Skip to content

P
Paddle

PaddlePaddle / Paddle
大约 1 年 前同步成功

通知 2298
Star 20931
Fork 5422
P
Paddle
提交 730d2b54 编写于 作者: T Travis CI
浏览文件
操作

Deploy to GitHub Pages: 99e3d1ec

上级 5cdaf71b
展开全部 显示空白变更内容
内联 并排
Showing with 908 addition and 2 deletion
develop/doc/_sources/design/ops/dist_train.md.txt 0 → 100644
浏览文件 @ 730d2b54
# Design Doc: Operation Graph Based Parameter Server
## Abstract
We propose an approach to implement the parameter server. In this
approach, there is no fundamental difference between the trainer and
the parameter server: they both run subgraphs, but subgraphs of
different purposes.
## Background
The previous implementations of the parameter server does not run a
subgraph. parameter initialization, optimizer computation, network
communication and checkpointing are implemented twice on both the
trainer and the parameter server.
It would be great if we can write code once and use them on both the
trainer and the parameter server: reduces code duplication and
improves extensibility. Given that after the current refactor, we are
representing everything as a computing graph on the
trainer. Representing everything as a computing graph on the parameter
server becomes a natural extension.
## Design
### Graph Converter
The *graph converter* converts the user-defined operation (OP) graph
into subgraphs to be scheduled on different nodes with the following
steps:
1. OP placement: the OPs will be placed on different nodes according
to heuristic that minimizes estimated total computation
time. Currently we will use a simple heuristic that puts parameter
varable on parameter server workers and everything else on trainer
workers.
1. Add communication OPs to enable the communication between nodes.
We will need these OPs: *Send*, *Recv*, *Enqueue*, *Dequeue*.
Below is an example of converting the user defined graph to the
subgraphs for the trainer and the parameter server:
<img src="src/local-graph.png" width="300"/>
After converting:
<img src="src/dist-graph.png" width="700"/>
1. The parameter variable W and it's optimizer subgraph are placed on the parameter server.
1. Operators are added to the subgraphs.
- *Send* sends data to the connected *Recv* operator. The
scheduler on the receive node will only schedule *Recv* operator
to run when the *Send* operator has ran (the *Send* OP will mark
the *Recv* OP runnable automatically).
- *Enueue* enqueues the input variable, it can block until space
become available in the queue.
- *Dequeue* outputs configurable numbers of tensors from the
queue. It will block until the queue have the required number of
tensors.
### Benefits
- Model parallelism become easier to implement: it's an extension to
the trainer - parameter server approach. we already have the
communication OPs, but need to extend the graph converter's
placement functionality.
- User-defined optimizer is easier to add - user can now express it as
a subgraph.
- No more duplication logic inside the trainer and the parameter
server mentioned in the background section.
### Challenges
- It might be hard for the graph converter to cut a general graph
(without any hint for which subgraph is the optimizer). We may need
to label which subgraph inside the OP graph is the optimizer.
- It's important to balance the parameter shards of on multiple
parameter server. If a single parameter is very big (some
word-embedding, fully connected, softmax layer), we need to
automatically partition the single parameter onto different
parameter servers when possible (only element-wise optimizer depends
on the parameter variable).
### Discussion
- In the "Aync SGD" figure, the "W" variable on the parameter server
could be read and wrote concurrently, what is our locking strategy?
E.g., each variable have a lock cpp method to be invoked by every
OP, or, have a lock OP.
- Can the Enqueue OP be implemented under our current tensor design
(puts the input tensor into the queue tensor)?
- *Dequeue* OP will have variable numbers of output (depends on the
`min_count` attribute), does our current design support it? (similar
question for the *Add* OP)
### References:
[1] [TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems](https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/45166.pdf)
develop/doc/design/ops/dist_train.html 0 → 100644
浏览文件 @ 730d2b54
<!DOCTYPE html>
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Design Doc: Operation Graph Based Parameter Server &mdash; PaddlePaddle documentation</title>
<link rel="stylesheet" href="../../_static/css/theme.css" type="text/css" />
<link rel="index" title="Index"
href="../../genindex.html"/>
<link rel="search" title="Search" href="../../search.html"/>
<link rel="top" title="PaddlePaddle documentation" href="../../index.html"/>
<link rel="stylesheet" href="https://cdn.jsdelivr.net/perfect-scrollbar/0.6.14/css/perfect-scrollbar.min.css" type="text/css" />
<link rel="stylesheet" href="../../_static/css/override.css" type="text/css" />
<script>
var _hmt = _hmt || [];
(function() {
var hm = document.createElement("script");
hm.src = "//hm.baidu.com/hm.js?b9a314ab40d04d805655aab1deee08ba";
var s = document.getElementsByTagName("script")[0];
s.parentNode.insertBefore(hm, s);
})();
</script>
<script src="../../_static/js/modernizr.min.js"></script>
</head>
<body class="wy-body-for-nav" role="document">
<header class="site-header">
<div class="site-logo">
<a href="/"><img src="../../_static/images/PP_w.png"></a>
</div>
<div class="site-nav-links">
<div class="site-menu">
<a class="fork-on-github" href="https://github.com/PaddlePaddle/Paddle" target="_blank"><i class="fa fa-github"></i>Fork me on Github</a>
<div class="language-switcher dropdown">
<a type="button" data-toggle="dropdown">
<span>English</span>
<i class="fa fa-angle-up"></i>
<i class="fa fa-angle-down"></i>
</a>
<ul class="dropdown-menu">
<li><a href="/doc_cn">中文</a></li>
<li><a href="/doc">English</a></li>
</ul>
</div>
<ul class="site-page-links">
<li><a href="/">Home</a></li>
</ul>
</div>
<div class="doc-module">
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../getstarted/index_en.html">GET STARTED</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../howto/index_en.html">HOW TO</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../api/index_en.html">API</a></li>
</ul>
<div role="search">
<form id="rtd-search-form" class="wy-form" action="../../search.html" method="get">
<input type="text" name="q" placeholder="Search docs" />
<input type="hidden" name="check_keywords" value="yes" />
<input type="hidden" name="area" value="default" />
</form>
</div>
</div>
</div>
</header>
<div class="main-content-wrap">
<nav class="doc-menu-vertical" role="navigation">
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../getstarted/index_en.html">GET STARTED</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../../getstarted/build_and_install/index_en.html">Install and Build</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../../getstarted/build_and_install/docker_install_en.html">PaddlePaddle in Docker Containers</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../getstarted/build_and_install/build_from_source_en.html">Installing from Sources</a></li>
</ul>
</li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../howto/index_en.html">HOW TO</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../../howto/usage/cmd_parameter/index_en.html">Set Command-line Parameters</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../../howto/usage/cmd_parameter/use_case_en.html">Use Case</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../howto/usage/cmd_parameter/arguments_en.html">Argument Outline</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../howto/usage/cmd_parameter/detail_introduction_en.html">Detail Description</a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/usage/cluster/cluster_train_en.html">Run Distributed Training</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/usage/k8s/k8s_en.html">Paddle On Kubernetes</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/usage/k8s/k8s_aws_en.html">Distributed PaddlePaddle Training on AWS with Kubernetes</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/dev/build_en.html">Build PaddlePaddle from Source Code and Run Unit Test</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/dev/new_layer_en.html">Write New Layers</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/dev/contribute_to_paddle_en.html">Contribute Code</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/deep_model/rnn/index_en.html">RNN Models</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../../howto/deep_model/rnn/rnn_config_en.html">RNN Configuration</a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/optimization/gpu_profiling_en.html">Tune GPU Performance</a></li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../api/index_en.html">API</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../../api/v2/model_configs.html">Model Configuration</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/config/activation.html">Activation</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/config/layer.html">Layers</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/config/evaluators.html">Evaluators</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/config/optimizer.html">Optimizer</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/config/pooling.html">Pooling</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/config/networks.html">Networks</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/config/attr.html">Parameter Attribute</a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../../api/v2/data.html">Data Reader Interface and DataSets</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../api/v2/run_logic.html">Training and Inference</a></li>
</ul>
</li>
</ul>
</nav>
<section class="doc-content-wrap">
<div role="navigation" aria-label="breadcrumbs navigation">
<ul class="wy-breadcrumbs">
<li>Design Doc: Operation Graph Based Parameter Server</li>
</ul>
</div>
<div class="wy-nav-content" id="doc-content">
<div class="rst-content">
<div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article">
<div itemprop="articleBody">
<div class="section" id="design-doc-operation-graph-based-parameter-server">
<span id="design-doc-operation-graph-based-parameter-server"></span><h1>Design Doc: Operation Graph Based Parameter Server<a class="headerlink" href="#design-doc-operation-graph-based-parameter-server" title="Permalink to this headline"></a></h1>
<div class="section" id="abstract">
<span id="abstract"></span><h2>Abstract<a class="headerlink" href="#abstract" title="Permalink to this headline"></a></h2>
<p>We propose an approach to implement the parameter server. In this
approach, there is no fundamental difference between the trainer and
the parameter server: they both run subgraphs, but subgraphs of
different purposes.</p>
</div>
<div class="section" id="background">
<span id="background"></span><h2>Background<a class="headerlink" href="#background" title="Permalink to this headline"></a></h2>
<p>The previous implementations of the parameter server does not run a
subgraph. parameter initialization, optimizer computation, network
communication and checkpointing are implemented twice on both the
trainer and the parameter server.</p>
<p>It would be great if we can write code once and use them on both the
trainer and the parameter server: reduces code duplication and
improves extensibility. Given that after the current refactor, we are
representing everything as a computing graph on the
trainer. Representing everything as a computing graph on the parameter
server becomes a natural extension.</p>
</div>
<div class="section" id="design">
<span id="design"></span><h2>Design<a class="headerlink" href="#design" title="Permalink to this headline"></a></h2>
<div class="section" id="graph-converter">
<span id="graph-converter"></span><h3>Graph Converter<a class="headerlink" href="#graph-converter" title="Permalink to this headline"></a></h3>
<p>The <em>graph converter</em> converts the user-defined operation (OP) graph
into subgraphs to be scheduled on different nodes with the following
steps:</p>
<ol class="simple">
<li>OP placement: the OPs will be placed on different nodes according
to heuristic that minimizes estimated total computation
time. Currently we will use a simple heuristic that puts parameter
varable on parameter server workers and everything else on trainer
workers.</li>
<li>Add communication OPs to enable the communication between nodes.</li>
</ol>
<p>We will need these OPs: <em>Send</em>, <em>Recv</em>, <em>Enqueue</em>, <em>Dequeue</em>.</p>
<p>Below is an example of converting the user defined graph to the
subgraphs for the trainer and the parameter server:</p>
<p><img src="src/local-graph.png" width="300"/></p>
<p>After converting:</p>
<p><img src="src/dist-graph.png" width="700"/></p>
<ol class="simple">
<li>The parameter variable W and it&#8217;s optimizer subgraph are placed on the parameter server.</li>
<li>Operators are added to the subgraphs.<ul>
<li><em>Send</em> sends data to the connected <em>Recv</em> operator. The
scheduler on the receive node will only schedule <em>Recv</em> operator
to run when the <em>Send</em> operator has ran (the <em>Send</em> OP will mark
the <em>Recv</em> OP runnable automatically).</li>
<li><em>Enueue</em> enqueues the input variable, it can block until space
become available in the queue.</li>
<li><em>Dequeue</em> outputs configurable numbers of tensors from the
queue. It will block until the queue have the required number of
tensors.</li>
</ul>
</li>
</ol>
</div>
<div class="section" id="benefits">
<span id="benefits"></span><h3>Benefits<a class="headerlink" href="#benefits" title="Permalink to this headline"></a></h3>
<ul class="simple">
<li>Model parallelism become easier to implement: it&#8217;s an extension to
the trainer - parameter server approach. we already have the
communication OPs, but need to extend the graph converter&#8217;s
placement functionality.</li>
<li>User-defined optimizer is easier to add - user can now express it as
a subgraph.</li>
<li>No more duplication logic inside the trainer and the parameter
server mentioned in the background section.</li>
</ul>
</div>
<div class="section" id="challenges">
<span id="challenges"></span><h3>Challenges<a class="headerlink" href="#challenges" title="Permalink to this headline"></a></h3>
<ul class="simple">
<li>It might be hard for the graph converter to cut a general graph
(without any hint for which subgraph is the optimizer). We may need
to label which subgraph inside the OP graph is the optimizer.</li>
<li>It&#8217;s important to balance the parameter shards of on multiple
parameter server. If a single parameter is very big (some
word-embedding, fully connected, softmax layer), we need to
automatically partition the single parameter onto different
parameter servers when possible (only element-wise optimizer depends
on the parameter variable).</li>
</ul>
</div>
<div class="section" id="discussion">
<span id="discussion"></span><h3>Discussion<a class="headerlink" href="#discussion" title="Permalink to this headline"></a></h3>
<ul class="simple">
<li>In the &#8220;Aync SGD&#8221; figure, the &#8220;W&#8221; variable on the parameter server
could be read and wrote concurrently, what is our locking strategy?
E.g., each variable have a lock cpp method to be invoked by every
OP, or, have a lock OP.</li>
<li>Can the Enqueue OP be implemented under our current tensor design
(puts the input tensor into the queue tensor)?</li>
<li><em>Dequeue</em> OP will have variable numbers of output (depends on the
<code class="docutils literal"><span class="pre">min_count</span></code> attribute), does our current design support it? (similar
question for the <em>Add</em> OP)</li>
</ul>
</div>
<div class="section" id="references">
<span id="references"></span><h3>References:<a class="headerlink" href="#references" title="Permalink to this headline"></a></h3>
<p>[1] <a class="reference external" href="https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/45166.pdf">TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems</a></p>
</div>
</div>
</div>
</div>
</div>
<footer>
<hr/>
<div role="contentinfo">
<p>
&copy; Copyright 2016, PaddlePaddle developers.
</p>
</div>
Built with <a href="http://sphinx-doc.org/">Sphinx</a> using a <a href="https://github.com/snide/sphinx_rtd_theme">theme</a> provided by <a href="https://readthedocs.org">Read the Docs</a>.
</footer>
</div>
</div>
</section>
</div>
<script type="text/javascript">
var DOCUMENTATION_OPTIONS = {
URL_ROOT:'../../',
VERSION:'',
COLLAPSE_INDEX:false,
FILE_SUFFIX:'.html',
HAS_SOURCE: true,
SOURCELINK_SUFFIX: ".txt",
};
</script>
<script type="text/javascript" src="../../_static/jquery.js"></script>
<script type="text/javascript" src="../../_static/underscore.js"></script>
<script type="text/javascript" src="../../_static/doctools.js"></script>
<script type="text/javascript" src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.0/MathJax.js?config=TeX-AMS-MML_HTMLorMML"></script>
<script type="text/javascript" src="../../_static/js/theme.js"></script>
<script src="https://maxcdn.bootstrapcdn.com/bootstrap/3.3.7/js/bootstrap.min.js" integrity="sha384-Tc5IQib027qvyjSMfHjOMaLkfuWVxZxUPnCJA7l2mCWNIpG9mGCD8wGNIcPD7Txa" crossorigin="anonymous"></script>
<script src="https://cdn.jsdelivr.net/perfect-scrollbar/0.6.14/js/perfect-scrollbar.jquery.min.js"></script>
<script src="../../_static/js/paddle_doc_init.js"></script>
</body>
</html>
\ No newline at end of file
develop/doc/searchindex.js
浏览文件 @ 730d2b54
因为 它太大了无法显示 source diff 。你可以改为 查看blob。
develop/doc_cn/_sources/design/ops/dist_train.md.txt 0 → 100644
浏览文件 @ 730d2b54
# Design Doc: Operation Graph Based Parameter Server
## Abstract
We propose an approach to implement the parameter server. In this
approach, there is no fundamental difference between the trainer and
the parameter server: they both run subgraphs, but subgraphs of
different purposes.
## Background
The previous implementations of the parameter server does not run a
subgraph. parameter initialization, optimizer computation, network
communication and checkpointing are implemented twice on both the
trainer and the parameter server.
It would be great if we can write code once and use them on both the
trainer and the parameter server: reduces code duplication and
improves extensibility. Given that after the current refactor, we are
representing everything as a computing graph on the
trainer. Representing everything as a computing graph on the parameter
server becomes a natural extension.
## Design
### Graph Converter
The *graph converter* converts the user-defined operation (OP) graph
into subgraphs to be scheduled on different nodes with the following
steps:
1. OP placement: the OPs will be placed on different nodes according
to heuristic that minimizes estimated total computation
time. Currently we will use a simple heuristic that puts parameter
varable on parameter server workers and everything else on trainer
workers.
1. Add communication OPs to enable the communication between nodes.
We will need these OPs: *Send*, *Recv*, *Enqueue*, *Dequeue*.
Below is an example of converting the user defined graph to the
subgraphs for the trainer and the parameter server:
<img src="src/local-graph.png" width="300"/>
After converting:
<img src="src/dist-graph.png" width="700"/>
1. The parameter variable W and it's optimizer subgraph are placed on the parameter server.
1. Operators are added to the subgraphs.
- *Send* sends data to the connected *Recv* operator. The
scheduler on the receive node will only schedule *Recv* operator
to run when the *Send* operator has ran (the *Send* OP will mark
the *Recv* OP runnable automatically).
- *Enueue* enqueues the input variable, it can block until space
become available in the queue.
- *Dequeue* outputs configurable numbers of tensors from the
queue. It will block until the queue have the required number of
tensors.
### Benefits
- Model parallelism become easier to implement: it's an extension to
the trainer - parameter server approach. we already have the
communication OPs, but need to extend the graph converter's
placement functionality.
- User-defined optimizer is easier to add - user can now express it as
a subgraph.
- No more duplication logic inside the trainer and the parameter
server mentioned in the background section.
### Challenges
- It might be hard for the graph converter to cut a general graph
(without any hint for which subgraph is the optimizer). We may need
to label which subgraph inside the OP graph is the optimizer.
- It's important to balance the parameter shards of on multiple
parameter server. If a single parameter is very big (some
word-embedding, fully connected, softmax layer), we need to
automatically partition the single parameter onto different
parameter servers when possible (only element-wise optimizer depends
on the parameter variable).
### Discussion
- In the "Aync SGD" figure, the "W" variable on the parameter server
could be read and wrote concurrently, what is our locking strategy?
E.g., each variable have a lock cpp method to be invoked by every
OP, or, have a lock OP.
- Can the Enqueue OP be implemented under our current tensor design
(puts the input tensor into the queue tensor)?
- *Dequeue* OP will have variable numbers of output (depends on the
`min_count` attribute), does our current design support it? (similar
question for the *Add* OP)
### References:
[1] [TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems](https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/45166.pdf)
develop/doc_cn/design/ops/dist_train.html 0 → 100644
浏览文件 @ 730d2b54
<!DOCTYPE html>
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Design Doc: Operation Graph Based Parameter Server &mdash; PaddlePaddle 文档</title>
<link rel="stylesheet" href="../../_static/css/theme.css" type="text/css" />
<link rel="index" title="索引"
href="../../genindex.html"/>
<link rel="search" title="搜索" href="../../search.html"/>
<link rel="top" title="PaddlePaddle 文档" href="../../index.html"/>
<link rel="stylesheet" href="https://cdn.jsdelivr.net/perfect-scrollbar/0.6.14/css/perfect-scrollbar.min.css" type="text/css" />
<link rel="stylesheet" href="../../_static/css/override.css" type="text/css" />
<script>
var _hmt = _hmt || [];
(function() {
var hm = document.createElement("script");
hm.src = "//hm.baidu.com/hm.js?b9a314ab40d04d805655aab1deee08ba";
var s = document.getElementsByTagName("script")[0];
s.parentNode.insertBefore(hm, s);
})();
</script>
<script src="../../_static/js/modernizr.min.js"></script>
</head>
<body class="wy-body-for-nav" role="document">
<header class="site-header">
<div class="site-logo">
<a href="/"><img src="../../_static/images/PP_w.png"></a>
</div>
<div class="site-nav-links">
<div class="site-menu">
<a class="fork-on-github" href="https://github.com/PaddlePaddle/Paddle" target="_blank"><i class="fa fa-github"></i>Fork me on Github</a>
<div class="language-switcher dropdown">
<a type="button" data-toggle="dropdown">
<span>English</span>
<i class="fa fa-angle-up"></i>
<i class="fa fa-angle-down"></i>
</a>
<ul class="dropdown-menu">
<li><a href="/doc_cn">中文</a></li>
<li><a href="/doc">English</a></li>
</ul>
</div>
<ul class="site-page-links">
<li><a href="/">Home</a></li>
</ul>
</div>
<div class="doc-module">
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../getstarted/index_cn.html">新手入门</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../howto/index_cn.html">进阶指南</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../api/index_cn.html">API</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../faq/index_cn.html">FAQ</a></li>
</ul>
<div role="search">
<form id="rtd-search-form" class="wy-form" action="../../search.html" method="get">
<input type="text" name="q" placeholder="Search docs" />
<input type="hidden" name="check_keywords" value="yes" />
<input type="hidden" name="area" value="default" />
</form>
</div>
</div>
</div>
</header>
<div class="main-content-wrap">
<nav class="doc-menu-vertical" role="navigation">
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../getstarted/index_cn.html">新手入门</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../../getstarted/build_and_install/index_cn.html">安装与编译</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../../getstarted/build_and_install/docker_install_cn.html">PaddlePaddle的Docker容器使用方式</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../getstarted/build_and_install/cmake/build_from_source_cn.html">PaddlePaddle的编译选项</a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../../getstarted/concepts/use_concepts_cn.html">基本使用概念</a></li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../howto/index_cn.html">进阶指南</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../../howto/usage/cmd_parameter/index_cn.html">设置命令行参数</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../../howto/usage/cmd_parameter/use_case_cn.html">使用案例</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../howto/usage/cmd_parameter/arguments_cn.html">参数概述</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../howto/usage/cmd_parameter/detail_introduction_cn.html">细节描述</a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/usage/cluster/cluster_train_cn.html">运行分布式训练</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/usage/k8s/k8s_basis_cn.html">Kubernetes 简介</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/usage/k8s/k8s_cn.html">Kubernetes单机训练</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/usage/k8s/k8s_distributed_cn.html">Kubernetes分布式训练</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/dev/build_cn.html">编译PaddlePaddle和运行单元测试</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/dev/write_docs_cn.html">如何贡献/修改文档</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/dev/contribute_to_paddle_cn.html">如何贡献代码</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/deep_model/rnn/index_cn.html">RNN相关模型</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../../howto/deep_model/rnn/rnn_config_cn.html">RNN配置</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../howto/deep_model/rnn/recurrent_group_cn.html">Recurrent Group教程</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../howto/deep_model/rnn/hierarchical_layer_cn.html">支持双层序列作为输入的Layer</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../howto/deep_model/rnn/hrnn_rnn_api_compare_cn.html">单双层RNN API对比介绍</a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/optimization/gpu_profiling_cn.html">GPU性能分析与调优</a></li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../api/index_cn.html">API</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../../api/v2/model_configs.html">模型配置</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/config/activation.html">Activation</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/config/layer.html">Layers</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/config/evaluators.html">Evaluators</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/config/optimizer.html">Optimizer</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/config/pooling.html">Pooling</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/config/networks.html">Networks</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/config/attr.html">Parameter Attribute</a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../../api/v2/data.html">数据访问</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../api/v2/run_logic.html">训练与应用</a></li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../faq/index_cn.html">FAQ</a></li>
</ul>
</nav>
<section class="doc-content-wrap">
<div role="navigation" aria-label="breadcrumbs navigation">
<ul class="wy-breadcrumbs">
<li>Design Doc: Operation Graph Based Parameter Server</li>
</ul>
</div>
<div class="wy-nav-content" id="doc-content">
<div class="rst-content">
<div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article">
<div itemprop="articleBody">
<div class="section" id="design-doc-operation-graph-based-parameter-server">
<span id="design-doc-operation-graph-based-parameter-server"></span><h1>Design Doc: Operation Graph Based Parameter Server<a class="headerlink" href="#design-doc-operation-graph-based-parameter-server" title="永久链接至标题"></a></h1>
<div class="section" id="abstract">
<span id="abstract"></span><h2>Abstract<a class="headerlink" href="#abstract" title="永久链接至标题"></a></h2>
<p>We propose an approach to implement the parameter server. In this
approach, there is no fundamental difference between the trainer and
the parameter server: they both run subgraphs, but subgraphs of
different purposes.</p>
</div>
<div class="section" id="background">
<span id="background"></span><h2>Background<a class="headerlink" href="#background" title="永久链接至标题"></a></h2>
<p>The previous implementations of the parameter server does not run a
subgraph. parameter initialization, optimizer computation, network
communication and checkpointing are implemented twice on both the
trainer and the parameter server.</p>
<p>It would be great if we can write code once and use them on both the
trainer and the parameter server: reduces code duplication and
improves extensibility. Given that after the current refactor, we are
representing everything as a computing graph on the
trainer. Representing everything as a computing graph on the parameter
server becomes a natural extension.</p>
</div>
<div class="section" id="design">
<span id="design"></span><h2>Design<a class="headerlink" href="#design" title="永久链接至标题"></a></h2>
<div class="section" id="graph-converter">
<span id="graph-converter"></span><h3>Graph Converter<a class="headerlink" href="#graph-converter" title="永久链接至标题"></a></h3>
<p>The <em>graph converter</em> converts the user-defined operation (OP) graph
into subgraphs to be scheduled on different nodes with the following
steps:</p>
<ol class="simple">
<li>OP placement: the OPs will be placed on different nodes according
to heuristic that minimizes estimated total computation
time. Currently we will use a simple heuristic that puts parameter
varable on parameter server workers and everything else on trainer
workers.</li>
<li>Add communication OPs to enable the communication between nodes.</li>
</ol>
<p>We will need these OPs: <em>Send</em>, <em>Recv</em>, <em>Enqueue</em>, <em>Dequeue</em>.</p>
<p>Below is an example of converting the user defined graph to the
subgraphs for the trainer and the parameter server:</p>
<p><img src="src/local-graph.png" width="300"/></p>
<p>After converting:</p>
<p><img src="src/dist-graph.png" width="700"/></p>
<ol class="simple">
<li>The parameter variable W and it&#8217;s optimizer subgraph are placed on the parameter server.</li>
<li>Operators are added to the subgraphs.<ul>
<li><em>Send</em> sends data to the connected <em>Recv</em> operator. The
scheduler on the receive node will only schedule <em>Recv</em> operator
to run when the <em>Send</em> operator has ran (the <em>Send</em> OP will mark
the <em>Recv</em> OP runnable automatically).</li>
<li><em>Enueue</em> enqueues the input variable, it can block until space
become available in the queue.</li>
<li><em>Dequeue</em> outputs configurable numbers of tensors from the
queue. It will block until the queue have the required number of
tensors.</li>
</ul>
</li>
</ol>
</div>
<div class="section" id="benefits">
<span id="benefits"></span><h3>Benefits<a class="headerlink" href="#benefits" title="永久链接至标题"></a></h3>
<ul class="simple">
<li>Model parallelism become easier to implement: it&#8217;s an extension to
the trainer - parameter server approach. we already have the
communication OPs, but need to extend the graph converter&#8217;s
placement functionality.</li>
<li>User-defined optimizer is easier to add - user can now express it as
a subgraph.</li>
<li>No more duplication logic inside the trainer and the parameter
server mentioned in the background section.</li>
</ul>
</div>
<div class="section" id="challenges">
<span id="challenges"></span><h3>Challenges<a class="headerlink" href="#challenges" title="永久链接至标题"></a></h3>
<ul class="simple">
<li>It might be hard for the graph converter to cut a general graph
(without any hint for which subgraph is the optimizer). We may need
to label which subgraph inside the OP graph is the optimizer.</li>
<li>It&#8217;s important to balance the parameter shards of on multiple
parameter server. If a single parameter is very big (some
word-embedding, fully connected, softmax layer), we need to
automatically partition the single parameter onto different
parameter servers when possible (only element-wise optimizer depends
on the parameter variable).</li>
</ul>
</div>
<div class="section" id="discussion">
<span id="discussion"></span><h3>Discussion<a class="headerlink" href="#discussion" title="永久链接至标题"></a></h3>
<ul class="simple">
<li>In the &#8220;Aync SGD&#8221; figure, the &#8220;W&#8221; variable on the parameter server
could be read and wrote concurrently, what is our locking strategy?
E.g., each variable have a lock cpp method to be invoked by every
OP, or, have a lock OP.</li>
<li>Can the Enqueue OP be implemented under our current tensor design
(puts the input tensor into the queue tensor)?</li>
<li><em>Dequeue</em> OP will have variable numbers of output (depends on the
<code class="docutils literal"><span class="pre">min_count</span></code> attribute), does our current design support it? (similar
question for the <em>Add</em> OP)</li>
</ul>
</div>
<div class="section" id="references">
<span id="references"></span><h3>References:<a class="headerlink" href="#references" title="永久链接至标题"></a></h3>
<p>[1] <a class="reference external" href="https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/45166.pdf">TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems</a></p>
</div>
</div>
</div>
</div>
</div>
<footer>
<hr/>
<div role="contentinfo">
<p>
&copy; Copyright 2016, PaddlePaddle developers.
</p>
</div>
Built with <a href="http://sphinx-doc.org/">Sphinx</a> using a <a href="https://github.com/snide/sphinx_rtd_theme">theme</a> provided by <a href="https://readthedocs.org">Read the Docs</a>.
</footer>
</div>
</div>
</section>
</div>
<script type="text/javascript">
var DOCUMENTATION_OPTIONS = {
URL_ROOT:'../../',
VERSION:'',
COLLAPSE_INDEX:false,
FILE_SUFFIX:'.html',
HAS_SOURCE: true,
SOURCELINK_SUFFIX: ".txt",
};
</script>
<script type="text/javascript" src="../../_static/jquery.js"></script>
<script type="text/javascript" src="../../_static/underscore.js"></script>
<script type="text/javascript" src="../../_static/doctools.js"></script>
<script type="text/javascript" src="../../_static/translations.js"></script>
<script type="text/javascript" src="https://cdn.bootcss.com/mathjax/2.7.0/MathJax.js"></script>
<script type="text/javascript" src="../../_static/js/theme.js"></script>
<script src="https://maxcdn.bootstrapcdn.com/bootstrap/3.3.7/js/bootstrap.min.js" integrity="sha384-Tc5IQib027qvyjSMfHjOMaLkfuWVxZxUPnCJA7l2mCWNIpG9mGCD8wGNIcPD7Txa" crossorigin="anonymous"></script>
<script src="https://cdn.jsdelivr.net/perfect-scrollbar/0.6.14/js/perfect-scrollbar.jquery.min.js"></script>
<script src="../../_static/js/paddle_doc_init.js"></script>
</body>
</html>
\ No newline at end of file
develop/doc_cn/searchindex.js
浏览文件 @ 730d2b54
此差异已折叠。
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或