pydoc/2.47.0/_modules/apache_beam/ml/inference/tensorrt_inference.html - beam-site - Git at Google



 <!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>apache_beam.ml.inference.tensorrt_inference &mdash; Apache Beam 2.47.0 documentation</title>


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


       <script type="text/javascript" id="documentation_options" data-url_root="../../../../" src="../../../../_static/documentation_options.js"></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/language_data.js"></script>
         <script async="async" type="text/javascript" src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/latest.js?config=TeX-AMS-MML_HTMLorMML"></script>

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


   <link rel="stylesheet" href="../../../../_static/css/theme.css" type="text/css" />
   <link rel="stylesheet" href="../../../../_static/pygments.css" type="text/css" />
     <link rel="index" title="Index" href="../../../../genindex.html" />
     <link rel="search" title="Search" href="../../../../search.html" />
 </head>

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


   <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.html" class="icon icon-home"> Apache Beam


           </a>


               <div class="version">
                 2.47.0
               </div>


 <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 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="../../../../apache_beam.coders.html">apache_beam.coders package</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../apache_beam.dataframe.html">apache_beam.dataframe package</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../apache_beam.io.html">apache_beam.io package</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../apache_beam.metrics.html">apache_beam.metrics package</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../apache_beam.ml.html">apache_beam.ml package</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../apache_beam.options.html">apache_beam.options package</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../apache_beam.portability.html">apache_beam.portability package</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../apache_beam.runners.html">apache_beam.runners package</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../apache_beam.testing.html">apache_beam.testing package</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../apache_beam.transforms.html">apache_beam.transforms package</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../apache_beam.typehints.html">apache_beam.typehints package</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../apache_beam.utils.html">apache_beam.utils package</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../apache_beam.yaml.html">apache_beam.yaml package</a></li>
 </ul>
 <ul>
 <li class="toctree-l1"><a class="reference internal" href="../../../../apache_beam.error.html">apache_beam.error module</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../apache_beam.pipeline.html">apache_beam.pipeline module</a></li>
 <li class="toctree-l1"><a class="reference internal" href="../../../../apache_beam.pvalue.html">apache_beam.pvalue module</a></li>
 </ul>


         </div>
       </div>
     </nav>

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


       <nav class="wy-nav-top" aria-label="top navigation">

           <i data-toggle="wy-nav-top" class="fa fa-bars"></i>
           <a href="../../../../index.html">Apache Beam</a>

       </nav>


       <div class="wy-nav-content">

         <div class="rst-content">


 <div role="navigation" aria-label="breadcrumbs navigation">

   <ul class="wy-breadcrumbs">

       <li><a href="../../../../index.html">Docs</a> &raquo;</li>

           <li><a href="../../../index.html">Module code</a> &raquo;</li>

       <li>apache_beam.ml.inference.tensorrt_inference</li>


       <li class="wy-breadcrumbs-aside">

       </li>

   </ul>


   <hr/>
 </div>
           <div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article">
            <div itemprop="articleBody">

   <h1>Source code for apache_beam.ml.inference.tensorrt_inference</h1><div class="highlight"><pre>
 <span></span><span class="c1">#</span>
 <span class="c1"># Licensed to the Apache Software Foundation (ASF) under one or more</span>
 <span class="c1"># contributor license agreements.  See the NOTICE file distributed with</span>
 <span class="c1"># this work for additional information regarding copyright ownership.</span>
 <span class="c1"># The ASF licenses this file to You under the Apache License, Version 2.0</span>
 <span class="c1"># (the &quot;License&quot;); you may not use this file except in compliance with</span>
 <span class="c1"># the License.  You may obtain a copy of the License at</span>
 <span class="c1">#</span>
 <span class="c1">#    http://www.apache.org/licenses/LICENSE-2.0</span>
 <span class="c1">#</span>
 <span class="c1"># Unless required by applicable law or agreed to in writing, software</span>
 <span class="c1"># distributed under the License is distributed on an &quot;AS IS&quot; BASIS,</span>
 <span class="c1"># WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
 <span class="c1"># See the License for the specific language governing permissions and</span>
 <span class="c1"># limitations under the License.</span>
 <span class="c1">#</span>

 <span class="c1"># pytype: skip-file</span>

 <span class="kn">from</span> <span class="nn">__future__</span> <span class="kn">import</span> <span class="n">annotations</span>

 <span class="kn">import</span> <span class="nn">logging</span>
 <span class="kn">import</span> <span class="nn">threading</span>
 <span class="kn">from</span> <span class="nn">typing</span> <span class="kn">import</span> <span class="n">Any</span>
 <span class="kn">from</span> <span class="nn">typing</span> <span class="kn">import</span> <span class="n">Callable</span>
 <span class="kn">from</span> <span class="nn">typing</span> <span class="kn">import</span> <span class="n">Dict</span>
 <span class="kn">from</span> <span class="nn">typing</span> <span class="kn">import</span> <span class="n">Iterable</span>
 <span class="kn">from</span> <span class="nn">typing</span> <span class="kn">import</span> <span class="n">Optional</span>
 <span class="kn">from</span> <span class="nn">typing</span> <span class="kn">import</span> <span class="n">Sequence</span>
 <span class="kn">from</span> <span class="nn">typing</span> <span class="kn">import</span> <span class="n">Tuple</span>

 <span class="kn">import</span> <span class="nn">numpy</span> <span class="k">as</span> <span class="nn">np</span>

 <span class="kn">from</span> <span class="nn">apache_beam.io.filesystems</span> <span class="kn">import</span> <span class="n">FileSystems</span>
 <span class="kn">from</span> <span class="nn">apache_beam.ml.inference</span> <span class="kn">import</span> <span class="n">utils</span>
 <span class="kn">from</span> <span class="nn">apache_beam.ml.inference.base</span> <span class="kn">import</span> <span class="n">ModelHandler</span>
 <span class="kn">from</span> <span class="nn">apache_beam.ml.inference.base</span> <span class="kn">import</span> <span class="n">PredictionResult</span>
 <span class="kn">from</span> <span class="nn">apache_beam.utils.annotations</span> <span class="kn">import</span> <span class="n">experimental</span>

 <span class="n">LOGGER</span> <span class="o">=</span> <span class="n">logging</span><span class="o">.</span><span class="n">getLogger</span><span class="p">(</span><span class="s2">&quot;TensorRTEngineHandlerNumPy&quot;</span><span class="p">)</span>
 <span class="c1"># This try/catch block allows users to submit jobs from a machine without</span>
 <span class="c1"># GPU and other dependencies (tensorrt, cuda, etc.) at job submission time.</span>
 <span class="k">try</span><span class="p">:</span>
   <span class="kn">import</span> <span class="nn">tensorrt</span> <span class="k">as</span> <span class="nn">trt</span>
   <span class="n">TRT_LOGGER</span> <span class="o">=</span> <span class="n">trt</span><span class="o">.</span><span class="n">Logger</span><span class="p">(</span><span class="n">trt</span><span class="o">.</span><span class="n">Logger</span><span class="o">.</span><span class="n">INFO</span><span class="p">)</span>
   <span class="n">trt</span><span class="o">.</span><span class="n">init_libnvinfer_plugins</span><span class="p">(</span><span class="n">TRT_LOGGER</span><span class="p">,</span> <span class="n">namespace</span><span class="o">=</span><span class="s2">&quot;&quot;</span><span class="p">)</span>
   <span class="n">LOGGER</span><span class="o">.</span><span class="n">info</span><span class="p">(</span><span class="s1">&#39;tensorrt module successfully imported.&#39;</span><span class="p">)</span>
 <span class="k">except</span> <span class="ne">ModuleNotFoundError</span><span class="p">:</span>
   <span class="n">TRT_LOGGER</span> <span class="o">=</span> <span class="kc">None</span>
   <span class="n">msg</span> <span class="o">=</span> <span class="s1">&#39;tensorrt module was not found. This is ok as long as the specified &#39;</span> \
     <span class="s1">&#39;runner has tensorrt dependencies installed.&#39;</span>
   <span class="n">LOGGER</span><span class="o">.</span><span class="n">warning</span><span class="p">(</span><span class="n">msg</span><span class="p">)</span>


 <span class="k">def</span> <span class="nf">_load_engine</span><span class="p">(</span><span class="n">engine_path</span><span class="p">):</span>
   <span class="kn">import</span> <span class="nn">tensorrt</span> <span class="k">as</span> <span class="nn">trt</span>
   <span class="n">file</span> <span class="o">=</span> <span class="n">FileSystems</span><span class="o">.</span><span class="n">open</span><span class="p">(</span><span class="n">engine_path</span><span class="p">,</span> <span class="s1">&#39;rb&#39;</span><span class="p">)</span>
   <span class="n">runtime</span> <span class="o">=</span> <span class="n">trt</span><span class="o">.</span><span class="n">Runtime</span><span class="p">(</span><span class="n">TRT_LOGGER</span><span class="p">)</span>
   <span class="n">engine</span> <span class="o">=</span> <span class="n">runtime</span><span class="o">.</span><span class="n">deserialize_cuda_engine</span><span class="p">(</span><span class="n">file</span><span class="o">.</span><span class="n">read</span><span class="p">())</span>
   <span class="k">assert</span> <span class="n">engine</span>
   <span class="k">return</span> <span class="n">engine</span>


 <span class="k">def</span> <span class="nf">_load_onnx</span><span class="p">(</span><span class="n">onnx_path</span><span class="p">):</span>
   <span class="kn">import</span> <span class="nn">tensorrt</span> <span class="k">as</span> <span class="nn">trt</span>
   <span class="n">builder</span> <span class="o">=</span> <span class="n">trt</span><span class="o">.</span><span class="n">Builder</span><span class="p">(</span><span class="n">TRT_LOGGER</span><span class="p">)</span>
   <span class="n">network</span> <span class="o">=</span> <span class="n">builder</span><span class="o">.</span><span class="n">create_network</span><span class="p">(</span>
       <span class="n">flags</span><span class="o">=</span><span class="mi">1</span> <span class="o">&lt;&lt;</span> <span class="nb">int</span><span class="p">(</span><span class="n">trt</span><span class="o">.</span><span class="n">NetworkDefinitionCreationFlag</span><span class="o">.</span><span class="n">EXPLICIT_BATCH</span><span class="p">))</span>
   <span class="n">parser</span> <span class="o">=</span> <span class="n">trt</span><span class="o">.</span><span class="n">OnnxParser</span><span class="p">(</span><span class="n">network</span><span class="p">,</span> <span class="n">TRT_LOGGER</span><span class="p">)</span>
   <span class="k">with</span> <span class="n">FileSystems</span><span class="o">.</span><span class="n">open</span><span class="p">(</span><span class="n">onnx_path</span><span class="p">)</span> <span class="k">as</span> <span class="n">f</span><span class="p">:</span>
     <span class="k">if</span> <span class="ow">not</span> <span class="n">parser</span><span class="o">.</span><span class="n">parse</span><span class="p">(</span><span class="n">f</span><span class="o">.</span><span class="n">read</span><span class="p">()):</span>
       <span class="n">LOGGER</span><span class="o">.</span><span class="n">error</span><span class="p">(</span><span class="s2">&quot;Failed to load ONNX file: </span><span class="si">%s</span><span class="s2">&quot;</span><span class="p">,</span> <span class="n">onnx_path</span><span class="p">)</span>
       <span class="k">for</span> <span class="n">error</span> <span class="ow">in</span> <span class="nb">range</span><span class="p">(</span><span class="n">parser</span><span class="o">.</span><span class="n">num_errors</span><span class="p">):</span>
         <span class="n">LOGGER</span><span class="o">.</span><span class="n">error</span><span class="p">(</span><span class="n">parser</span><span class="o">.</span><span class="n">get_error</span><span class="p">(</span><span class="n">error</span><span class="p">))</span>
       <span class="k">raise</span> <span class="ne">ValueError</span><span class="p">(</span><span class="sa">f</span><span class="s2">&quot;Failed to load ONNX file: </span><span class="si">{</span><span class="n">onnx_path</span><span class="si">}</span><span class="s2">&quot;</span><span class="p">)</span>
   <span class="k">return</span> <span class="n">network</span><span class="p">,</span> <span class="n">builder</span>


 <span class="k">def</span> <span class="nf">_build_engine</span><span class="p">(</span><span class="n">network</span><span class="p">,</span> <span class="n">builder</span><span class="p">):</span>
   <span class="kn">import</span> <span class="nn">tensorrt</span> <span class="k">as</span> <span class="nn">trt</span>
   <span class="n">config</span> <span class="o">=</span> <span class="n">builder</span><span class="o">.</span><span class="n">create_builder_config</span><span class="p">()</span>
   <span class="n">runtime</span> <span class="o">=</span> <span class="n">trt</span><span class="o">.</span><span class="n">Runtime</span><span class="p">(</span><span class="n">TRT_LOGGER</span><span class="p">)</span>
   <span class="n">plan</span> <span class="o">=</span> <span class="n">builder</span><span class="o">.</span><span class="n">build_serialized_network</span><span class="p">(</span><span class="n">network</span><span class="p">,</span> <span class="n">config</span><span class="p">)</span>
   <span class="n">engine</span> <span class="o">=</span> <span class="n">runtime</span><span class="o">.</span><span class="n">deserialize_cuda_engine</span><span class="p">(</span><span class="n">plan</span><span class="p">)</span>
   <span class="n">builder</span><span class="o">.</span><span class="n">reset</span><span class="p">()</span>
   <span class="k">return</span> <span class="n">engine</span>


 <span class="k">def</span> <span class="nf">_assign_or_fail</span><span class="p">(</span><span class="n">args</span><span class="p">):</span>
 <span class="w">  </span><span class="sd">&quot;&quot;&quot;CUDA error checking.&quot;&quot;&quot;</span>
   <span class="kn">from</span> <span class="nn">cuda</span> <span class="kn">import</span> <span class="n">cuda</span>
   <span class="n">err</span><span class="p">,</span> <span class="n">ret</span> <span class="o">=</span> <span class="n">args</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> <span class="n">args</span><span class="p">[</span><span class="mi">1</span><span class="p">:]</span>
   <span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">err</span><span class="p">,</span> <span class="n">cuda</span><span class="o">.</span><span class="n">CUresult</span><span class="p">):</span>
     <span class="k">if</span> <span class="n">err</span> <span class="o">!=</span> <span class="n">cuda</span><span class="o">.</span><span class="n">CUresult</span><span class="o">.</span><span class="n">CUDA_SUCCESS</span><span class="p">:</span>
       <span class="k">raise</span> <span class="ne">RuntimeError</span><span class="p">(</span><span class="s2">&quot;Cuda Error: </span><span class="si">{}</span><span class="s2">&quot;</span><span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="n">err</span><span class="p">))</span>
   <span class="k">else</span><span class="p">:</span>
     <span class="k">raise</span> <span class="ne">RuntimeError</span><span class="p">(</span><span class="s2">&quot;Unknown error type: </span><span class="si">{}</span><span class="s2">&quot;</span><span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="n">err</span><span class="p">))</span>
   <span class="c1"># Special case so that no unpacking is needed at call-site.</span>
   <span class="k">if</span> <span class="nb">len</span><span class="p">(</span><span class="n">ret</span><span class="p">)</span> <span class="o">==</span> <span class="mi">1</span><span class="p">:</span>
     <span class="k">return</span> <span class="n">ret</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span>
   <span class="k">return</span> <span class="n">ret</span>


 <div class="viewcode-block" id="TensorRTEngine"><a class="viewcode-back" href="../../../../apache_beam.ml.inference.tensorrt_inference.html#apache_beam.ml.inference.tensorrt_inference.TensorRTEngine">[docs]</a><span class="k">class</span> <span class="nc">TensorRTEngine</span><span class="p">:</span>
   <span class="k">def</span> <span class="fm">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">engine</span><span class="p">:</span> <span class="n">trt</span><span class="o">.</span><span class="n">ICudaEngine</span><span class="p">):</span>
 <span class="w">    </span><span class="sd">&quot;&quot;&quot;Implementation of the TensorRTEngine class which handles</span>
 <span class="sd">    allocations associated with TensorRT engine.</span>

 <span class="sd">    Example Usage::</span>

 <span class="sd">      TensorRTEngine(engine)</span>

 <span class="sd">    Args:</span>
 <span class="sd">      engine: trt.ICudaEngine object that contains TensorRT engine</span>
 <span class="sd">    &quot;&quot;&quot;</span>
     <span class="kn">from</span> <span class="nn">cuda</span> <span class="kn">import</span> <span class="n">cuda</span>
     <span class="kn">import</span> <span class="nn">tensorrt</span> <span class="k">as</span> <span class="nn">trt</span>
     <span class="bp">self</span><span class="o">.</span><span class="n">engine</span> <span class="o">=</span> <span class="n">engine</span>
     <span class="bp">self</span><span class="o">.</span><span class="n">context</span> <span class="o">=</span> <span class="n">engine</span><span class="o">.</span><span class="n">create_execution_context</span><span class="p">()</span>
     <span class="bp">self</span><span class="o">.</span><span class="n">context_lock</span> <span class="o">=</span> <span class="n">threading</span><span class="o">.</span><span class="n">RLock</span><span class="p">()</span>
     <span class="bp">self</span><span class="o">.</span><span class="n">inputs</span> <span class="o">=</span> <span class="p">[]</span>
     <span class="bp">self</span><span class="o">.</span><span class="n">outputs</span> <span class="o">=</span> <span class="p">[]</span>
     <span class="bp">self</span><span class="o">.</span><span class="n">gpu_allocations</span> <span class="o">=</span> <span class="p">[]</span>
     <span class="bp">self</span><span class="o">.</span><span class="n">cpu_allocations</span> <span class="o">=</span> <span class="p">[]</span>

     <span class="c1"># TODO(https://github.com/NVIDIA/TensorRT/issues/2557):</span>
     <span class="c1"># Clean up when fixed upstream.</span>
     <span class="k">try</span><span class="p">:</span>
       <span class="n">_</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">bool</span>  <span class="c1"># type: ignore</span>
     <span class="k">except</span> <span class="ne">AttributeError</span><span class="p">:</span>
       <span class="c1"># numpy &gt;= 1.24.0</span>
       <span class="n">np</span><span class="o">.</span><span class="n">bool</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">bool_</span>  <span class="c1"># type: ignore</span>

     <span class="c1"># Setup I/O bindings.</span>
     <span class="k">for</span> <span class="n">i</span> <span class="ow">in</span> <span class="nb">range</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">engine</span><span class="o">.</span><span class="n">num_bindings</span><span class="p">):</span>
       <span class="n">name</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">engine</span><span class="o">.</span><span class="n">get_binding_name</span><span class="p">(</span><span class="n">i</span><span class="p">)</span>
       <span class="n">dtype</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">engine</span><span class="o">.</span><span class="n">get_binding_dtype</span><span class="p">(</span><span class="n">i</span><span class="p">)</span>
       <span class="n">shape</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">engine</span><span class="o">.</span><span class="n">get_binding_shape</span><span class="p">(</span><span class="n">i</span><span class="p">)</span>
       <span class="n">size</span> <span class="o">=</span> <span class="n">trt</span><span class="o">.</span><span class="n">volume</span><span class="p">(</span><span class="n">shape</span><span class="p">)</span> <span class="o">*</span> <span class="n">dtype</span><span class="o">.</span><span class="n">itemsize</span>
       <span class="n">allocation</span> <span class="o">=</span> <span class="n">_assign_or_fail</span><span class="p">(</span><span class="n">cuda</span><span class="o">.</span><span class="n">cuMemAlloc</span><span class="p">(</span><span class="n">size</span><span class="p">))</span>
       <span class="n">binding</span> <span class="o">=</span> <span class="p">{</span>
           <span class="s1">&#39;index&#39;</span><span class="p">:</span> <span class="n">i</span><span class="p">,</span>
           <span class="s1">&#39;name&#39;</span><span class="p">:</span> <span class="n">name</span><span class="p">,</span>
           <span class="s1">&#39;dtype&#39;</span><span class="p">:</span> <span class="n">np</span><span class="o">.</span><span class="n">dtype</span><span class="p">(</span><span class="n">trt</span><span class="o">.</span><span class="n">nptype</span><span class="p">(</span><span class="n">dtype</span><span class="p">)),</span>
           <span class="s1">&#39;shape&#39;</span><span class="p">:</span> <span class="nb">list</span><span class="p">(</span><span class="n">shape</span><span class="p">),</span>
           <span class="s1">&#39;allocation&#39;</span><span class="p">:</span> <span class="n">allocation</span><span class="p">,</span>
           <span class="s1">&#39;size&#39;</span><span class="p">:</span> <span class="n">size</span>
       <span class="p">}</span>
       <span class="bp">self</span><span class="o">.</span><span class="n">gpu_allocations</span><span class="o">.</span><span class="n">append</span><span class="p">(</span><span class="n">allocation</span><span class="p">)</span>
       <span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">engine</span><span class="o">.</span><span class="n">binding_is_input</span><span class="p">(</span><span class="n">i</span><span class="p">):</span>
         <span class="bp">self</span><span class="o">.</span><span class="n">inputs</span><span class="o">.</span><span class="n">append</span><span class="p">(</span><span class="n">binding</span><span class="p">)</span>
       <span class="k">else</span><span class="p">:</span>
         <span class="bp">self</span><span class="o">.</span><span class="n">outputs</span><span class="o">.</span><span class="n">append</span><span class="p">(</span><span class="n">binding</span><span class="p">)</span>

     <span class="k">assert</span> <span class="bp">self</span><span class="o">.</span><span class="n">context</span>
     <span class="k">assert</span> <span class="nb">len</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">inputs</span><span class="p">)</span> <span class="o">&gt;</span> <span class="mi">0</span>
     <span class="k">assert</span> <span class="nb">len</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">outputs</span><span class="p">)</span> <span class="o">&gt;</span> <span class="mi">0</span>
     <span class="k">assert</span> <span class="nb">len</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">gpu_allocations</span><span class="p">)</span> <span class="o">&gt;</span> <span class="mi">0</span>

     <span class="k">for</span> <span class="n">output</span> <span class="ow">in</span> <span class="bp">self</span><span class="o">.</span><span class="n">outputs</span><span class="p">:</span>
       <span class="bp">self</span><span class="o">.</span><span class="n">cpu_allocations</span><span class="o">.</span><span class="n">append</span><span class="p">(</span><span class="n">np</span><span class="o">.</span><span class="n">zeros</span><span class="p">(</span><span class="n">output</span><span class="p">[</span><span class="s1">&#39;shape&#39;</span><span class="p">],</span> <span class="n">output</span><span class="p">[</span><span class="s1">&#39;dtype&#39;</span><span class="p">]))</span>
     <span class="c1"># Create CUDA Stream.</span>
     <span class="bp">self</span><span class="o">.</span><span class="n">stream</span> <span class="o">=</span> <span class="n">_assign_or_fail</span><span class="p">(</span><span class="n">cuda</span><span class="o">.</span><span class="n">cuStreamCreate</span><span class="p">(</span><span class="mi">0</span><span class="p">))</span>

 <div class="viewcode-block" id="TensorRTEngine.get_engine_attrs"><a class="viewcode-back" href="../../../../apache_beam.ml.inference.tensorrt_inference.html#apache_beam.ml.inference.tensorrt_inference.TensorRTEngine.get_engine_attrs">[docs]</a>  <span class="k">def</span> <span class="nf">get_engine_attrs</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
 <span class="w">    </span><span class="sd">&quot;&quot;&quot;Returns TensorRT engine attributes.&quot;&quot;&quot;</span>
     <span class="k">return</span> <span class="p">(</span>
         <span class="bp">self</span><span class="o">.</span><span class="n">engine</span><span class="p">,</span>
         <span class="bp">self</span><span class="o">.</span><span class="n">context</span><span class="p">,</span>
         <span class="bp">self</span><span class="o">.</span><span class="n">context_lock</span><span class="p">,</span>
         <span class="bp">self</span><span class="o">.</span><span class="n">inputs</span><span class="p">,</span>
         <span class="bp">self</span><span class="o">.</span><span class="n">outputs</span><span class="p">,</span>
         <span class="bp">self</span><span class="o">.</span><span class="n">gpu_allocations</span><span class="p">,</span>
         <span class="bp">self</span><span class="o">.</span><span class="n">cpu_allocations</span><span class="p">,</span>
         <span class="bp">self</span><span class="o">.</span><span class="n">stream</span><span class="p">)</span></div></div>


 <span class="n">TensorRTInferenceFn</span> <span class="o">=</span> <span class="n">Callable</span><span class="p">[</span>
     <span class="p">[</span><span class="n">Sequence</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">],</span> <span class="n">TensorRTEngine</span><span class="p">,</span> <span class="n">Optional</span><span class="p">[</span><span class="n">Dict</span><span class="p">[</span><span class="nb">str</span><span class="p">,</span> <span class="n">Any</span><span class="p">]]],</span>
     <span class="n">Iterable</span><span class="p">[</span><span class="n">PredictionResult</span><span class="p">]]</span>


 <span class="k">def</span> <span class="nf">_default_tensorRT_inference_fn</span><span class="p">(</span>
     <span class="n">batch</span><span class="p">:</span> <span class="n">Sequence</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">],</span>
     <span class="n">engine</span><span class="p">:</span> <span class="n">TensorRTEngine</span><span class="p">,</span>
     <span class="n">inference_args</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">Dict</span><span class="p">[</span><span class="nb">str</span><span class="p">,</span>
                                   <span class="n">Any</span><span class="p">]]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="n">Iterable</span><span class="p">[</span><span class="n">PredictionResult</span><span class="p">]:</span>
   <span class="kn">from</span> <span class="nn">cuda</span> <span class="kn">import</span> <span class="n">cuda</span>
   <span class="p">(</span>
       <span class="n">engine</span><span class="p">,</span>
       <span class="n">context</span><span class="p">,</span>
       <span class="n">context_lock</span><span class="p">,</span>
       <span class="n">inputs</span><span class="p">,</span>
       <span class="n">outputs</span><span class="p">,</span>
       <span class="n">gpu_allocations</span><span class="p">,</span>
       <span class="n">cpu_allocations</span><span class="p">,</span>
       <span class="n">stream</span><span class="p">)</span> <span class="o">=</span> <span class="n">engine</span><span class="o">.</span><span class="n">get_engine_attrs</span><span class="p">()</span>

   <span class="c1"># Process I/O and execute the network</span>
   <span class="k">with</span> <span class="n">context_lock</span><span class="p">:</span>
     <span class="n">_assign_or_fail</span><span class="p">(</span>
         <span class="n">cuda</span><span class="o">.</span><span class="n">cuMemcpyHtoDAsync</span><span class="p">(</span>
             <span class="n">inputs</span><span class="p">[</span><span class="mi">0</span><span class="p">][</span><span class="s1">&#39;allocation&#39;</span><span class="p">],</span>
             <span class="n">np</span><span class="o">.</span><span class="n">ascontiguousarray</span><span class="p">(</span><span class="n">batch</span><span class="p">),</span>
             <span class="n">inputs</span><span class="p">[</span><span class="mi">0</span><span class="p">][</span><span class="s1">&#39;size&#39;</span><span class="p">],</span>
             <span class="n">stream</span><span class="p">))</span>
     <span class="n">context</span><span class="o">.</span><span class="n">execute_async_v2</span><span class="p">(</span><span class="n">gpu_allocations</span><span class="p">,</span> <span class="n">stream</span><span class="p">)</span>
     <span class="k">for</span> <span class="n">output</span> <span class="ow">in</span> <span class="nb">range</span><span class="p">(</span><span class="nb">len</span><span class="p">(</span><span class="n">cpu_allocations</span><span class="p">)):</span>
       <span class="n">_assign_or_fail</span><span class="p">(</span>
           <span class="n">cuda</span><span class="o">.</span><span class="n">cuMemcpyDtoHAsync</span><span class="p">(</span>
               <span class="n">cpu_allocations</span><span class="p">[</span><span class="n">output</span><span class="p">],</span>
               <span class="n">outputs</span><span class="p">[</span><span class="n">output</span><span class="p">][</span><span class="s1">&#39;allocation&#39;</span><span class="p">],</span>
               <span class="n">outputs</span><span class="p">[</span><span class="n">output</span><span class="p">][</span><span class="s1">&#39;size&#39;</span><span class="p">],</span>
               <span class="n">stream</span><span class="p">))</span>
     <span class="n">_assign_or_fail</span><span class="p">(</span><span class="n">cuda</span><span class="o">.</span><span class="n">cuStreamSynchronize</span><span class="p">(</span><span class="n">stream</span><span class="p">))</span>

     <span class="n">predictions</span> <span class="o">=</span> <span class="p">[]</span>
     <span class="k">for</span> <span class="n">idx</span> <span class="ow">in</span> <span class="nb">range</span><span class="p">(</span><span class="nb">len</span><span class="p">(</span><span class="n">batch</span><span class="p">)):</span>
       <span class="n">predictions</span><span class="o">.</span><span class="n">append</span><span class="p">([</span><span class="n">prediction</span><span class="p">[</span><span class="n">idx</span><span class="p">]</span> <span class="k">for</span> <span class="n">prediction</span> <span class="ow">in</span> <span class="n">cpu_allocations</span><span class="p">])</span>

     <span class="k">return</span> <span class="n">utils</span><span class="o">.</span><span class="n">_convert_to_result</span><span class="p">(</span><span class="n">batch</span><span class="p">,</span> <span class="n">predictions</span><span class="p">)</span>


 <div class="viewcode-block" id="TensorRTEngineHandlerNumPy"><a class="viewcode-back" href="../../../../apache_beam.ml.inference.tensorrt_inference.html#apache_beam.ml.inference.tensorrt_inference.TensorRTEngineHandlerNumPy">[docs]</a><span class="nd">@experimental</span><span class="p">(</span><span class="n">extra_message</span><span class="o">=</span><span class="s2">&quot;No backwards-compatibility guarantees.&quot;</span><span class="p">)</span>
 <span class="k">class</span> <span class="nc">TensorRTEngineHandlerNumPy</span><span class="p">(</span><span class="n">ModelHandler</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">,</span>
                                               <span class="n">PredictionResult</span><span class="p">,</span>
                                               <span class="n">TensorRTEngine</span><span class="p">]):</span>
   <span class="k">def</span> <span class="fm">__init__</span><span class="p">(</span>
       <span class="bp">self</span><span class="p">,</span>
       <span class="n">min_batch_size</span><span class="p">:</span> <span class="nb">int</span><span class="p">,</span>
       <span class="n">max_batch_size</span><span class="p">:</span> <span class="nb">int</span><span class="p">,</span>
       <span class="o">*</span><span class="p">,</span>
       <span class="n">inference_fn</span><span class="p">:</span> <span class="n">TensorRTInferenceFn</span> <span class="o">=</span> <span class="n">_default_tensorRT_inference_fn</span><span class="p">,</span>
       <span class="o">**</span><span class="n">kwargs</span><span class="p">):</span>
 <span class="w">    </span><span class="sd">&quot;&quot;&quot;Implementation of the ModelHandler interface for TensorRT.</span>

 <span class="sd">    Example Usage::</span>

 <span class="sd">      pcoll | RunInference(</span>
 <span class="sd">          TensorRTEngineHandlerNumPy(</span>
 <span class="sd">            min_batch_size=1,</span>
 <span class="sd">            max_batch_size=1,</span>
 <span class="sd">            engine_path=&quot;my_uri&quot;))</span>

 <span class="sd">    **NOTE:** This API and its implementation are under development and</span>
 <span class="sd">    do not provide backward compatibility guarantees.</span>

 <span class="sd">    Args:</span>
 <span class="sd">      min_batch_size: minimum accepted batch size.</span>
 <span class="sd">      max_batch_size: maximum accepted batch size.</span>
 <span class="sd">      inference_fn: the inference function to use on RunInference calls.</span>
 <span class="sd">        default: _default_tensorRT_inference_fn</span>
 <span class="sd">      kwargs: Additional arguments like &#39;engine_path&#39; and &#39;onnx_path&#39; are</span>
 <span class="sd">        currently supported.</span>

 <span class="sd">    See https://docs.nvidia.com/deeplearning/tensorrt/api/python_api/</span>
 <span class="sd">    for details</span>
 <span class="sd">    &quot;&quot;&quot;</span>
     <span class="bp">self</span><span class="o">.</span><span class="n">min_batch_size</span> <span class="o">=</span> <span class="n">min_batch_size</span>
     <span class="bp">self</span><span class="o">.</span><span class="n">max_batch_size</span> <span class="o">=</span> <span class="n">max_batch_size</span>
     <span class="bp">self</span><span class="o">.</span><span class="n">inference_fn</span> <span class="o">=</span> <span class="n">inference_fn</span>
     <span class="k">if</span> <span class="s1">&#39;engine_path&#39;</span> <span class="ow">in</span> <span class="n">kwargs</span><span class="p">:</span>
       <span class="bp">self</span><span class="o">.</span><span class="n">engine_path</span> <span class="o">=</span> <span class="n">kwargs</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;engine_path&#39;</span><span class="p">)</span>
     <span class="k">elif</span> <span class="s1">&#39;onnx_path&#39;</span> <span class="ow">in</span> <span class="n">kwargs</span><span class="p">:</span>
       <span class="bp">self</span><span class="o">.</span><span class="n">onnx_path</span> <span class="o">=</span> <span class="n">kwargs</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;onnx_path&#39;</span><span class="p">)</span>

 <div class="viewcode-block" id="TensorRTEngineHandlerNumPy.batch_elements_kwargs"><a class="viewcode-back" href="../../../../apache_beam.ml.inference.tensorrt_inference.html#apache_beam.ml.inference.tensorrt_inference.TensorRTEngineHandlerNumPy.batch_elements_kwargs">[docs]</a>  <span class="k">def</span> <span class="nf">batch_elements_kwargs</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
 <span class="w">    </span><span class="sd">&quot;&quot;&quot;Sets min_batch_size and max_batch_size of a TensorRT engine.&quot;&quot;&quot;</span>
     <span class="k">return</span> <span class="p">{</span>
         <span class="s1">&#39;min_batch_size&#39;</span><span class="p">:</span> <span class="bp">self</span><span class="o">.</span><span class="n">min_batch_size</span><span class="p">,</span>
         <span class="s1">&#39;max_batch_size&#39;</span><span class="p">:</span> <span class="bp">self</span><span class="o">.</span><span class="n">max_batch_size</span>
     <span class="p">}</span></div>

 <div class="viewcode-block" id="TensorRTEngineHandlerNumPy.load_model"><a class="viewcode-back" href="../../../../apache_beam.ml.inference.tensorrt_inference.html#apache_beam.ml.inference.tensorrt_inference.TensorRTEngineHandlerNumPy.load_model">[docs]</a>  <span class="k">def</span> <span class="nf">load_model</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="n">TensorRTEngine</span><span class="p">:</span>
 <span class="w">    </span><span class="sd">&quot;&quot;&quot;Loads and initializes a TensorRT engine for processing.&quot;&quot;&quot;</span>
     <span class="n">engine</span> <span class="o">=</span> <span class="n">_load_engine</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">engine_path</span><span class="p">)</span>
     <span class="k">return</span> <span class="n">TensorRTEngine</span><span class="p">(</span><span class="n">engine</span><span class="p">)</span></div>

 <div class="viewcode-block" id="TensorRTEngineHandlerNumPy.load_onnx"><a class="viewcode-back" href="../../../../apache_beam.ml.inference.tensorrt_inference.html#apache_beam.ml.inference.tensorrt_inference.TensorRTEngineHandlerNumPy.load_onnx">[docs]</a>  <span class="k">def</span> <span class="nf">load_onnx</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="n">Tuple</span><span class="p">[</span><span class="n">trt</span><span class="o">.</span><span class="n">INetworkDefinition</span><span class="p">,</span> <span class="n">trt</span><span class="o">.</span><span class="n">Builder</span><span class="p">]:</span>
 <span class="w">    </span><span class="sd">&quot;&quot;&quot;Loads and parses an onnx model for processing.&quot;&quot;&quot;</span>
     <span class="k">return</span> <span class="n">_load_onnx</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">onnx_path</span><span class="p">)</span></div>

 <div class="viewcode-block" id="TensorRTEngineHandlerNumPy.build_engine"><a class="viewcode-back" href="../../../../apache_beam.ml.inference.tensorrt_inference.html#apache_beam.ml.inference.tensorrt_inference.TensorRTEngineHandlerNumPy.build_engine">[docs]</a>  <span class="k">def</span> <span class="nf">build_engine</span><span class="p">(</span>
       <span class="bp">self</span><span class="p">,</span> <span class="n">network</span><span class="p">:</span> <span class="n">trt</span><span class="o">.</span><span class="n">INetworkDefinition</span><span class="p">,</span>
       <span class="n">builder</span><span class="p">:</span> <span class="n">trt</span><span class="o">.</span><span class="n">Builder</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="n">TensorRTEngine</span><span class="p">:</span>
 <span class="w">    </span><span class="sd">&quot;&quot;&quot;Build an engine according to parsed/created network.&quot;&quot;&quot;</span>
     <span class="n">engine</span> <span class="o">=</span> <span class="n">_build_engine</span><span class="p">(</span><span class="n">network</span><span class="p">,</span> <span class="n">builder</span><span class="p">)</span>
     <span class="k">return</span> <span class="n">TensorRTEngine</span><span class="p">(</span><span class="n">engine</span><span class="p">)</span></div>

 <div class="viewcode-block" id="TensorRTEngineHandlerNumPy.run_inference"><a class="viewcode-back" href="../../../../apache_beam.ml.inference.tensorrt_inference.html#apache_beam.ml.inference.tensorrt_inference.TensorRTEngineHandlerNumPy.run_inference">[docs]</a>  <span class="k">def</span> <span class="nf">run_inference</span><span class="p">(</span>
       <span class="bp">self</span><span class="p">,</span>
       <span class="n">batch</span><span class="p">:</span> <span class="n">Sequence</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">],</span>
       <span class="n">engine</span><span class="p">:</span> <span class="n">TensorRTEngine</span><span class="p">,</span>
       <span class="n">inference_args</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">Dict</span><span class="p">[</span><span class="nb">str</span><span class="p">,</span> <span class="n">Any</span><span class="p">]]</span> <span class="o">=</span> <span class="kc">None</span>
   <span class="p">)</span> <span class="o">-&gt;</span> <span class="n">Iterable</span><span class="p">[</span><span class="n">PredictionResult</span><span class="p">]:</span>
 <span class="w">    </span><span class="sd">&quot;&quot;&quot;</span>
 <span class="sd">    Runs inferences on a batch of Tensors and returns an Iterable of</span>
 <span class="sd">    TensorRT Predictions.</span>

 <span class="sd">    Args:</span>
 <span class="sd">      batch: A np.ndarray or a np.ndarray that represents a concatenation</span>
 <span class="sd">        of multiple arrays as a batch.</span>
 <span class="sd">      engine: A TensorRT engine.</span>
 <span class="sd">      inference_args: Any additional arguments for an inference</span>
 <span class="sd">        that are not applicable to TensorRT.</span>

 <span class="sd">    Returns:</span>
 <span class="sd">      An Iterable of type PredictionResult.</span>
 <span class="sd">    &quot;&quot;&quot;</span>
     <span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">inference_fn</span><span class="p">(</span><span class="n">batch</span><span class="p">,</span> <span class="n">engine</span><span class="p">,</span> <span class="n">inference_args</span><span class="p">)</span></div>

 <div class="viewcode-block" id="TensorRTEngineHandlerNumPy.get_num_bytes"><a class="viewcode-back" href="../../../../apache_beam.ml.inference.tensorrt_inference.html#apache_beam.ml.inference.tensorrt_inference.TensorRTEngineHandlerNumPy.get_num_bytes">[docs]</a>  <span class="k">def</span> <span class="nf">get_num_bytes</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">batch</span><span class="p">:</span> <span class="n">Sequence</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">])</span> <span class="o">-&gt;</span> <span class="nb">int</span><span class="p">:</span>
 <span class="w">    </span><span class="sd">&quot;&quot;&quot;</span>
 <span class="sd">    Returns:</span>
 <span class="sd">      The number of bytes of data for a batch of Tensors.</span>
 <span class="sd">    &quot;&quot;&quot;</span>
     <span class="k">return</span> <span class="nb">sum</span><span class="p">((</span><span class="n">np_array</span><span class="o">.</span><span class="n">itemsize</span> <span class="k">for</span> <span class="n">np_array</span> <span class="ow">in</span> <span class="n">batch</span><span class="p">))</span></div>

 <div class="viewcode-block" id="TensorRTEngineHandlerNumPy.get_metrics_namespace"><a class="viewcode-back" href="../../../../apache_beam.ml.inference.tensorrt_inference.html#apache_beam.ml.inference.tensorrt_inference.TensorRTEngineHandlerNumPy.get_metrics_namespace">[docs]</a>  <span class="k">def</span> <span class="nf">get_metrics_namespace</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">str</span><span class="p">:</span>
 <span class="w">    </span><span class="sd">&quot;&quot;&quot;</span>
 <span class="sd">    Returns a namespace for metrics collected by the RunInference transform.</span>
 <span class="sd">    &quot;&quot;&quot;</span>
     <span class="k">return</span> <span class="s1">&#39;BeamML_TensorRT&#39;</span></div></div>
 </pre></div>

            </div>

           </div>
           <footer>


   <hr/>

   <div role="contentinfo">
     <p>
         &copy; Copyright

     </p>
   </div>
   Built with <a href="http://sphinx-doc.org/">Sphinx</a> using a <a href="https://github.com/rtfd/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">
       jQuery(function () {
           SphinxRtdTheme.Navigation.enable(true);
       });
   </script>


 </body>
 </html>