arm_compute v17.09
Change-Id: I4bf8f4e6e5f84ce0d5b6f5ba570d276879f42a81
diff --git a/documentation/convolution__layer_8cl.xhtml b/documentation/convolution__layer_8cl.xhtml
index e083207..889c6ea 100644
--- a/documentation/convolution__layer_8cl.xhtml
+++ b/documentation/convolution__layer_8cl.xhtml
@@ -4,7 +4,7 @@
<head>
<meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/>
<meta http-equiv="X-UA-Compatible" content="IE=9"/>
-<meta name="generator" content="Doxygen 1.8.11"/>
+<meta name="generator" content="Doxygen 1.8.6"/>
<meta name="robots" content="NOINDEX, NOFOLLOW" /> <!-- Prevent indexing by search engines -->
<title>Compute Library: src/core/CL/cl_kernels/convolution_layer.cl File Reference</title>
<link href="tabs.css" rel="stylesheet" type="text/css"/>
@@ -12,24 +12,22 @@
<script type="text/javascript" src="dynsections.js"></script>
<link href="navtree.css" rel="stylesheet" type="text/css"/>
<script type="text/javascript" src="resize.js"></script>
-<script type="text/javascript" src="navtreedata.js"></script>
<script type="text/javascript" src="navtree.js"></script>
<script type="text/javascript">
$(document).ready(initResizable);
$(window).load(resizeHeight);
</script>
<link href="search/search.css" rel="stylesheet" type="text/css"/>
-<script type="text/javascript" src="search/searchdata.js"></script>
<script type="text/javascript" src="search/search.js"></script>
<script type="text/javascript">
- $(document).ready(function() { init_search(); });
+ $(document).ready(function() { searchBox.OnSelectItem(0); });
</script>
<script type="text/x-mathjax-config">
MathJax.Hub.Config({
extensions: ["tex2jax.js"],
jax: ["input/TeX","output/HTML-CSS"],
});
-</script><script type="text/javascript" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js"></script>
+</script><script src="http://cdn.mathjax.org/mathjax/latest/MathJax.js"></script>
<link href="doxygen.css" rel="stylesheet" type="text/css" />
</head>
<body>
@@ -40,7 +38,7 @@
<tr style="height: 56px;">
<td style="padding-left: 0.5em;">
<div id="projectname">Compute Library
-  <span id="projectnumber">v17.06</span>
+  <span id="projectnumber">17.09</span>
</div>
</td>
</tr>
@@ -48,7 +46,7 @@
</table>
</div>
<!-- end header part -->
-<!-- Generated by Doxygen 1.8.11 -->
+<!-- Generated by Doxygen 1.8.6 -->
<script type="text/javascript">
var searchBox = new SearchBox("searchBox", "search",false,'Search');
</script>
@@ -103,7 +101,7 @@
onmouseover="return searchBox.OnSearchSelectShow()"
onmouseout="return searchBox.OnSearchSelectHide()"
onkeydown="return searchBox.OnSearchSelectKey(event)">
-</div>
+<a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(0)"><span class="SelectionMark"> </span>All</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(1)"><span class="SelectionMark"> </span>Data Structures</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(2)"><span class="SelectionMark"> </span>Namespaces</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(3)"><span class="SelectionMark"> </span>Files</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(4)"><span class="SelectionMark"> </span>Functions</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(5)"><span class="SelectionMark"> </span>Variables</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(6)"><span class="SelectionMark"> </span>Typedefs</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(7)"><span class="SelectionMark"> </span>Enumerations</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(8)"><span class="SelectionMark"> </span>Enumerator</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(9)"><span class="SelectionMark"> </span>Friends</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(10)"><span class="SelectionMark"> </span>Macros</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(11)"><span class="SelectionMark"> </span>Pages</a></div>
<!-- iframe showing the search results (closed by default) -->
<div id="MSearchResultsWindow">
@@ -119,344 +117,20 @@
<div class="title">convolution_layer.cl File Reference</div> </div>
</div><!--header-->
<div class="contents">
-<div class="textblock"><code>#include "<a class="el" href="helpers_8h_source.xhtml">helpers.h</a>"</code><br />
+<div class="textblock"><code>#include "<a class="el" href="helpers_8h_source.xhtml">helpers.h</a>"</code><br/>
</div>
<p><a href="convolution__layer_8cl_source.xhtml">Go to the source code of this file.</a></p>
<table class="memberdecls">
<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="func-members"></a>
Functions</h2></td></tr>
<tr class="memitem:ab6358b28d1e7bc946950075d94412013"><td class="memItemLeft" align="right" valign="top">__kernel void </td><td class="memItemRight" valign="bottom"><a class="el" href="convolution__layer_8cl.xhtml#ab6358b28d1e7bc946950075d94412013">reshape_to_columns</a> (__global uchar *src_ptr, uint src_stride_x, uint src_step_x, uint src_stride_y, uint src_step_y, uint src_stride_z, uint src_step_z, uint src_offset_first_element_in_bytes, __global uchar *dst_ptr, uint dst_stride_x, uint dst_step_x, uint dst_stride_y, uint dst_step_y, uint dst_offset_first_element_in_bytes, __global uchar *bias_ptr, uint bias_stride_x, uint bias_step_x, uint bias_offset_first_element_in_bytes, uint width, uint height, uint depth, uint total_filters)</td></tr>
-<tr class="memdesc:ab6358b28d1e7bc946950075d94412013"><td class="mdescLeft"> </td><td class="mdescRight">This kernel reshapes the tensor's low three dimensions to single column. <a href="#ab6358b28d1e7bc946950075d94412013">More...</a><br /></td></tr>
+<tr class="memdesc:ab6358b28d1e7bc946950075d94412013"><td class="mdescLeft"> </td><td class="mdescRight">This kernel reshapes the tensor's low three dimensions to single column. <a href="#ab6358b28d1e7bc946950075d94412013">More...</a><br/></td></tr>
<tr class="separator:ab6358b28d1e7bc946950075d94412013"><td class="memSeparator" colspan="2"> </td></tr>
-<tr class="memitem:afaddfccc926506181141b2dc86cb73f1"><td class="memItemLeft" align="right" valign="top">__kernel void </td><td class="memItemRight" valign="bottom"><a class="el" href="convolution__layer_8cl.xhtml#afaddfccc926506181141b2dc86cb73f1">im2col_generic</a> (__global uchar *src_ptr, uint src_stride_x, uint src_step_x, uint src_stride_y, uint src_step_y, uint src_stride_z, uint src_step_z, uint src_offset_first_element_in_bytes, __global uchar *dst_ptr, uint dst_stride_x, uint dst_step_x, uint dst_stride_y, uint dst_step_y, uint dst_offset_first_element_in_bytes, int kernel_size, int kernel_depth, int width, int2 input_dims, int2 strides, int2 paddings)</td></tr>
-<tr class="memdesc:afaddfccc926506181141b2dc86cb73f1"><td class="mdescLeft"> </td><td class="mdescRight">This kernel performs a reshaping of the input tensor to a tensor used to perform convolution using GEMM. <a href="#afaddfccc926506181141b2dc86cb73f1">More...</a><br /></td></tr>
-<tr class="separator:afaddfccc926506181141b2dc86cb73f1"><td class="memSeparator" colspan="2"> </td></tr>
-<tr class="memitem:af11bad06b2f7c26cc6ee2f93add5dcbf"><td class="memItemLeft" align="right" valign="top">__kernel void </td><td class="memItemRight" valign="bottom"><a class="el" href="convolution__layer_8cl.xhtml#af11bad06b2f7c26cc6ee2f93add5dcbf">col2im</a> (__global uchar *src_ptr, uint src_stride_x, uint src_step_x, uint src_stride_y, uint src_step_y, uint src_offset_first_element_in_bytes, __global uchar *dst_ptr, uint dst_stride_x, uint dst_step_x, uint dst_stride_y, uint dst_step_y, uint dst_stride_z, uint dst_step_z, uint dst_offset_first_element_in_bytes, uint width)</td></tr>
-<tr class="memdesc:af11bad06b2f7c26cc6ee2f93add5dcbf"><td class="mdescLeft"> </td><td class="mdescRight">This kernel performs a reshaping of the output of the convolution layer. <a href="#af11bad06b2f7c26cc6ee2f93add5dcbf">More...</a><br /></td></tr>
-<tr class="separator:af11bad06b2f7c26cc6ee2f93add5dcbf"><td class="memSeparator" colspan="2"> </td></tr>
<tr class="memitem:a6f31f2c8d606e1b57d9bdfacd416024c"><td class="memItemLeft" align="right" valign="top">__kernel void </td><td class="memItemRight" valign="bottom"><a class="el" href="convolution__layer_8cl.xhtml#a6f31f2c8d606e1b57d9bdfacd416024c">im2col_reduced</a> (__global uchar *src_ptr, uint src_stride_x, uint src_step_x, uint src_stride_y, uint src_step_y, uint src_stride_z, uint src_step_z, uint src_offset_first_element_in_bytes, __global uchar *dst_ptr, uint dst_stride_x, uint dst_step_x, uint dst_offset_first_element_in_bytes, uint width, uint height)</td></tr>
-<tr class="memdesc:a6f31f2c8d606e1b57d9bdfacd416024c"><td class="mdescLeft"> </td><td class="mdescRight">This kernel reshapes the tensor's low three dimensions to single row for GEMM operation. <a href="#a6f31f2c8d606e1b57d9bdfacd416024c">More...</a><br /></td></tr>
+<tr class="memdesc:a6f31f2c8d606e1b57d9bdfacd416024c"><td class="mdescLeft"> </td><td class="mdescRight">This kernel reshapes the tensor's low three dimensions to single row for GEMM operation. <a href="#a6f31f2c8d606e1b57d9bdfacd416024c">More...</a><br/></td></tr>
<tr class="separator:a6f31f2c8d606e1b57d9bdfacd416024c"><td class="memSeparator" colspan="2"> </td></tr>
</table>
<h2 class="groupheader">Function Documentation</h2>
-<a class="anchor" id="af11bad06b2f7c26cc6ee2f93add5dcbf"></a>
-<div class="memitem">
-<div class="memproto">
- <table class="memname">
- <tr>
- <td class="memname">__kernel void col2im </td>
- <td>(</td>
- <td class="paramtype">__global uchar * </td>
- <td class="paramname"><em>src_ptr</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>src_stride_x</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>src_step_x</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>src_stride_y</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>src_step_y</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>src_offset_first_element_in_bytes</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">__global uchar * </td>
- <td class="paramname"><em>dst_ptr</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>dst_stride_x</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>dst_step_x</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>dst_stride_y</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>dst_step_y</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>dst_stride_z</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>dst_step_z</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>dst_offset_first_element_in_bytes</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>width</em> </td>
- </tr>
- <tr>
- <td></td>
- <td>)</td>
- <td></td><td></td>
- </tr>
- </table>
-</div><div class="memdoc">
-
-<p>This kernel performs a reshaping of the output of the convolution layer. </p>
-<dl class="section note"><dt>Note</dt><dd>The data type must be passed at compile time using -DDATA_TYPE: e.g. -DDATA_TYPE=float</dd></dl>
-<dl class="params"><dt>Parameters</dt><dd>
- <table class="params">
- <tr><td class="paramdir">[in]</td><td class="paramname">src_ptr</td><td>Pointer to the source tensor. Supported data types: F16, F32 </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_stride_x</td><td>Stride of the source tensor in X dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_step_x</td><td>src_stride_x * number of elements along X processed per workitem(in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_stride_y</td><td>Stride of the source tensor in Y dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_step_y</td><td>src_stride_y * number of elements along Y processed per workitem(in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_offset_first_element_in_bytes</td><td>The offset of the first element in the source tensor </td></tr>
- <tr><td class="paramdir">[out]</td><td class="paramname">dst_ptr</td><td>Pointer to the destination tensor. Supported data types: F16, F32 </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_stride_x</td><td>Stride of the destination tensor in X dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_x</td><td>dst_stride_x * number of elements along X processed per workitem(in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_stride_y</td><td>Stride of the destination tensor in Y dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_y</td><td>dst_stride_y * number of elements along Y processed per workitem(in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_stride_z</td><td>Stride of the destination tensor in Z dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_z</td><td>dst_stride_z * number of elements along Z processed per workitem(in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_offset_first_element_in_bytes</td><td>The offset of the first element in the destination tensor </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">width</td><td>The output tensor width </td></tr>
- </table>
- </dd>
-</dl>
-
-<p>Definition at line <a class="el" href="convolution__layer_8cl_source.xhtml#l00187">187</a> of file <a class="el" href="convolution__layer_8cl_source.xhtml">convolution_layer.cl</a>.</p>
-
-<p>References <a class="el" href="helpers_8h_source.xhtml#l00073">CONVERT_TO_IMAGE_STRUCT</a>, <a class="el" href="helpers_8h_source.xhtml#l00083">CONVERT_TO_TENSOR3D_STRUCT_NO_STEP</a>, <a class="el" href="convolution3x3_8cl_source.xhtml#l00027">DATA_TYPE</a>, <a class="el" href="helpers_8h_source.xhtml#l00097">Image::ptr</a>, <a class="el" href="helpers_8h_source.xhtml#l00106">Tensor3D::ptr</a>, <a class="el" href="helpers_8h_source.xhtml#l00108">Tensor3D::stride_x</a>, <a class="el" href="helpers_8h_source.xhtml#l00109">Tensor3D::stride_y</a>, and <a class="el" href="helpers_8h_source.xhtml#l00110">Tensor3D::stride_z</a>.</p>
-<div class="fragment"><div class="line"><a name="l00191"></a><span class="lineno"> 191</span> {</div><div class="line"><a name="l00192"></a><span class="lineno"> 192</span>  <a class="code" href="struct_image.xhtml">Image</a> src = <a class="code" href="helpers_8h.xhtml#aebe814363556c244be043b13e7969197">CONVERT_TO_IMAGE_STRUCT</a>(src);</div><div class="line"><a name="l00193"></a><span class="lineno"> 193</span>  <a class="code" href="struct_tensor3_d.xhtml">Tensor3D</a> dst = <a class="code" href="helpers_8h.xhtml#a79e8e562daa6599317d2d1cd86ef1bf2">CONVERT_TO_TENSOR3D_STRUCT_NO_STEP</a>(dst);</div><div class="line"><a name="l00194"></a><span class="lineno"> 194</span> </div><div class="line"><a name="l00195"></a><span class="lineno"> 195</span>  <span class="keywordtype">int</span> idx = get_global_id(0) * dst.<a class="code" href="struct_tensor3_d.xhtml#ad5ff7a2b2bd0eec50fe09c254b127d1c">stride_z</a> + (get_global_id(1) / width) * dst.<a class="code" href="struct_tensor3_d.xhtml#a4f0b90c9ecd6e57ceb3f37332fefe8f1">stride_y</a> + (get_global_id(1) % width) * dst.<a class="code" href="struct_tensor3_d.xhtml#ae01febbfd0689ef709f3ff6fdd2abc7e">stride_x</a>;</div><div class="line"><a name="l00196"></a><span class="lineno"> 196</span>  __global uchar *tmp_out_ptr = dst.<a class="code" href="struct_tensor3_d.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a> + idx;</div><div class="line"><a name="l00197"></a><span class="lineno"> 197</span>  *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)tmp_out_ptr) = *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)(src.<a class="code" href="struct_image.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a>));</div><div class="line"><a name="l00198"></a><span class="lineno"> 198</span> }</div><div class="ttc" id="convolution3x3_8cl_xhtml_afb8c72ce35c4a1f4a2588d6573e54aa1"><div class="ttname"><a href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a></div><div class="ttdeci">#define DATA_TYPE</div><div class="ttdef"><b>Definition:</b> <a href="convolution3x3_8cl_source.xhtml#l00027">convolution3x3.cl:27</a></div></div>
-<div class="ttc" id="struct_tensor3_d_xhtml_ad5ff7a2b2bd0eec50fe09c254b127d1c"><div class="ttname"><a href="struct_tensor3_d.xhtml#ad5ff7a2b2bd0eec50fe09c254b127d1c">Tensor3D::stride_z</a></div><div class="ttdeci">int stride_z</div><div class="ttdoc">Stride of the image in Z dimension (in bytes) </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00110">helpers.h:110</a></div></div>
-<div class="ttc" id="struct_tensor3_d_xhtml_ae01febbfd0689ef709f3ff6fdd2abc7e"><div class="ttname"><a href="struct_tensor3_d.xhtml#ae01febbfd0689ef709f3ff6fdd2abc7e">Tensor3D::stride_x</a></div><div class="ttdeci">int stride_x</div><div class="ttdoc">Stride of the image in X dimension (in bytes) </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00108">helpers.h:108</a></div></div>
-<div class="ttc" id="struct_tensor3_d_xhtml"><div class="ttname"><a href="struct_tensor3_d.xhtml">Tensor3D</a></div><div class="ttdoc">Structure to hold 3D tensor information. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00104">helpers.h:104</a></div></div>
-<div class="ttc" id="helpers_8h_xhtml_aebe814363556c244be043b13e7969197"><div class="ttname"><a href="helpers_8h.xhtml#aebe814363556c244be043b13e7969197">CONVERT_TO_IMAGE_STRUCT</a></div><div class="ttdeci">#define CONVERT_TO_IMAGE_STRUCT(name)</div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00073">helpers.h:73</a></div></div>
-<div class="ttc" id="struct_image_xhtml"><div class="ttname"><a href="struct_image.xhtml">Image</a></div><div class="ttdoc">Structure to hold Image information. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00095">helpers.h:95</a></div></div>
-<div class="ttc" id="struct_image_xhtml_acf52c23cbd7424606c10a606524e3e32"><div class="ttname"><a href="struct_image.xhtml#acf52c23cbd7424606c10a606524e3e32">Image::ptr</a></div><div class="ttdeci">__global uchar * ptr</div><div class="ttdoc">Pointer to the starting postion of the buffer. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00097">helpers.h:97</a></div></div>
-<div class="ttc" id="helpers_8h_xhtml_a79e8e562daa6599317d2d1cd86ef1bf2"><div class="ttname"><a href="helpers_8h.xhtml#a79e8e562daa6599317d2d1cd86ef1bf2">CONVERT_TO_TENSOR3D_STRUCT_NO_STEP</a></div><div class="ttdeci">#define CONVERT_TO_TENSOR3D_STRUCT_NO_STEP(name)</div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00083">helpers.h:83</a></div></div>
-<div class="ttc" id="struct_tensor3_d_xhtml_acf52c23cbd7424606c10a606524e3e32"><div class="ttname"><a href="struct_tensor3_d.xhtml#acf52c23cbd7424606c10a606524e3e32">Tensor3D::ptr</a></div><div class="ttdeci">__global uchar * ptr</div><div class="ttdoc">Pointer to the starting postion of the buffer. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00106">helpers.h:106</a></div></div>
-<div class="ttc" id="struct_tensor3_d_xhtml_a4f0b90c9ecd6e57ceb3f37332fefe8f1"><div class="ttname"><a href="struct_tensor3_d.xhtml#a4f0b90c9ecd6e57ceb3f37332fefe8f1">Tensor3D::stride_y</a></div><div class="ttdeci">int stride_y</div><div class="ttdoc">Stride of the image in Y dimension (in bytes) </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00109">helpers.h:109</a></div></div>
-</div><!-- fragment -->
-</div>
-</div>
-<a class="anchor" id="afaddfccc926506181141b2dc86cb73f1"></a>
-<div class="memitem">
-<div class="memproto">
- <table class="memname">
- <tr>
- <td class="memname">__kernel void im2col_generic </td>
- <td>(</td>
- <td class="paramtype">__global uchar * </td>
- <td class="paramname"><em>src_ptr</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>src_stride_x</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>src_step_x</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>src_stride_y</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>src_step_y</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>src_stride_z</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>src_step_z</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>src_offset_first_element_in_bytes</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">__global uchar * </td>
- <td class="paramname"><em>dst_ptr</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>dst_stride_x</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>dst_step_x</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>dst_stride_y</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>dst_step_y</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">uint </td>
- <td class="paramname"><em>dst_offset_first_element_in_bytes</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">int </td>
- <td class="paramname"><em>kernel_size</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">int </td>
- <td class="paramname"><em>kernel_depth</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">int </td>
- <td class="paramname"><em>width</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">int2 </td>
- <td class="paramname"><em>input_dims</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">int2 </td>
- <td class="paramname"><em>strides</em>, </td>
- </tr>
- <tr>
- <td class="paramkey"></td>
- <td></td>
- <td class="paramtype">int2 </td>
- <td class="paramname"><em>paddings</em> </td>
- </tr>
- <tr>
- <td></td>
- <td>)</td>
- <td></td><td></td>
- </tr>
- </table>
-</div><div class="memdoc">
-
-<p>This kernel performs a reshaping of the input tensor to a tensor used to perform convolution using GEMM. </p>
-<dl class="section note"><dt>Note</dt><dd>The data type must be passed at compile time using -DDATA_TYPE: e.g. -DDATA_TYPE=float </dd>
-<dd>
-In case biases will be added to the convolution -DHAS_BIAS has to be passed to append the final matrix with 1 in each row.</dd></dl>
-<dl class="params"><dt>Parameters</dt><dd>
- <table class="params">
- <tr><td class="paramdir">[in]</td><td class="paramname">src_ptr</td><td>Pointer to the source tensor. Supported data types: F16, F32 </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_stride_x</td><td>Stride of the source tensor in X dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_step_x</td><td>src_stride_x * number of elements along X processed per workitem(in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_stride_y</td><td>Stride of the source tensor in Y dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_step_y</td><td>src_stride_y * number of elements along Y processed per workitem(in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_stride_z</td><td>Stride of the source tensor in Z dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_step_z</td><td>src_stride_z * number of elements along Z processed per workitem(in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_offset_first_element_in_bytes</td><td>The offset of the first element in the source tensor </td></tr>
- <tr><td class="paramdir">[out]</td><td class="paramname">dst_ptr</td><td>Pointer to the destination tensor. Supported data types: F16, F32 </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_stride_x</td><td>Stride of the destination tensor in X dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_x</td><td>dst_stride_x * number of elements along X processed per workitem(in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_stride_y</td><td>Stride of the destination tensor in Y dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_y</td><td>dst_stride_y * number of elements along Y processed per workitem(in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_offset_first_element_in_bytes</td><td>The offset of the first element in the destination tensor </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">kernel_size</td><td>The convolution kernel size </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">kernel_depth</td><td>The kernel depth </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">width</td><td>The output tensor width </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">input_dims</td><td>The input tensor dimensions </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">strides</td><td>The strides of the im2col operation </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">paddings</td><td>The input tensor paddings </td></tr>
- </table>
- </dd>
-</dl>
-
-<p>Definition at line <a class="el" href="convolution__layer_8cl_source.xhtml#l00122">122</a> of file <a class="el" href="convolution__layer_8cl_source.xhtml">convolution_layer.cl</a>.</p>
-
-<p>References <a class="el" href="helpers_8h_source.xhtml#l00076">CONVERT_TO_IMAGE_STRUCT_NO_STEP</a>, <a class="el" href="helpers_8h_source.xhtml#l00079">CONVERT_TO_TENSOR3D_STRUCT</a>, <a class="el" href="convolution3x3_8cl_source.xhtml#l00027">DATA_TYPE</a>, <a class="el" href="helpers_8h_source.xhtml#l00097">Image::ptr</a>, <a class="el" href="helpers_8h_source.xhtml#l00099">Image::stride_x</a>, <a class="el" href="helpers_8h_source.xhtml#l00100">Image::stride_y</a>, and <a class="el" href="helpers_8h_source.xhtml#l00213">tensor3D_offset()</a>.</p>
-<div class="fragment"><div class="line"><a name="l00131"></a><span class="lineno"> 131</span> {</div><div class="line"><a name="l00132"></a><span class="lineno"> 132</span>  <a class="code" href="struct_tensor3_d.xhtml">Tensor3D</a> src = <a class="code" href="helpers_8h.xhtml#a31c8c760f08fb1a331b16b7c204321dc">CONVERT_TO_TENSOR3D_STRUCT</a>(src);</div><div class="line"><a name="l00133"></a><span class="lineno"> 133</span>  <a class="code" href="struct_image.xhtml">Image</a> dst = <a class="code" href="helpers_8h.xhtml#a4334a4a76f8e9628c0fb9e1acf616e2a">CONVERT_TO_IMAGE_STRUCT_NO_STEP</a>(dst);</div><div class="line"><a name="l00134"></a><span class="lineno"> 134</span> </div><div class="line"><a name="l00135"></a><span class="lineno"> 135</span>  <span class="comment">// Determine output index</span></div><div class="line"><a name="l00136"></a><span class="lineno"> 136</span>  uint idx = (get_global_id(1) * width + get_global_id(0)) * dst.<a class="code" href="struct_image.xhtml#a4f0b90c9ecd6e57ceb3f37332fefe8f1">stride_y</a>;</div><div class="line"><a name="l00137"></a><span class="lineno"> 137</span>  __global uchar *output_ptr = dst.<a class="code" href="struct_image.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a> + idx;</div><div class="line"><a name="l00138"></a><span class="lineno"> 138</span> </div><div class="line"><a name="l00139"></a><span class="lineno"> 139</span>  <span class="comment">// Determine current input index</span></div><div class="line"><a name="l00140"></a><span class="lineno"> 140</span>  <span class="keyword">const</span> <span class="keywordtype">int</span> top_left_x = get_global_id(0) * strides.x - paddings.x;</div><div class="line"><a name="l00141"></a><span class="lineno"> 141</span>  <span class="keyword">const</span> <span class="keywordtype">int</span> top_left_y = get_global_id(1) * strides.y - paddings.y;</div><div class="line"><a name="l00142"></a><span class="lineno"> 142</span> </div><div class="line"><a name="l00143"></a><span class="lineno"> 143</span>  <span class="comment">// Linearize convolution elements</span></div><div class="line"><a name="l00144"></a><span class="lineno"> 144</span>  <span class="keywordflow">for</span>(<span class="keywordtype">int</span> d = 0; d < kernel_depth; ++d)</div><div class="line"><a name="l00145"></a><span class="lineno"> 145</span>  {</div><div class="line"><a name="l00146"></a><span class="lineno"> 146</span>  <span class="keywordflow">for</span>(<span class="keywordtype">int</span> y = top_left_y, y_e = top_left_y + kernel_size; y < y_e; ++y)</div><div class="line"><a name="l00147"></a><span class="lineno"> 147</span>  {</div><div class="line"><a name="l00148"></a><span class="lineno"> 148</span>  <span class="keywordflow">for</span>(<span class="keywordtype">int</span> x = top_left_x, x_e = top_left_x + kernel_size; x < x_e; ++x, output_ptr += dst.<a class="code" href="struct_image.xhtml#ae01febbfd0689ef709f3ff6fdd2abc7e">stride_x</a>)</div><div class="line"><a name="l00149"></a><span class="lineno"> 149</span>  {</div><div class="line"><a name="l00150"></a><span class="lineno"> 150</span>  <span class="keywordflow">if</span>(x < 0 || x >= input_dims.x || y < 0 || y >= input_dims.y)</div><div class="line"><a name="l00151"></a><span class="lineno"> 151</span>  {</div><div class="line"><a name="l00152"></a><span class="lineno"> 152</span>  *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)output_ptr) = 0;</div><div class="line"><a name="l00153"></a><span class="lineno"> 153</span>  }</div><div class="line"><a name="l00154"></a><span class="lineno"> 154</span>  <span class="keywordflow">else</span></div><div class="line"><a name="l00155"></a><span class="lineno"> 155</span>  {</div><div class="line"><a name="l00156"></a><span class="lineno"> 156</span>  *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)output_ptr) = *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)(<a class="code" href="helpers_8h.xhtml#a2101b2fe0193ce227ae4e0945e321d85">tensor3D_offset</a>(&src, x, y, d)));</div><div class="line"><a name="l00157"></a><span class="lineno"> 157</span>  }</div><div class="line"><a name="l00158"></a><span class="lineno"> 158</span>  }</div><div class="line"><a name="l00159"></a><span class="lineno"> 159</span>  }</div><div class="line"><a name="l00160"></a><span class="lineno"> 160</span>  }</div><div class="line"><a name="l00161"></a><span class="lineno"> 161</span> </div><div class="line"><a name="l00162"></a><span class="lineno"> 162</span> <span class="preprocessor">#if defined HAS_BIAS</span></div><div class="line"><a name="l00163"></a><span class="lineno"> 163</span>  *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)output_ptr) = 1;</div><div class="line"><a name="l00164"></a><span class="lineno"> 164</span> <span class="preprocessor">#endif</span></div><div class="line"><a name="l00165"></a><span class="lineno"> 165</span> }</div><div class="ttc" id="helpers_8h_xhtml_a31c8c760f08fb1a331b16b7c204321dc"><div class="ttname"><a href="helpers_8h.xhtml#a31c8c760f08fb1a331b16b7c204321dc">CONVERT_TO_TENSOR3D_STRUCT</a></div><div class="ttdeci">#define CONVERT_TO_TENSOR3D_STRUCT(name)</div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00079">helpers.h:79</a></div></div>
-<div class="ttc" id="convolution3x3_8cl_xhtml_afb8c72ce35c4a1f4a2588d6573e54aa1"><div class="ttname"><a href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a></div><div class="ttdeci">#define DATA_TYPE</div><div class="ttdef"><b>Definition:</b> <a href="convolution3x3_8cl_source.xhtml#l00027">convolution3x3.cl:27</a></div></div>
-<div class="ttc" id="helpers_8h_xhtml_a4334a4a76f8e9628c0fb9e1acf616e2a"><div class="ttname"><a href="helpers_8h.xhtml#a4334a4a76f8e9628c0fb9e1acf616e2a">CONVERT_TO_IMAGE_STRUCT_NO_STEP</a></div><div class="ttdeci">#define CONVERT_TO_IMAGE_STRUCT_NO_STEP(name)</div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00076">helpers.h:76</a></div></div>
-<div class="ttc" id="struct_tensor3_d_xhtml"><div class="ttname"><a href="struct_tensor3_d.xhtml">Tensor3D</a></div><div class="ttdoc">Structure to hold 3D tensor information. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00104">helpers.h:104</a></div></div>
-<div class="ttc" id="helpers_8h_xhtml_a2101b2fe0193ce227ae4e0945e321d85"><div class="ttname"><a href="helpers_8h.xhtml#a2101b2fe0193ce227ae4e0945e321d85">tensor3D_offset</a></div><div class="ttdeci">__global const uchar * tensor3D_offset(const Tensor3D *tensor, int x, int y, int z)</div><div class="ttdoc">Get the pointer position of a Tensor3D. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00213">helpers.h:213</a></div></div>
-<div class="ttc" id="struct_image_xhtml"><div class="ttname"><a href="struct_image.xhtml">Image</a></div><div class="ttdoc">Structure to hold Image information. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00095">helpers.h:95</a></div></div>
-<div class="ttc" id="struct_image_xhtml_acf52c23cbd7424606c10a606524e3e32"><div class="ttname"><a href="struct_image.xhtml#acf52c23cbd7424606c10a606524e3e32">Image::ptr</a></div><div class="ttdeci">__global uchar * ptr</div><div class="ttdoc">Pointer to the starting postion of the buffer. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00097">helpers.h:97</a></div></div>
-<div class="ttc" id="struct_image_xhtml_a4f0b90c9ecd6e57ceb3f37332fefe8f1"><div class="ttname"><a href="struct_image.xhtml#a4f0b90c9ecd6e57ceb3f37332fefe8f1">Image::stride_y</a></div><div class="ttdeci">int stride_y</div><div class="ttdoc">Stride of the image in Y dimension (in bytes) </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00100">helpers.h:100</a></div></div>
-<div class="ttc" id="struct_image_xhtml_ae01febbfd0689ef709f3ff6fdd2abc7e"><div class="ttname"><a href="struct_image.xhtml#ae01febbfd0689ef709f3ff6fdd2abc7e">Image::stride_x</a></div><div class="ttdeci">int stride_x</div><div class="ttdoc">Stride of the image in X dimension (in bytes) </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00099">helpers.h:99</a></div></div>
-</div><!-- fragment -->
-</div>
-</div>
<a class="anchor" id="a6f31f2c8d606e1b57d9bdfacd416024c"></a>
<div class="memitem">
<div class="memproto">
@@ -559,7 +233,7 @@
In case biases will be added in late stage, -DHAS_BIAS has to be passed to append the final matrix with 1 in each row.</dd></dl>
<dl class="params"><dt>Parameters</dt><dd>
<table class="params">
- <tr><td class="paramdir">[in]</td><td class="paramname">src_ptr</td><td>Pointer to the source tensor. Supported data types: F16, F32 </td></tr>
+ <tr><td class="paramdir">[in]</td><td class="paramname">src_ptr</td><td>Pointer to the source tensor. Supported data types: QS8/F16/F32 </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">src_stride_x</td><td>Stride of the source tensor in X dimension (in bytes) </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">src_step_x</td><td>src_stride_x * number of elements along X processed per workitem(in bytes) </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">src_stride_y</td><td>Stride of the source tensor in Y dimension (in bytes) </td></tr>
@@ -567,7 +241,7 @@
<tr><td class="paramdir">[in]</td><td class="paramname">src_stride_z</td><td>Stride of the source tensor in Z dimension (in bytes) </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">src_step_z</td><td>src_stride_z * number of elements along Y processed per workitem(in bytes) </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">src_offset_first_element_in_bytes</td><td>The offset of the first element in the source tensor </td></tr>
- <tr><td class="paramdir">[out]</td><td class="paramname">dst_ptr</td><td>Pointer to the destination tensor. Same as input. </td></tr>
+ <tr><td class="paramdir">[out]</td><td class="paramname">dst_ptr</td><td>Pointer to the destination tensor. Same as <code>src_ptr</code> </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">dst_stride_x</td><td>Stride of the destination tensor in X dimension (in bytes) </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">dst_step_x</td><td>dst_stride_x * number of elements along X processed per workitem(in bytes) </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">dst_offset_first_element_in_bytes</td><td>The offset of the first element in the destination tensor </td></tr>
@@ -577,12 +251,34 @@
</dd>
</dl>
-<p>Definition at line <a class="el" href="convolution__layer_8cl_source.xhtml#l00220">220</a> of file <a class="el" href="convolution__layer_8cl_source.xhtml">convolution_layer.cl</a>.</p>
+<p>Definition at line <a class="el" href="convolution__layer_8cl_source.xhtml#l00309">309</a> of file <a class="el" href="convolution__layer_8cl_source.xhtml">convolution_layer.cl</a>.</p>
-<p>References <a class="el" href="helpers_8h_source.xhtml#l00079">CONVERT_TO_TENSOR3D_STRUCT</a>, <a class="el" href="convolution3x3_8cl_source.xhtml#l00027">DATA_TYPE</a>, and <a class="el" href="helpers_8h_source.xhtml#l00106">Tensor3D::ptr</a>.</p>
-<div class="fragment"><div class="line"><a name="l00224"></a><span class="lineno"> 224</span> {</div><div class="line"><a name="l00225"></a><span class="lineno"> 225</span>  <a class="code" href="struct_tensor3_d.xhtml">Tensor3D</a> src = <a class="code" href="helpers_8h.xhtml#a31c8c760f08fb1a331b16b7c204321dc">CONVERT_TO_TENSOR3D_STRUCT</a>(src);</div><div class="line"><a name="l00226"></a><span class="lineno"> 226</span> </div><div class="line"><a name="l00227"></a><span class="lineno"> 227</span>  <span class="keyword">const</span> uint image_size = width * height;</div><div class="line"><a name="l00228"></a><span class="lineno"> 228</span> </div><div class="line"><a name="l00229"></a><span class="lineno"> 229</span>  __global uchar *tmp_out_ptr = dst_ptr + dst_offset_first_element_in_bytes + (get_global_id(0) + get_global_id(1) * width + get_global_id(2) * image_size) * dst_stride_x;</div><div class="line"><a name="l00230"></a><span class="lineno"> 230</span> </div><div class="line"><a name="l00231"></a><span class="lineno"> 231</span>  *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)tmp_out_ptr) = *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)src.ptr);</div><div class="line"><a name="l00232"></a><span class="lineno"> 232</span> </div><div class="line"><a name="l00233"></a><span class="lineno"> 233</span> <span class="preprocessor">#if defined HAS_BIAS</span></div><div class="line"><a name="l00234"></a><span class="lineno"> 234</span>  <span class="comment">// If it is the last thread in the 3 dimensional workgroup</span></div><div class="line"><a name="l00235"></a><span class="lineno"> 235</span>  <span class="keywordflow">if</span>(get_global_id(0) == (get_global_size(0) - 1) && get_global_id(1) == (get_global_size(1) - 1) && get_global_id(2) == (get_global_size(2) - 1))</div><div class="line"><a name="l00236"></a><span class="lineno"> 236</span>  {</div><div class="line"><a name="l00237"></a><span class="lineno"> 237</span>  tmp_out_ptr += dst_stride_x;</div><div class="line"><a name="l00238"></a><span class="lineno"> 238</span>  *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)tmp_out_ptr) = (<a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a>)1;</div><div class="line"><a name="l00239"></a><span class="lineno"> 239</span>  }</div><div class="line"><a name="l00240"></a><span class="lineno"> 240</span> <span class="preprocessor">#endif</span></div><div class="line"><a name="l00241"></a><span class="lineno"> 241</span> }</div><div class="ttc" id="helpers_8h_xhtml_a31c8c760f08fb1a331b16b7c204321dc"><div class="ttname"><a href="helpers_8h.xhtml#a31c8c760f08fb1a331b16b7c204321dc">CONVERT_TO_TENSOR3D_STRUCT</a></div><div class="ttdeci">#define CONVERT_TO_TENSOR3D_STRUCT(name)</div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00079">helpers.h:79</a></div></div>
+<p>References <a class="el" href="helpers_8h_source.xhtml#l00108">CONVERT_TO_TENSOR3D_STRUCT</a>, <a class="el" href="convolution3x3_8cl_source.xhtml#l00027">DATA_TYPE</a>, and <a class="el" href="helpers_8h_source.xhtml#l00142">Tensor3D::ptr</a>.</p>
+<div class="fragment"><div class="line"><a name="l00313"></a><span class="lineno"> 313</span> {</div>
+<div class="line"><a name="l00314"></a><span class="lineno"> 314</span>  <a class="code" href="struct_tensor3_d.xhtml">Tensor3D</a> src = <a class="code" href="helpers_8h.xhtml#a31c8c760f08fb1a331b16b7c204321dc">CONVERT_TO_TENSOR3D_STRUCT</a>(src);</div>
+<div class="line"><a name="l00315"></a><span class="lineno"> 315</span> </div>
+<div class="line"><a name="l00316"></a><span class="lineno"> 316</span>  <span class="keyword">const</span> uint image_size = width * height;</div>
+<div class="line"><a name="l00317"></a><span class="lineno"> 317</span> </div>
+<div class="line"><a name="l00318"></a><span class="lineno"> 318</span>  __global uchar *tmp_out_ptr = dst_ptr + dst_offset_first_element_in_bytes + (get_global_id(0) + get_global_id(1) * width + get_global_id(2) * image_size) * dst_stride_x;</div>
+<div class="line"><a name="l00319"></a><span class="lineno"> 319</span> </div>
+<div class="line"><a name="l00320"></a><span class="lineno"> 320</span>  *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)tmp_out_ptr) = *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)src.ptr);</div>
+<div class="line"><a name="l00321"></a><span class="lineno"> 321</span> </div>
+<div class="line"><a name="l00322"></a><span class="lineno"> 322</span> <span class="preprocessor">#ifdef HAS_BIAS</span></div>
+<div class="line"><a name="l00323"></a><span class="lineno"> 323</span> <span class="preprocessor"></span> <span class="comment">// If it is the last thread in the 3 dimensional workgroup</span></div>
+<div class="line"><a name="l00324"></a><span class="lineno"> 324</span>  <span class="keywordflow">if</span>(get_global_id(0) == (get_global_size(0) - 1) && get_global_id(1) == (get_global_size(1) - 1) && get_global_id(2) == (get_global_size(2) - 1))</div>
+<div class="line"><a name="l00325"></a><span class="lineno"> 325</span>  {</div>
+<div class="line"><a name="l00326"></a><span class="lineno"> 326</span>  tmp_out_ptr += dst_stride_x;</div>
+<div class="line"><a name="l00327"></a><span class="lineno"> 327</span> <span class="preprocessor">#ifdef FIXED_POINT_POSITION</span></div>
+<div class="line"><a name="l00328"></a><span class="lineno"> 328</span> <span class="preprocessor"></span> *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)tmp_out_ptr) = (<a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a>)(1 << FIXED_POINT_POSITION);</div>
+<div class="line"><a name="l00329"></a><span class="lineno"> 329</span> <span class="preprocessor">#else // FIXED_POINT_POSITION</span></div>
+<div class="line"><a name="l00330"></a><span class="lineno"> 330</span> <span class="preprocessor"></span> *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)tmp_out_ptr) = (<a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a>)1;</div>
+<div class="line"><a name="l00331"></a><span class="lineno"> 331</span> <span class="preprocessor">#endif // FIXED_POINT_POSITION</span></div>
+<div class="line"><a name="l00332"></a><span class="lineno"> 332</span> <span class="preprocessor"></span> }</div>
+<div class="line"><a name="l00333"></a><span class="lineno"> 333</span> <span class="preprocessor">#endif // HAS_BIAS</span></div>
+<div class="line"><a name="l00334"></a><span class="lineno"> 334</span> <span class="preprocessor"></span>}</div>
+<div class="ttc" id="helpers_8h_xhtml_a31c8c760f08fb1a331b16b7c204321dc"><div class="ttname"><a href="helpers_8h.xhtml#a31c8c760f08fb1a331b16b7c204321dc">CONVERT_TO_TENSOR3D_STRUCT</a></div><div class="ttdeci">#define CONVERT_TO_TENSOR3D_STRUCT(name)</div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00108">helpers.h:108</a></div></div>
<div class="ttc" id="convolution3x3_8cl_xhtml_afb8c72ce35c4a1f4a2588d6573e54aa1"><div class="ttname"><a href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a></div><div class="ttdeci">#define DATA_TYPE</div><div class="ttdef"><b>Definition:</b> <a href="convolution3x3_8cl_source.xhtml#l00027">convolution3x3.cl:27</a></div></div>
-<div class="ttc" id="struct_tensor3_d_xhtml"><div class="ttname"><a href="struct_tensor3_d.xhtml">Tensor3D</a></div><div class="ttdoc">Structure to hold 3D tensor information. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00104">helpers.h:104</a></div></div>
+<div class="ttc" id="struct_tensor3_d_xhtml"><div class="ttname"><a href="struct_tensor3_d.xhtml">Tensor3D</a></div><div class="ttdoc">Structure to hold 3D tensor information. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00140">helpers.h:140</a></div></div>
</div><!-- fragment -->
</div>
</div>
@@ -734,7 +430,7 @@
<dl class="section note"><dt>Note</dt><dd>Datatype should be given as a preprocessor argument using -DDATA_TYPE=type. e.g. -DDATA_TYPE=short</dd></dl>
<dl class="params"><dt>Parameters</dt><dd>
<table class="params">
- <tr><td class="paramdir">[in]</td><td class="paramname">src_ptr</td><td>Pointer to the source tensor. Supported data types: F16, F32 </td></tr>
+ <tr><td class="paramdir">[in]</td><td class="paramname">src_ptr</td><td>Pointer to the source tensor. Supported data types: F16/F32 </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">src_stride_x</td><td>Stride of the source tensor in X dimension (in bytes) </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">src_step_x</td><td>src_stride_x * number of elements along X processed per workitem(in bytes) </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">src_stride_y</td><td>Stride of the source tensor in Y dimension (in bytes) </td></tr>
@@ -742,13 +438,13 @@
<tr><td class="paramdir">[in]</td><td class="paramname">src_stride_z</td><td>Stride of the source tensor in Z dimension (in bytes) </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">src_step_z</td><td>src_stride_z * number of elements along Y processed per workitem(in bytes) </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">src_offset_first_element_in_bytes</td><td>The offset of the first element in the source tensor </td></tr>
- <tr><td class="paramdir">[out]</td><td class="paramname">dst_ptr</td><td>Pointer to the destination tensor. Same as input </td></tr>
+ <tr><td class="paramdir">[out]</td><td class="paramname">dst_ptr</td><td>Pointer to the destination tensor. Same as <code>src_ptr</code> </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">dst_stride_x</td><td>Stride of the destination tensor in X dimension (in bytes) </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">dst_step_x</td><td>dst_stride_x * number of elements along X processed per workitem(in bytes) </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">dst_stride_y</td><td>Stride of the destination tensor in Y dimension (in bytes) </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">dst_step_y</td><td>dst_stride_y * number of elements along Y processed per workitem(in bytes) </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">dst_offset_first_element_in_bytes</td><td>The offset of the first element in the destination tensor </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">bias_ptr</td><td>Pointer to the bias tensor. Same as input </td></tr>
+ <tr><td class="paramdir">[in]</td><td class="paramname">bias_ptr</td><td>Pointer to the bias tensor. Same as <code>src_ptr</code> </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">bias_stride_x</td><td>Stride of the bias tensor in X dimension (in bytes) </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">bias_step_x</td><td>bias_stride_x * number of elements along X processed per workitem(in bytes) </td></tr>
<tr><td class="paramdir">[in]</td><td class="paramname">bias_offset_first_element_in_bytes</td><td>The offset of the first element in the source tensor </td></tr>
@@ -760,13 +456,48 @@
</dd>
</dl>
-<p>Definition at line <a class="el" href="convolution__layer_8cl_source.xhtml#l00053">53</a> of file <a class="el" href="convolution__layer_8cl_source.xhtml">convolution_layer.cl</a>.</p>
+<p>Definition at line <a class="el" href="convolution__layer_8cl_source.xhtml#l00057">57</a> of file <a class="el" href="convolution__layer_8cl_source.xhtml">convolution_layer.cl</a>.</p>
-<p>References <a class="el" href="helpers_8h_source.xhtml#l00079">CONVERT_TO_TENSOR3D_STRUCT</a>, <a class="el" href="convolution3x3_8cl_source.xhtml#l00027">DATA_TYPE</a>, and <a class="el" href="helpers_8h_source.xhtml#l00106">Tensor3D::ptr</a>.</p>
-<div class="fragment"><div class="line"><a name="l00060"></a><span class="lineno"> 60</span> {</div><div class="line"><a name="l00061"></a><span class="lineno"> 61</span>  <a class="code" href="struct_tensor3_d.xhtml">Tensor3D</a> src = <a class="code" href="helpers_8h.xhtml#a31c8c760f08fb1a331b16b7c204321dc">CONVERT_TO_TENSOR3D_STRUCT</a>(src);</div><div class="line"><a name="l00062"></a><span class="lineno"> 62</span>  <span class="keywordtype">bool</span> is_last_thread = (get_global_id(0) == (get_global_size(0) - 1) && get_global_id(1) == (get_global_size(1) - 1) && get_global_id(2) == (get_global_size(2) - 1));</div><div class="line"><a name="l00063"></a><span class="lineno"> 63</span> </div><div class="line"><a name="l00064"></a><span class="lineno"> 64</span>  __global uchar *tmp_src_ptr = src.<a class="code" href="struct_tensor3_d.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a>;</div><div class="line"><a name="l00065"></a><span class="lineno"> 65</span>  __global uchar *tmp_dst_ptr = dst_ptr + dst_offset_first_element_in_bytes + get_global_id(0) * dst_stride_y + get_global_id(1) * width * dst_stride_y + get_global_id(</div><div class="line"><a name="l00066"></a><span class="lineno"> 66</span>  2) * width * height * dst_stride_y;</div><div class="line"><a name="l00067"></a><span class="lineno"> 67</span> <span class="preprocessor">#if defined HAS_BIAS</span></div><div class="line"><a name="l00068"></a><span class="lineno"> 68</span>  __global uchar *tmp_bias_ptr = bias_ptr + bias_offset_first_element_in_bytes;</div><div class="line"><a name="l00069"></a><span class="lineno"> 69</span> <span class="preprocessor">#endif</span></div><div class="line"><a name="l00070"></a><span class="lineno"> 70</span> </div><div class="line"><a name="l00071"></a><span class="lineno"> 71</span>  <span class="keywordflow">if</span>(is_last_thread)</div><div class="line"><a name="l00072"></a><span class="lineno"> 72</span>  {</div><div class="line"><a name="l00073"></a><span class="lineno"> 73</span>  <span class="keywordflow">for</span>(uint i = 0; i < total_filters; ++i)</div><div class="line"><a name="l00074"></a><span class="lineno"> 74</span>  {</div><div class="line"><a name="l00075"></a><span class="lineno"> 75</span>  *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)tmp_dst_ptr) = *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)tmp_src_ptr);</div><div class="line"><a name="l00076"></a><span class="lineno"> 76</span> </div><div class="line"><a name="l00077"></a><span class="lineno"> 77</span> <span class="preprocessor">#if defined HAS_BIAS</span></div><div class="line"><a name="l00078"></a><span class="lineno"> 78</span>  *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)(tmp_dst_ptr + dst_stride_y)) = *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)(tmp_bias_ptr));</div><div class="line"><a name="l00079"></a><span class="lineno"> 79</span>  tmp_bias_ptr += bias_stride_x;</div><div class="line"><a name="l00080"></a><span class="lineno"> 80</span> <span class="preprocessor">#endif</span></div><div class="line"><a name="l00081"></a><span class="lineno"> 81</span>  tmp_src_ptr += depth * src_stride_z;</div><div class="line"><a name="l00082"></a><span class="lineno"> 82</span>  tmp_dst_ptr += dst_stride_x;</div><div class="line"><a name="l00083"></a><span class="lineno"> 83</span>  }</div><div class="line"><a name="l00084"></a><span class="lineno"> 84</span>  }</div><div class="line"><a name="l00085"></a><span class="lineno"> 85</span>  <span class="keywordflow">else</span></div><div class="line"><a name="l00086"></a><span class="lineno"> 86</span>  {</div><div class="line"><a name="l00087"></a><span class="lineno"> 87</span>  <span class="keywordflow">for</span>(uint i = 0; i < total_filters; ++i)</div><div class="line"><a name="l00088"></a><span class="lineno"> 88</span>  {</div><div class="line"><a name="l00089"></a><span class="lineno"> 89</span>  *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)tmp_dst_ptr) = *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)tmp_src_ptr);</div><div class="line"><a name="l00090"></a><span class="lineno"> 90</span>  tmp_src_ptr += depth * src_stride_z;</div><div class="line"><a name="l00091"></a><span class="lineno"> 91</span>  tmp_dst_ptr += dst_stride_x;</div><div class="line"><a name="l00092"></a><span class="lineno"> 92</span>  }</div><div class="line"><a name="l00093"></a><span class="lineno"> 93</span>  }</div><div class="line"><a name="l00094"></a><span class="lineno"> 94</span> }</div><div class="ttc" id="helpers_8h_xhtml_a31c8c760f08fb1a331b16b7c204321dc"><div class="ttname"><a href="helpers_8h.xhtml#a31c8c760f08fb1a331b16b7c204321dc">CONVERT_TO_TENSOR3D_STRUCT</a></div><div class="ttdeci">#define CONVERT_TO_TENSOR3D_STRUCT(name)</div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00079">helpers.h:79</a></div></div>
+<p>References <a class="el" href="helpers_8h_source.xhtml#l00108">CONVERT_TO_TENSOR3D_STRUCT</a>, <a class="el" href="convolution3x3_8cl_source.xhtml#l00027">DATA_TYPE</a>, and <a class="el" href="helpers_8h_source.xhtml#l00142">Tensor3D::ptr</a>.</p>
+<div class="fragment"><div class="line"><a name="l00064"></a><span class="lineno"> 64</span> {</div>
+<div class="line"><a name="l00065"></a><span class="lineno"> 65</span>  <a class="code" href="struct_tensor3_d.xhtml">Tensor3D</a> src = <a class="code" href="helpers_8h.xhtml#a31c8c760f08fb1a331b16b7c204321dc">CONVERT_TO_TENSOR3D_STRUCT</a>(src);</div>
+<div class="line"><a name="l00066"></a><span class="lineno"> 66</span>  <span class="keywordtype">bool</span> is_last_thread = (get_global_id(0) == (get_global_size(0) - 1) && get_global_id(1) == (get_global_size(1) - 1) && get_global_id(2) == (get_global_size(2) - 1));</div>
+<div class="line"><a name="l00067"></a><span class="lineno"> 67</span> </div>
+<div class="line"><a name="l00068"></a><span class="lineno"> 68</span>  __global uchar *tmp_src_ptr = src.<a class="code" href="struct_tensor3_d.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a>;</div>
+<div class="line"><a name="l00069"></a><span class="lineno"> 69</span>  __global uchar *tmp_dst_ptr = dst_ptr + dst_offset_first_element_in_bytes + get_global_id(0) * dst_stride_y + get_global_id(1) * width * dst_stride_y + get_global_id(</div>
+<div class="line"><a name="l00070"></a><span class="lineno"> 70</span>  2) * width * height * dst_stride_y;</div>
+<div class="line"><a name="l00071"></a><span class="lineno"> 71</span> <span class="preprocessor">#ifdef HAS_BIAS</span></div>
+<div class="line"><a name="l00072"></a><span class="lineno"> 72</span> <span class="preprocessor"></span> __global uchar *tmp_bias_ptr = bias_ptr + bias_offset_first_element_in_bytes;</div>
+<div class="line"><a name="l00073"></a><span class="lineno"> 73</span> <span class="preprocessor">#endif </span><span class="comment">/* HAS_BIAS */</span><span class="preprocessor"></span></div>
+<div class="line"><a name="l00074"></a><span class="lineno"> 74</span> <span class="preprocessor"></span></div>
+<div class="line"><a name="l00075"></a><span class="lineno"> 75</span>  <span class="keywordflow">if</span>(is_last_thread)</div>
+<div class="line"><a name="l00076"></a><span class="lineno"> 76</span>  {</div>
+<div class="line"><a name="l00077"></a><span class="lineno"> 77</span>  <span class="keywordflow">for</span>(uint i = 0; i < total_filters; ++i)</div>
+<div class="line"><a name="l00078"></a><span class="lineno"> 78</span>  {</div>
+<div class="line"><a name="l00079"></a><span class="lineno"> 79</span>  *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)tmp_dst_ptr) = *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)tmp_src_ptr);</div>
+<div class="line"><a name="l00080"></a><span class="lineno"> 80</span> </div>
+<div class="line"><a name="l00081"></a><span class="lineno"> 81</span> <span class="preprocessor">#ifdef HAS_BIAS</span></div>
+<div class="line"><a name="l00082"></a><span class="lineno"> 82</span> <span class="preprocessor"></span> *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)(tmp_dst_ptr + dst_stride_y)) = *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)(tmp_bias_ptr));</div>
+<div class="line"><a name="l00083"></a><span class="lineno"> 83</span>  tmp_bias_ptr += bias_stride_x;</div>
+<div class="line"><a name="l00084"></a><span class="lineno"> 84</span> <span class="preprocessor">#endif </span><span class="comment">/* HAS_BIAS */</span><span class="preprocessor"></span></div>
+<div class="line"><a name="l00085"></a><span class="lineno"> 85</span> <span class="preprocessor"></span> tmp_src_ptr += depth * src_stride_z;</div>
+<div class="line"><a name="l00086"></a><span class="lineno"> 86</span>  tmp_dst_ptr += dst_stride_x;</div>
+<div class="line"><a name="l00087"></a><span class="lineno"> 87</span>  }</div>
+<div class="line"><a name="l00088"></a><span class="lineno"> 88</span>  }</div>
+<div class="line"><a name="l00089"></a><span class="lineno"> 89</span>  <span class="keywordflow">else</span></div>
+<div class="line"><a name="l00090"></a><span class="lineno"> 90</span>  {</div>
+<div class="line"><a name="l00091"></a><span class="lineno"> 91</span>  <span class="keywordflow">for</span>(uint i = 0; i < total_filters; ++i)</div>
+<div class="line"><a name="l00092"></a><span class="lineno"> 92</span>  {</div>
+<div class="line"><a name="l00093"></a><span class="lineno"> 93</span>  *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)tmp_dst_ptr) = *((__global <a class="code" href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a> *)tmp_src_ptr);</div>
+<div class="line"><a name="l00094"></a><span class="lineno"> 94</span>  tmp_src_ptr += depth * src_stride_z;</div>
+<div class="line"><a name="l00095"></a><span class="lineno"> 95</span>  tmp_dst_ptr += dst_stride_x;</div>
+<div class="line"><a name="l00096"></a><span class="lineno"> 96</span>  }</div>
+<div class="line"><a name="l00097"></a><span class="lineno"> 97</span>  }</div>
+<div class="line"><a name="l00098"></a><span class="lineno"> 98</span> }</div>
+<div class="ttc" id="helpers_8h_xhtml_a31c8c760f08fb1a331b16b7c204321dc"><div class="ttname"><a href="helpers_8h.xhtml#a31c8c760f08fb1a331b16b7c204321dc">CONVERT_TO_TENSOR3D_STRUCT</a></div><div class="ttdeci">#define CONVERT_TO_TENSOR3D_STRUCT(name)</div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00108">helpers.h:108</a></div></div>
<div class="ttc" id="convolution3x3_8cl_xhtml_afb8c72ce35c4a1f4a2588d6573e54aa1"><div class="ttname"><a href="convolution3x3_8cl.xhtml#afb8c72ce35c4a1f4a2588d6573e54aa1">DATA_TYPE</a></div><div class="ttdeci">#define DATA_TYPE</div><div class="ttdef"><b>Definition:</b> <a href="convolution3x3_8cl_source.xhtml#l00027">convolution3x3.cl:27</a></div></div>
-<div class="ttc" id="struct_tensor3_d_xhtml"><div class="ttname"><a href="struct_tensor3_d.xhtml">Tensor3D</a></div><div class="ttdoc">Structure to hold 3D tensor information. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00104">helpers.h:104</a></div></div>
-<div class="ttc" id="struct_tensor3_d_xhtml_acf52c23cbd7424606c10a606524e3e32"><div class="ttname"><a href="struct_tensor3_d.xhtml#acf52c23cbd7424606c10a606524e3e32">Tensor3D::ptr</a></div><div class="ttdeci">__global uchar * ptr</div><div class="ttdoc">Pointer to the starting postion of the buffer. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00106">helpers.h:106</a></div></div>
+<div class="ttc" id="struct_tensor3_d_xhtml"><div class="ttname"><a href="struct_tensor3_d.xhtml">Tensor3D</a></div><div class="ttdoc">Structure to hold 3D tensor information. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00140">helpers.h:140</a></div></div>
+<div class="ttc" id="struct_tensor3_d_xhtml_acf52c23cbd7424606c10a606524e3e32"><div class="ttname"><a href="struct_tensor3_d.xhtml#acf52c23cbd7424606c10a606524e3e32">Tensor3D::ptr</a></div><div class="ttdeci">__global uchar * ptr</div><div class="ttdoc">Pointer to the starting postion of the buffer. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00142">helpers.h:142</a></div></div>
</div><!-- fragment -->
</div>
</div>
@@ -775,10 +506,10 @@
<!-- start footer part -->
<div id="nav-path" class="navpath"><!-- id is needed for treeview function! -->
<ul>
- <li class="navelem"><a class="el" href="dir_68267d1309a1af8e8297ef4c3efbcdba.xhtml">src</a></li><li class="navelem"><a class="el" href="dir_aebb8dcc11953d78e620bbef0b9e2183.xhtml">core</a></li><li class="navelem"><a class="el" href="dir_8c278f79c760e5c5fbd911f9870614c1.xhtml">CL</a></li><li class="navelem"><a class="el" href="dir_25885286e9dad4fa105b7b25a8031bbf.xhtml">cl_kernels</a></li><li class="navelem"><a class="el" href="convolution__layer_8cl.xhtml">convolution_layer.cl</a></li>
- <li class="footer">Generated on Fri Jun 23 2017 15:44:34 for Compute Library by
+ <li class="navelem"><a class="el" href="dir_55b5a5006e943fb664ff8cff3cfe7768.xhtml">src</a></li><li class="navelem"><a class="el" href="dir_f6c3ae42d7e27145d0a3c3e38ca1c000.xhtml">core</a></li><li class="navelem"><a class="el" href="dir_1b7ae9123c88e650426d50d95c19d414.xhtml">CL</a></li><li class="navelem"><a class="el" href="dir_b7c3d80b0dadb2139bc73b7161751c12.xhtml">cl_kernels</a></li><li class="navelem"><a class="el" href="convolution__layer_8cl.xhtml">convolution_layer.cl</a></li>
+ <li class="footer">Generated on Thu Sep 28 2017 14:37:53 for Compute Library by
<a href="http://www.doxygen.org/index.html">
- <img class="footer" src="doxygen.png" alt="doxygen"/></a> 1.8.11 </li>
+ <img class="footer" src="doxygen.png" alt="doxygen"/></a> 1.8.6 </li>
</ul>
</div>
</body>