-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathindex.html
More file actions
188 lines (149 loc) · 5.06 KB
/
Copy pathindex.html
File metadata and controls
188 lines (149 loc) · 5.06 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
<!doctype html>
<html>
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0, maximum-scale=1.0, user-scalable=no">
<title>Nvidia-docker Quantlane</title>
<link rel="stylesheet" href="css/reveal.css">
<link rel="stylesheet" href="css/theme/quantlane.css">
<style type="text/css">
section > img {
border: none !important;
box-shadow: none !important;
}
</style>
<!-- Theme used for syntax highlighting of code -->
<link rel="stylesheet" href="lib/css/zenburn.css">
<!-- Printing and PDF exports -->
<script>
var link = document.createElement( 'link' );
link.rel = 'stylesheet';
link.type = 'text/css';
link.href = window.location.search.match( /print-pdf/gi ) ? 'css/print/pdf.css' : 'css/print/paper.css';
document.getElementsByTagName( 'head' )[0].appendChild( link );
</script>
</head>
<body>
<div class="reveal">
<div class="slides">
<section class="intro-slide">
<img class="logo" src="img/logo-white.png" alt="Quantlane">
<h2>NVIDIA in Docker</h2>
<h4>Libor Jonát</h4>
</section>
<section>
<h2>What we wanted?</h2>
<ul>
<li>Run TensorFlow easily</li>
<li>Quickly reproducible environment</li>
<li>Multiple TensorFlow versions and easy upgrades</li>
<li>Project isolation</li>
</ul>
</section>
<section>
<h2>Use Docker</h2>
<ul>
<li>Quickly reproducible environment</li>
<li>Project isolation</li>
<li>Hardware isolation</li>
</ul>
</section>
<section>
<h2>How?</h2>
<a href="https://github.com/NVIDIA/nvidia-docker/">https://github.com/NVIDIA/nvidia-docker/</a>
<ul>
<li>Just a wrapper around the <code>docker</code> command line tool</li>
<li>Supports GPU isolation when you have multiple GPUs</li>
<li>
Specialized Docker images by NVIDIA:
<ul>
<li>Ubuntu</li>
<li>CentOS</li>
</ul>
</li>
</ul>
</section>
<section>
<h2>Usage</h2>
<pre><code>nvidia-docker run -it tensorflow/tensorflow:latest-gpu</code></pre>
<p>If you don’t want to use the nvidia-docker wrapper, you can add the command line arguments manually:</p>
<pre><code>docker run \
--device=/dev/nvidiactl \
--device=/dev/nvidia-uvm \
--device=/dev/nvidia0 \
...</code></pre>
</section>
<section>
<h2>Ready-made images</h2>
<dl>
<dt>Caffe</dt>
<dd><a href="https://github.com/BVLC/caffe/">https://github.com/BVLC/caffe/</a></dd>
<dt>Caffe2</dt>
<dd><a href="https://hub.docker.com/r/caffe2ai/caffe2/">https://hub.docker.com/r/caffe2ai/caffe2/</a></dd>
<dt>TensorFlow</dt>
<dd>
<a href="https://hub.docker.com/r/tensorflow/tensorflow/">https://hub.docker.com/r/tensorflow/tensorflow/</a>
</dd>
<dt>MXNet</dt>
<dd><a href="https://hub.docker.com/r/mxnet/">https://hub.docker.com/r/mxnet/</a></dd>
<dt>PyTorch</dt>
<dd><a href="https://hub.docker.com/r/pytorch/pytorch/">https://hub.docker.com/r/pytorch/pytorch/</a></dd>
</dl>
<p>
We forked the TensorFlow image and added support for Python 3
<a href="https://hub.docker.com/u/quantlane/tensorflow">https://hub.docker.com/u/quantlane/tensorflow</a>
</p>
</section>
<section>
<h2>How to let a non-admin user use this</h2>
<ol>
<li>Add a non-privileged UNIX user…</li>
<li>…that is <strong>not</strong> a member of the <code>docker</code> group.</li>
<li>Have a long-running TensorFlow container</li>
<li>
Allow the non-privileged user to run
<br \>
<code>docker exec -it ...</code>
<br \>
(E.g. using <code>sudo</code>.)
</li>
</ol>
</section>
<section>
<h2>Future</h2>
<ul>
<li>
NVIDIA Docker 2
<ul>
<li>No wrapper command</li>
<li>GPU support for any Docker image</li>
</ul>
</li>
<li>Experimental support for scheduling GPUs in Kubernetes <br \> <a href="https://kubernetes.io/docs/tasks/manage-gpus/scheduling-gpus">https://kubernetes.io/docs/tasks/manage-gpus/scheduling-gpus/</a></li>
</ul>
</section>
<section class="thankyou-slide" data-background-color="#000000">
<img class="logo" src="img/logo-black.png" alt="Quantlane">
<br>
<h4>Thank you</h4>
<a href="libor@quantlane.com"><span class="email-name">libor@</span>quantlane.com</a>
</section>
</div>
</div>
<script src="lib/js/head.min.js"></script>
<script src="js/reveal.js"></script>
<script>
// More info https://github.com/hakimel/reveal.js#configuration
Reveal.initialize({
history: true,
// More info https://github.com/hakimel/reveal.js#dependencies
dependencies: [
{ src: 'plugin/markdown/marked.js' },
{ src: 'plugin/markdown/markdown.js' },
{ src: 'plugin/notes/notes.js', async: true },
{ src: 'plugin/highlight/highlight.js', async: true, callback: function() { hljs.initHighlightingOnLoad(); } }
]
});
</script>
</body>
</html>