\"color:#03A9F4;font-weight:bold;text-shadow: #fff 3px 0px 0px, #fff 2.83487px 0.981584px 0px, #fff 2.35766px 1.85511px 0px, #fff 1.62091px 2.52441px 0px, #fff 0.705713px 2.91581px 0px, #fff -0.287171px 2.98622px 0px, #fff -1.24844px 2.72789px 0px, #fff -2.07227px 2.16926px 0px, #fff -2.66798px 1.37182px 0px, #fff -2.96998px 0.42336px 0px, #fff -2.94502px -0.571704px 0px, #fff -2.59586px -1.50383px 0px, #fff -1.96093px -2.27041px 0px, #fff -1.11013px -2.78704px 0px, #fff -0.137119px -2.99686px 0px, #fff 0.850987px -2.87677px 0px, #fff 1.74541px -2.43999px 0px, #fff 2.44769px -1.73459px 0px, #fff 2.88051px -0.838246px 0px;\"><i class=\"fa fa-check-circle st-css-no\" aria-hidden=\"true\"><\/i>\u30dd\u30a4\u30f3\u30c8<\/p><div class=\"st-in-mybox\">Jupyter Notebook\u3092\u4f7f\u3063\u3066\u3044\u308b\u5834\u5408\u306f\u518d\u8d77\u52d5\u3057\u3066\u304a\u304f\u3068\u3088\u3044\u3068\u601d\u3044\u307e\u3059\u3002\u305d\u308c\u3067\u3082\u3060\u3081\u306a\u3089Windows\u3092\u518d\u8d77\u52d5\u3002\u518d\u8d77\u52d5\u3001\u518d\u8d77\u52d5\u3001\u518d\u8d77\u52d5\uff01<\/div><\/div>\n<p>\u5f53\u7136\u30e2\u30c7\u30eb\u306e\u69cb\u7bc9\u306b\u3082\u3088\u308a\u307e\u3059\u304c\u3001\u81ea\u5206\u306e\u74b0\u5883\u3067\u306fCPU\u3060\u30681\u30a8\u30dd\u30c3\u30af\u7d0452\u79d2\u304b\u304b\u3063\u3066\u3044\u305f\u51e6\u7406\u304cGPU\u3060\u30685\u79d2\u306b\u307e\u3067\u77ed\u7e2e\u3055\u308c\u307e\u3057\u305f\u3002<br \/>\n\u51e6\u7406\u6642\u9593\u304c\u7d041\/10\u3067\u3059\u306d\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>cuDNN\u306e\u521d\u671f\u5316\u306b\u5931\u6557 Keras\u3092\u4f7f\u3063\u3066\u3044\u3066\u30d0\u30c3\u30af\u30a8\u30f3\u30c9\u306bTensorFlow\u3092\u4f7f\u3063\u3066\u3044\u308b\u3068\u304d\u306bConv2D\u30ec\u30a4\u30e4\u30fc\u304c\u542b\u307e\u308c\u308b\u30e2\u30c7\u30eb\u306e\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3092\u3057\u305f\u3068\u3053\u308d\u4ee5\u4e0b\u306e\u3088\u3046\u306a\u30a8\u30e9\u30fc\u304c\u767a\u751f\u3057\u305f\u306e\u3067\u5bfe\u5fdc\u3057 &#8230; <\/p>\n","protected":false},"author":1,"featured_media":1494,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[49],"tags":[106],"views":3229,"_links":{"self":[{"href":"https:\/\/www.beeete2.com\/blog\/index.php?rest_route=\/wp\/v2\/posts\/2731"}],"collection":[{"href":"https:\/\/www.beeete2.com\/blog\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.beeete2.com\/blog\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.beeete2.com\/blog\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.beeete2.com\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=2731"}],"version-history":[{"count":8,"href":"https:\/\/www.beeete2.com\/blog\/index.php?rest_route=\/wp\/v2\/posts\/2731\/revisions"}],"predecessor-version":[{"id":2739,"href":"https:\/\/www.beeete2.com\/blog\/index.php?rest_route=\/wp\/v2\/posts\/2731\/revisions\/2739"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.beeete2.com\/blog\/index.php?rest_route=\/wp\/v2\/media\/1494"}],"wp:attachment":[{"href":"https:\/\/www.beeete2.com\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2731"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.beeete2.com\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=2731"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.beeete2.com\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=2731"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}