fluid_compiler.html 14.8 KB
Newer Older
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


<!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>PaddlePaddle Fluid: Towards a Compiled Programming Language &mdash; PaddlePaddle  documentation</title>
  

  
  

  

  
  
    

  

  
  
    <link rel="stylesheet" href="../_static/css/theme.css" type="text/css" />
  

  
31

32 33 34 35 36 37 38 39 40 41 42 43 44
  
        <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"/> 

  
  <script src="../_static/js/modernizr.min.js"></script>

</head>

<body class="wy-body-for-nav" role="document">

45 46 47 48 49 50 51 52 53 54 55 56 57
  <div class="wy-grid-for-nav">

    
    <nav data-toggle="wy-nav-shift" class="wy-nav-side">
      <div class="wy-side-scroll">
        <div class="wy-side-nav-search">
          

          
            <a href="../index_en.html" class="icon icon-home"> PaddlePaddle
          

          
58 59
          </a>

60 61 62 63 64 65
          
            
            
          

          
66 67 68 69 70 71
<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>
72
</div>
73 74

          
75 76 77 78 79 80 81 82 83 84 85 86
        </div>

        <div class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="main navigation">
          
            
            
                <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="../build_and_install/index_en.html">Install and Build</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="../dev/index_en.html">Development</a></li>
<li class="toctree-l1"><a class="reference internal" href="../faq/index_en.html">FAQ</a></li>
87 88
</ul>

89 90 91 92
            
          
        </div>
      </div>
93 94
    </nav>

95
    <section data-toggle="wy-nav-shift" class="wy-nav-content-wrap">
96

97 98 99 100 101
      
      <nav class="wy-nav-top" role="navigation" aria-label="top navigation">
        <i data-toggle="wy-nav-top" class="fa fa-bars"></i>
        <a href="../index_en.html">PaddlePaddle</a>
      </nav>
102 103


104 105 106 107
      
      <div class="wy-nav-content">
        <div class="rst-content">
          
108

109
 
110 111 112 113 114



<div role="navigation" aria-label="breadcrumbs navigation">
  <ul class="wy-breadcrumbs">
115
    <li><a href="../index_en.html">Docs</a> &raquo;</li>
116 117
      
    <li>PaddlePaddle Fluid: Towards a Compiled Programming Language</li>
118 119 120 121 122 123 124
      <li class="wy-breadcrumbs-aside">
        
          
            <a href="../_sources/design/fluid_compiler.md.txt" rel="nofollow"> View page source</a>
          
        
      </li>
125
  </ul>
126
  <hr/>
127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266
</div>
          <div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article">
           <div itemprop="articleBody">
            
  <div class="section" id="paddlepaddle-fluid-towards-a-compiled-programming-language">
<span id="paddlepaddle-fluid-towards-a-compiled-programming-language"></span><h1>PaddlePaddle Fluid: Towards a Compiled Programming Language<a class="headerlink" href="#paddlepaddle-fluid-towards-a-compiled-programming-language" title="Permalink to this headline"></a></h1>
<p>As described in <a class="reference internal" href="fluid.html"><span class="doc">fluid.md</span></a>, when a Fluid application program
runs, it generates a <code class="docutils literal"><span class="pre">ProgramDesc</span></code> protobuf message as an intermediate
representation of itself.  The C++ class <code class="docutils literal"><span class="pre">Executor</span></code> can run this
protobuf message as an interpreter.  This article describes the Fluid
compiler.</p>
<p><img alt="" src="../_images/fluid-compiler.png" /></p>
<div class="section" id="programdesc">
<span id="programdesc"></span><h2>ProgramDesc<a class="headerlink" href="#programdesc" title="Permalink to this headline"></a></h2>
<p>Before we go deeper into the idea of compiled language, let us take a
look at a simple example Fluid application.</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="s2">&quot;fluid&quot;</span>

<span class="n">func</span> <span class="n">paddlepaddle</span><span class="p">()</span> <span class="p">{</span>
  <span class="n">X</span> <span class="o">=</span> <span class="n">fluid</span><span class="o">.</span><span class="n">read</span><span class="p">(</span><span class="o">...</span><span class="p">)</span>
  <span class="n">W</span> <span class="o">=</span> <span class="n">fluid</span><span class="o">.</span><span class="n">Tensor</span><span class="p">(</span><span class="o">...</span><span class="p">)</span>
  <span class="n">Y</span> <span class="o">=</span> <span class="n">fluid</span><span class="o">.</span><span class="n">mult</span><span class="p">(</span><span class="n">X</span><span class="p">,</span> <span class="n">W</span><span class="p">)</span>
<span class="p">}</span>
</pre></div>
</div>
<p>This program consists of a <a class="reference internal" href="block.html"><span class="doc">block</span></a> of three operators &#8211;
<code class="docutils literal"><span class="pre">read</span></code>, <code class="docutils literal"><span class="pre">assign</span></code>, and <code class="docutils literal"><span class="pre">mult</span></code>.  Its <code class="docutils literal"><span class="pre">ProgramDesc</span></code> message looks like
the following</p>
<div class="highlight-protobuf"><div class="highlight"><pre><span></span><span class="kd">message</span> <span class="nc">ProgramDesc</span> <span class="p">{</span>
  <span class="n">block</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span> <span class="o">=</span> <span class="n">Block</span> <span class="p">{</span>
    <span class="na">vars</span> <span class="o">=</span> <span class="p">[</span><span class="n">X</span><span class="p">,</span> <span class="n">W</span><span class="p">,</span> <span class="n">Y</span><span class="p">],</span>
    <span class="na">ops</span> <span class="o">=</span> <span class="p">[</span>
      <span class="n">read</span><span class="p">(</span><span class="na">output</span> <span class="o">=</span> <span class="n">X</span><span class="p">)</span>
      <span class="n">assign</span><span class="p">(</span><span class="na">input</span> <span class="o">=</span> <span class="o">...</span><span class="p">,</span> <span class="na">output</span> <span class="o">=</span> <span class="n">W</span><span class="p">)</span>
      <span class="n">mult</span><span class="p">(</span><span class="na">input</span> <span class="o">=</span> <span class="p">{</span><span class="n">X</span><span class="p">,</span> <span class="n">W</span><span class="p">},</span> <span class="na">output</span> <span class="o">=</span> <span class="n">Y</span><span class="p">)</span>
    <span class="p">],</span>
  <span class="p">}</span>
<span class="p">}</span>
</pre></div>
</div>
</div>
<div class="section" id="transpilers">
<span id="transpilers"></span><h2>Transpilers<a class="headerlink" href="#transpilers" title="Permalink to this headline"></a></h2>
<p>We can write a transpiler program that takes a <code class="docutils literal"><span class="pre">ProgramDesc</span></code>, e.g.,
the above one, and outputs another <code class="docutils literal"><span class="pre">ProgramDesc</span></code>.  Let us take some
examples:</p>
<ol class="simple">
<li><em>Memory optimization transpiler</em>: We can write a transpiler that
inserts some <code class="docutils literal"><span class="pre">FreeMemoryOp</span></code>s in the above example <code class="docutils literal"><span class="pre">ProgramDesc</span></code> so
to free memory early, before the end of an iteration, so to keep a
small memory footprint.</li>
<li><em>Distributed training transpiler</em>: We can write a transpiler that
converts a<code class="docutils literal"><span class="pre">ProgramDesc</span></code> into its distributed version of two
<code class="docutils literal"><span class="pre">ProgramDesc</span></code>s &#8211; one for running by the trainer processes and the
other for the parameter server.</li>
</ol>
<p>In the rest of this article, we talk about a special kind of
transpiler, <em>Native code generator</em>, which takes a <code class="docutils literal"><span class="pre">ProgramDesc</span></code> and
generates a <code class="docutils literal"><span class="pre">.cu</span></code> (or <code class="docutils literal"><span class="pre">.cc</span></code>) file, which could be built by C++
compilers (gcc, nvcc, icc) into binaries.</p>
</div>
<div class="section" id="native-code-generator">
<span id="native-code-generator"></span><h2>Native Code Generator<a class="headerlink" href="#native-code-generator" title="Permalink to this headline"></a></h2>
<p>For the above example, the native code generator transpiler, say, the
CUDA code generator, should generate a <code class="docutils literal"><span class="pre">main</span></code> function:</p>
<div class="highlight-c++"><div class="highlight"><pre><span></span><span class="kt">void</span> <span class="nf">main</span><span class="p">()</span> <span class="p">{</span>
  <span class="k">auto</span> <span class="n">X</span> <span class="o">=</span> <span class="n">fluid_cuda_read</span><span class="p">(...);</span>
  <span class="k">auto</span> <span class="n">W</span> <span class="o">=</span> <span class="n">fluid_cuda_create_tensor</span><span class="p">(...);</span>
  <span class="k">auto</span> <span class="n">Y</span> <span class="o">=</span> <span class="n">fluid_cuda_mult</span><span class="p">(</span><span class="n">X</span><span class="p">,</span> <span class="n">W</span><span class="p">);</span>
<span class="p">}</span>
</pre></div>
</div>
<p>and the definitions of functions <code class="docutils literal"><span class="pre">fluid_cuda_read</span></code>,
<code class="docutils literal"><span class="pre">fluid_cuda_create_tensor</span></code>, and <code class="docutils literal"><span class="pre">fluid_cuda_mult</span></code>.  Please be aware
that each function could just define a C++ instance of an operator and
run it.  For example</p>
<div class="highlight-c++"><div class="highlight"><pre><span></span><span class="n">paddle</span><span class="o">::</span><span class="n">Tensor</span> <span class="n">fluid_cuda_read</span><span class="p">(...)</span> <span class="p">{</span>
  <span class="n">paddle</span><span class="o">::</span><span class="n">Tensor</span> <span class="n">t</span><span class="p">;</span>
  <span class="n">paddle</span><span class="o">::</span><span class="k">operator</span><span class="o">::</span><span class="n">Read</span> <span class="n">r</span><span class="p">(</span><span class="o">&amp;</span><span class="n">t</span><span class="p">,</span> <span class="p">...);</span>
  <span class="n">r</span><span class="p">.</span><span class="n">Run</span><span class="p">();</span>
  <span class="k">return</span> <span class="n">t</span><span class="p">;</span>
<span class="p">}</span>
</pre></div>
</div>
<p>For computational operators that have multiple <em>kernels</em>, each for a
specific hardware platform, for example, the <code class="docutils literal"><span class="pre">mult</span></code> operator, the
generated code should call its CUDA kernel:</p>
<div class="highlight-c++"><div class="highlight"><pre><span></span><span class="n">paddle</span><span class="o">::</span><span class="n">Tensor</span> <span class="n">fluid_cuda_mult</span><span class="p">(</span><span class="k">const</span> <span class="n">paddle</span><span class="o">::</span><span class="n">Tensor</span><span class="o">&amp;</span> <span class="n">a</span><span class="p">,</span> 
                               <span class="k">const</span> <span class="n">paddle</span><span class="o">::</span><span class="n">Tensor</span><span class="o">&amp;</span> <span class="n">b</span><span class="p">)</span> <span class="p">{</span>
  <span class="n">paddle</span><span class="o">::</span><span class="n">Tensor</span> <span class="n">t</span><span class="p">;</span>
  <span class="n">paddle</span><span class="o">::</span><span class="k">operator</span><span class="o">::</span><span class="n">Mult</span> <span class="n">m</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="p">...);</span>
  <span class="n">Mult</span><span class="p">.</span><span class="n">Run</span><span class="p">(</span><span class="n">cuda_context</span><span class="p">);</span>
<span class="p">}</span>
</pre></div>
</div>
<p>where <code class="docutils literal"><span class="pre">cuda_context</span></code> could be a global variable of type
<code class="docutils literal"><span class="pre">paddle::CUDADeviceContext</span></code>.</p>
</div>
<div class="section" id="multi-block-code-generation">
<span id="multi-block-code-generation"></span><h2>Multi-Block Code Generation<a class="headerlink" href="#multi-block-code-generation" title="Permalink to this headline"></a></h2>
<p>Most Fluid application programs may have more than one blocks.  To
execute them, we need to trace <a class="reference internal" href="scope.html"><span class="doc">scopes</span></a>.</p>
</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',
267
            HAS_SOURCE:  true
268 269 270 271 272 273
        };
    </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>
274

275 276 277 278 279 280
  

  
  
    <script type="text/javascript" src="../_static/js/theme.js"></script>
  
281 282

  
283
  
284 285 286 287 288 289
  <script type="text/javascript">
      jQuery(function () {
          SphinxRtdTheme.StickyNav.enable();
      });
  </script>
   
290 291 292

</body>
</html>