arm_compute v18.02
Change-Id: I7207aa488e5470f235f39b6c188b4678dc38d1a6
diff --git a/documentation/gemm_8cl.xhtml b/documentation/gemm_8cl.xhtml
index 6b254aa..e0fe208 100644
--- a/documentation/gemm_8cl.xhtml
+++ b/documentation/gemm_8cl.xhtml
@@ -40,7 +40,7 @@
<tr style="height: 56px;">
<td style="padding-left: 0.5em;">
<div id="projectname">Compute Library
-  <span id="projectnumber">18.01</span>
+  <span id="projectnumber">18.02</span>
</div>
</td>
</tr>
@@ -113,8 +113,6 @@
</div>
<div class="header">
- <div class="summary">
-<a href="#func-members">Functions</a> </div>
<div class="headertitle">
<div class="title">gemm.cl File Reference</div> </div>
</div><!--header-->
@@ -122,717 +120,13 @@
<div class="textblock"><code>#include "<a class="el" href="helpers_8h_source.xhtml">helpers.h</a>"</code><br />
</div>
<p><a href="gemm_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:ac5eba21b9ac015e1454044aeba93546c"><td class="memItemLeft" align="right" valign="top">__kernel void </td><td class="memItemRight" valign="bottom"><a class="el" href="gemm_8cl.xhtml#ac5eba21b9ac015e1454044aeba93546c">gemm_transpose1x4</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_offset_first_element_in_bytes)</td></tr>
-<tr class="memdesc:ac5eba21b9ac015e1454044aeba93546c"><td class="mdescLeft"> </td><td class="mdescRight">This OpenCL kernel computes the "vector" 1x4 transposition of input matrix. <a href="#ac5eba21b9ac015e1454044aeba93546c">More...</a><br /></td></tr>
-<tr class="separator:ac5eba21b9ac015e1454044aeba93546c"><td class="memSeparator" colspan="2"> </td></tr>
-<tr class="memitem:a75c35ac120f123e9cce56c43ba3f48d2"><td class="memItemLeft" align="right" valign="top">__kernel void </td><td class="memItemRight" valign="bottom"><a class="el" href="gemm_8cl.xhtml#a75c35ac120f123e9cce56c43ba3f48d2">gemm_transpose1x8</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_offset_first_element_in_bytes)</td></tr>
-<tr class="memdesc:a75c35ac120f123e9cce56c43ba3f48d2"><td class="mdescLeft"> </td><td class="mdescRight">This OpenCL kernel computes the "vector" 1x8 transposition of input matrix. <a href="#a75c35ac120f123e9cce56c43ba3f48d2">More...</a><br /></td></tr>
-<tr class="separator:a75c35ac120f123e9cce56c43ba3f48d2"><td class="memSeparator" colspan="2"> </td></tr>
-<tr class="memitem:a97a299ffb620aa98fe5e39bb8ab0833f"><td class="memItemLeft" align="right" valign="top">__kernel void </td><td class="memItemRight" valign="bottom"><a class="el" href="gemm_8cl.xhtml#a97a299ffb620aa98fe5e39bb8ab0833f">gemm_transpose1x16</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_offset_first_element_in_bytes)</td></tr>
-<tr class="memdesc:a97a299ffb620aa98fe5e39bb8ab0833f"><td class="mdescLeft"> </td><td class="mdescRight">This OpenCL kernel computes the "vector" 1x16 transposition of input matrix. <a href="#a97a299ffb620aa98fe5e39bb8ab0833f">More...</a><br /></td></tr>
-<tr class="separator:a97a299ffb620aa98fe5e39bb8ab0833f"><td class="memSeparator" colspan="2"> </td></tr>
-<tr class="memitem:a560b2d50aa886edc0f5daf4fe729717f"><td class="memItemLeft" align="right" valign="top">__kernel void </td><td class="memItemRight" valign="bottom"><a class="el" href="gemm_8cl.xhtml#a560b2d50aa886edc0f5daf4fe729717f">gemm_interleave4x4_32bit</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_offset_first_element_in_bytes)</td></tr>
-<tr class="memdesc:a560b2d50aa886edc0f5daf4fe729717f"><td class="mdescLeft"> </td><td class="mdescRight">This OpenCL kernel reshapes the input matrix transposing each 4x4 block and interleaving the values. <a href="#a560b2d50aa886edc0f5daf4fe729717f">More...</a><br /></td></tr>
-<tr class="separator:a560b2d50aa886edc0f5daf4fe729717f"><td class="memSeparator" colspan="2"> </td></tr>
-<tr class="memitem:a2f32d740e780059f68da8aa589ed0a5b"><td class="memItemLeft" align="right" valign="top">__kernel void </td><td class="memItemRight" valign="bottom"><a class="el" href="gemm_8cl.xhtml#a2f32d740e780059f68da8aa589ed0a5b">gemm_interleave4x4_16bit</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_offset_first_element_in_bytes)</td></tr>
-<tr class="memdesc:a2f32d740e780059f68da8aa589ed0a5b"><td class="mdescLeft"> </td><td class="mdescRight">This OpenCL kernel reshapes the input matrix transposing each 4x4 block and interleaving the values. <a href="#a2f32d740e780059f68da8aa589ed0a5b">More...</a><br /></td></tr>
-<tr class="separator:a2f32d740e780059f68da8aa589ed0a5b"><td class="memSeparator" colspan="2"> </td></tr>
-<tr class="memitem:aa66a7b6b0420e54ec173743e6c5bfc45"><td class="memItemLeft" align="right" valign="top">__kernel void </td><td class="memItemRight" valign="bottom"><a class="el" href="gemm_8cl.xhtml#aa66a7b6b0420e54ec173743e6c5bfc45">gemm_interleave4x4_8bit</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_offset_first_element_in_bytes)</td></tr>
-<tr class="memdesc:aa66a7b6b0420e54ec173743e6c5bfc45"><td class="mdescLeft"> </td><td class="mdescRight">This OpenCL kernel reshapes the input matrix transposing each 4x4 block and interleaving the values. <a href="#aa66a7b6b0420e54ec173743e6c5bfc45">More...</a><br /></td></tr>
-<tr class="separator:aa66a7b6b0420e54ec173743e6c5bfc45"><td class="memSeparator" colspan="2"> </td></tr>
-</table>
-<h2 class="groupheader">Function Documentation</h2>
-<a class="anchor" id="a2f32d740e780059f68da8aa589ed0a5b"></a>
-<div class="memitem">
-<div class="memproto">
- <table class="memname">
- <tr>
- <td class="memname">__kernel void gemm_interleave4x4_16bit </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_offset_first_element_in_bytes</em> </td>
- </tr>
- <tr>
- <td></td>
- <td>)</td>
- <td></td><td></td>
- </tr>
- </table>
-</div><div class="memdoc">
-
-<p>This OpenCL kernel reshapes the input matrix transposing each 4x4 block and interleaving the values. </p>
-<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 matrix. Supported data types: U16/S16/QS16/F16 </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_stride_x</td><td>Stride of the source matrix 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 matrix 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 matrix </td></tr>
- <tr><td class="paramdir">[out]</td><td class="paramname">dst_ptr</td><td>Pointer to the destination matrix Supported data types: 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 matrix in X dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_x</td><td>dst_gx_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 matrix in Y dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_y</td><td>dst_gx_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 matrix </td></tr>
- </table>
- </dd>
-</dl>
-
-<p>Definition at line <a class="el" href="gemm_8cl_source.xhtml#l00182">182</a> of file <a class="el" href="gemm_8cl_source.xhtml">gemm.cl</a>.</p>
-
-<p>References <a class="el" href="helpers_8h_source.xhtml#l00102">CONVERT_TO_IMAGE_STRUCT</a>, <a class="el" href="_c_l_2_convolution_8cpp_source.xhtml#l00123">arm_compute::test::validation::dst</a>, <a class="el" href="helpers_8h_source.xhtml#l00301">offset()</a>, <a class="el" href="helpers_8h_source.xhtml#l00142">Image::ptr</a>, and <a class="el" href="_c_l_2_convolution_8cpp_source.xhtml#l00133">arm_compute::test::validation::src</a>.</p>
-<div class="fragment"><div class="line"><a name="l00184"></a><span class="lineno"> 184</span> {</div><div class="line"><a name="l00185"></a><span class="lineno"> 185</span>  <span class="comment">// Compute source and destination addresses</span></div><div class="line"><a name="l00186"></a><span class="lineno"> 186</span>  <a class="code" href="struct_image.xhtml">Image</a> <a class="code" href="namespacearm__compute_1_1test_1_1validation.xhtml#a6743f0a130e8311e6f5b1a23df102472">src</a> = <a class="code" href="helpers_8h.xhtml#aebe814363556c244be043b13e7969197">CONVERT_TO_IMAGE_STRUCT</a>(src);</div><div class="line"><a name="l00187"></a><span class="lineno"> 187</span>  <a class="code" href="struct_image.xhtml">Image</a> <a class="code" href="namespacearm__compute_1_1test_1_1validation.xhtml#adbf67dcee294e673cf796f1ed8aeb6a4">dst</a> = <a class="code" href="helpers_8h.xhtml#aebe814363556c244be043b13e7969197">CONVERT_TO_IMAGE_STRUCT</a>(dst);</div><div class="line"><a name="l00188"></a><span class="lineno"> 188</span> </div><div class="line"><a name="l00189"></a><span class="lineno"> 189</span>  <span class="comment">// Load values from Matrix A</span></div><div class="line"><a name="l00190"></a><span class="lineno"> 190</span>  ushort8 a0 = vload8(0, (__global ushort *)(<a class="code" href="helpers_8h.xhtml#a009469e4d9b8fce3b6d5e97d2077827d">offset</a>(&src, 0, 0)));</div><div class="line"><a name="l00191"></a><span class="lineno"> 191</span>  ushort8 a1 = vload8(0, (__global ushort *)(<a class="code" href="helpers_8h.xhtml#a009469e4d9b8fce3b6d5e97d2077827d">offset</a>(&src, 0, 1)));</div><div class="line"><a name="l00192"></a><span class="lineno"> 192</span>  ushort8 a2 = vload8(0, (__global ushort *)(<a class="code" href="helpers_8h.xhtml#a009469e4d9b8fce3b6d5e97d2077827d">offset</a>(&src, 0, 2)));</div><div class="line"><a name="l00193"></a><span class="lineno"> 193</span>  ushort8 a3 = vload8(0, (__global ushort *)(<a class="code" href="helpers_8h.xhtml#a009469e4d9b8fce3b6d5e97d2077827d">offset</a>(&src, 0, 3)));</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>  ushort8 val0 = (ushort8)((ushort4)(a0.s0, a1.s0, a2.s0, a3.s0), (ushort4)(a0.s1, a1.s1, a2.s1, a3.s1));</div><div class="line"><a name="l00196"></a><span class="lineno"> 196</span>  vstore8(val0, 0, ((__global ushort *)dst.<a class="code" href="struct_image.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a>) + 0);</div><div class="line"><a name="l00197"></a><span class="lineno"> 197</span> </div><div class="line"><a name="l00198"></a><span class="lineno"> 198</span>  val0 = (ushort8)((ushort4)(a0.s2, a1.s2, a2.s2, a3.s2), (ushort4)(a0.s3, a1.s3, a2.s3, a3.s3));</div><div class="line"><a name="l00199"></a><span class="lineno"> 199</span>  vstore8(val0, 0, ((__global ushort *)dst.<a class="code" href="struct_image.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a>) + 8);</div><div class="line"><a name="l00200"></a><span class="lineno"> 200</span> </div><div class="line"><a name="l00201"></a><span class="lineno"> 201</span>  val0 = (ushort8)((ushort4)(a0.s4, a1.s4, a2.s4, a3.s4), (ushort4)(a0.s5, a1.s5, a2.s5, a3.s5));</div><div class="line"><a name="l00202"></a><span class="lineno"> 202</span>  vstore8(val0, 0, ((__global ushort *)dst.<a class="code" href="struct_image.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a>) + 16);</div><div class="line"><a name="l00203"></a><span class="lineno"> 203</span> </div><div class="line"><a name="l00204"></a><span class="lineno"> 204</span>  val0 = (ushort8)((ushort4)(a0.s6, a1.s6, a2.s6, a3.s6), (ushort4)(a0.s7, a1.s7, a2.s7, a3.s7));</div><div class="line"><a name="l00205"></a><span class="lineno"> 205</span>  vstore8(val0, 0, ((__global ushort *)dst.<a class="code" href="struct_image.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a>) + 24);</div><div class="line"><a name="l00206"></a><span class="lineno"> 206</span> }</div><div class="ttc" id="helpers_8h_xhtml_a009469e4d9b8fce3b6d5e97d2077827d"><div class="ttname"><a href="helpers_8h.xhtml#a009469e4d9b8fce3b6d5e97d2077827d">offset</a></div><div class="ttdeci">__global uchar * offset(const Image *img, int x, int y)</div><div class="ttdoc">Get the pointer position of a Image. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00301">helpers.h:301</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#l00102">helpers.h:102</a></div></div>
-<div class="ttc" id="namespacearm__compute_1_1test_1_1validation_xhtml_adbf67dcee294e673cf796f1ed8aeb6a4"><div class="ttname"><a href="namespacearm__compute_1_1test_1_1validation.xhtml#adbf67dcee294e673cf796f1ed8aeb6a4">arm_compute::test::validation::dst</a></div><div class="ttdeci">CLTensor dst</div><div class="ttdef"><b>Definition:</b> <a href="_c_l_2_convolution_8cpp_source.xhtml#l00123">Convolution.cpp:123</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#l00140">helpers.h:140</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#l00142">helpers.h:142</a></div></div>
-<div class="ttc" id="namespacearm__compute_1_1test_1_1validation_xhtml_a6743f0a130e8311e6f5b1a23df102472"><div class="ttname"><a href="namespacearm__compute_1_1test_1_1validation.xhtml#a6743f0a130e8311e6f5b1a23df102472">arm_compute::test::validation::src</a></div><div class="ttdeci">convolution configure & src</div><div class="ttdef"><b>Definition:</b> <a href="_c_l_2_convolution_8cpp_source.xhtml#l00133">Convolution.cpp:133</a></div></div>
-</div><!-- fragment -->
-</div>
-</div>
-<a class="anchor" id="a560b2d50aa886edc0f5daf4fe729717f"></a>
-<div class="memitem">
-<div class="memproto">
- <table class="memname">
- <tr>
- <td class="memname">__kernel void gemm_interleave4x4_32bit </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_offset_first_element_in_bytes</em> </td>
- </tr>
- <tr>
- <td></td>
- <td>)</td>
- <td></td><td></td>
- </tr>
- </table>
-</div><div class="memdoc">
-
-<p>This OpenCL kernel reshapes the input matrix transposing each 4x4 block and interleaving the values. </p>
-<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 matrix. Supported data types: U32/S32/F32 </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_stride_x</td><td>Stride of the source matrix 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 matrix 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 matrix </td></tr>
- <tr><td class="paramdir">[out]</td><td class="paramname">dst_ptr</td><td>Pointer to the destination matrix Supported data types: 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 matrix in X dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_x</td><td>dst_gx_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 matrix in Y dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_y</td><td>dst_gx_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 matrix </td></tr>
- </table>
- </dd>
-</dl>
-
-<p>Definition at line <a class="el" href="gemm_8cl_source.xhtml#l00141">141</a> of file <a class="el" href="gemm_8cl_source.xhtml">gemm.cl</a>.</p>
-
-<p>References <a class="el" href="helpers_8h_source.xhtml#l00102">CONVERT_TO_IMAGE_STRUCT</a>, <a class="el" href="_c_l_2_convolution_8cpp_source.xhtml#l00123">arm_compute::test::validation::dst</a>, <a class="el" href="helpers_8h_source.xhtml#l00301">offset()</a>, <a class="el" href="helpers_8h_source.xhtml#l00142">Image::ptr</a>, and <a class="el" href="_c_l_2_convolution_8cpp_source.xhtml#l00133">arm_compute::test::validation::src</a>.</p>
-<div class="fragment"><div class="line"><a name="l00143"></a><span class="lineno"> 143</span> {</div><div class="line"><a name="l00144"></a><span class="lineno"> 144</span>  <span class="comment">// Compute source and destination addresses</span></div><div class="line"><a name="l00145"></a><span class="lineno"> 145</span>  <a class="code" href="struct_image.xhtml">Image</a> <a class="code" href="namespacearm__compute_1_1test_1_1validation.xhtml#a6743f0a130e8311e6f5b1a23df102472">src</a> = <a class="code" href="helpers_8h.xhtml#aebe814363556c244be043b13e7969197">CONVERT_TO_IMAGE_STRUCT</a>(src);</div><div class="line"><a name="l00146"></a><span class="lineno"> 146</span>  <a class="code" href="struct_image.xhtml">Image</a> <a class="code" href="namespacearm__compute_1_1test_1_1validation.xhtml#adbf67dcee294e673cf796f1ed8aeb6a4">dst</a> = <a class="code" href="helpers_8h.xhtml#aebe814363556c244be043b13e7969197">CONVERT_TO_IMAGE_STRUCT</a>(dst);</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="comment">// Load values from Matrix A</span></div><div class="line"><a name="l00149"></a><span class="lineno"> 149</span>  uint4 a0 = vload4(0, (__global uint *)(<a class="code" href="helpers_8h.xhtml#a009469e4d9b8fce3b6d5e97d2077827d">offset</a>(&src, 0, 0)));</div><div class="line"><a name="l00150"></a><span class="lineno"> 150</span>  uint4 a1 = vload4(0, (__global uint *)(<a class="code" href="helpers_8h.xhtml#a009469e4d9b8fce3b6d5e97d2077827d">offset</a>(&src, 0, 1)));</div><div class="line"><a name="l00151"></a><span class="lineno"> 151</span>  uint4 a2 = vload4(0, (__global uint *)(<a class="code" href="helpers_8h.xhtml#a009469e4d9b8fce3b6d5e97d2077827d">offset</a>(&src, 0, 2)));</div><div class="line"><a name="l00152"></a><span class="lineno"> 152</span>  uint4 a3 = vload4(0, (__global uint *)(<a class="code" href="helpers_8h.xhtml#a009469e4d9b8fce3b6d5e97d2077827d">offset</a>(&src, 0, 3)));</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>  uint4 val0 = (uint4)(a0.s0, a1.s0, a2.s0, a3.s0);</div><div class="line"><a name="l00155"></a><span class="lineno"> 155</span>  vstore4(val0, 0, ((__global uint *)dst.<a class="code" href="struct_image.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a>) + 0);</div><div class="line"><a name="l00156"></a><span class="lineno"> 156</span> </div><div class="line"><a name="l00157"></a><span class="lineno"> 157</span>  val0 = (uint4)(a0.s1, a1.s1, a2.s1, a3.s1);</div><div class="line"><a name="l00158"></a><span class="lineno"> 158</span>  vstore4(val0, 0, ((__global uint *)dst.<a class="code" href="struct_image.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a>) + 4);</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>  val0 = (uint4)(a0.s2, a1.s2, a2.s2, a3.s2);</div><div class="line"><a name="l00161"></a><span class="lineno"> 161</span>  vstore4(val0, 0, ((__global uint *)dst.<a class="code" href="struct_image.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a>) + 8);</div><div class="line"><a name="l00162"></a><span class="lineno"> 162</span> </div><div class="line"><a name="l00163"></a><span class="lineno"> 163</span>  val0 = (uint4)(a0.s3, a1.s3, a2.s3, a3.s3);</div><div class="line"><a name="l00164"></a><span class="lineno"> 164</span>  vstore4(val0, 0, ((__global uint *)dst.<a class="code" href="struct_image.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a>) + 12);</div><div class="line"><a name="l00165"></a><span class="lineno"> 165</span> }</div><div class="ttc" id="helpers_8h_xhtml_a009469e4d9b8fce3b6d5e97d2077827d"><div class="ttname"><a href="helpers_8h.xhtml#a009469e4d9b8fce3b6d5e97d2077827d">offset</a></div><div class="ttdeci">__global uchar * offset(const Image *img, int x, int y)</div><div class="ttdoc">Get the pointer position of a Image. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00301">helpers.h:301</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#l00102">helpers.h:102</a></div></div>
-<div class="ttc" id="namespacearm__compute_1_1test_1_1validation_xhtml_adbf67dcee294e673cf796f1ed8aeb6a4"><div class="ttname"><a href="namespacearm__compute_1_1test_1_1validation.xhtml#adbf67dcee294e673cf796f1ed8aeb6a4">arm_compute::test::validation::dst</a></div><div class="ttdeci">CLTensor dst</div><div class="ttdef"><b>Definition:</b> <a href="_c_l_2_convolution_8cpp_source.xhtml#l00123">Convolution.cpp:123</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#l00140">helpers.h:140</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#l00142">helpers.h:142</a></div></div>
-<div class="ttc" id="namespacearm__compute_1_1test_1_1validation_xhtml_a6743f0a130e8311e6f5b1a23df102472"><div class="ttname"><a href="namespacearm__compute_1_1test_1_1validation.xhtml#a6743f0a130e8311e6f5b1a23df102472">arm_compute::test::validation::src</a></div><div class="ttdeci">convolution configure & src</div><div class="ttdef"><b>Definition:</b> <a href="_c_l_2_convolution_8cpp_source.xhtml#l00133">Convolution.cpp:133</a></div></div>
-</div><!-- fragment -->
-</div>
-</div>
-<a class="anchor" id="aa66a7b6b0420e54ec173743e6c5bfc45"></a>
-<div class="memitem">
-<div class="memproto">
- <table class="memname">
- <tr>
- <td class="memname">__kernel void gemm_interleave4x4_8bit </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_offset_first_element_in_bytes</em> </td>
- </tr>
- <tr>
- <td></td>
- <td>)</td>
- <td></td><td></td>
- </tr>
- </table>
-</div><div class="memdoc">
-
-<p>This OpenCL kernel reshapes the input matrix transposing each 4x4 block and interleaving the values. </p>
-<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 matrix. Supported data types: U8/S8/QS8 </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_stride_x</td><td>Stride of the source matrix 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 matrix 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 matrix </td></tr>
- <tr><td class="paramdir">[out]</td><td class="paramname">dst_ptr</td><td>Pointer to the destination matrix Supported data types: 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 matrix in X dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_x</td><td>dst_gx_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 matrix in Y dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_y</td><td>dst_gx_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 matrix </td></tr>
- </table>
- </dd>
-</dl>
-
-<p>Definition at line <a class="el" href="gemm_8cl_source.xhtml#l00223">223</a> of file <a class="el" href="gemm_8cl_source.xhtml">gemm.cl</a>.</p>
-
-<p>References <a class="el" href="fixed__point_8h_source.xhtml#l00178">ADD_SAT_OP_EXPAND</a>, <a class="el" href="validation_2_c_l_2_g_e_m_m_8cpp_source.xhtml#l00120">arm_compute::test::validation::c</a>, <a class="el" href="helpers_8h_source.xhtml#l00102">CONVERT_TO_IMAGE_STRUCT</a>, <a class="el" href="helpers_8h_source.xhtml#l00096">CONVERT_TO_VECTOR_STRUCT</a>, <a class="el" href="convolution3x3_8cl_source.xhtml#l00027">DATA_TYPE</a>, <a class="el" href="_c_l_2_convolution_8cpp_source.xhtml#l00123">arm_compute::test::validation::dst</a>, <a class="el" href="helpers_8h_source.xhtml#l00066">IMAGE_DECLARATION</a>, <a class="el" href="fixed__point_8h_source.xhtml#l00276">mla_sat_qs16x8()</a>, <a class="el" href="fixed__point_8h_source.xhtml#l00275">mla_sat_qs8x16()</a>, <a class="el" href="fixed__point_8h_source.xhtml#l00296">mlal_sat_qs16x8()</a>, <a class="el" href="fixed__point_8h_source.xhtml#l00295">mlal_sat_qs8x8()</a>, <a class="el" href="fixed__point_8h_source.xhtml#l00254">mul_sat_qs16x8()</a>, <a class="el" href="fixed__point_8h_source.xhtml#l00249">mul_sat_qs8x16()</a>, <a class="el" href="helpers_8h_source.xhtml#l00301">offset()</a>, <a class="el" href="helpers_8h_source.xhtml#l00134">Vector::ptr</a>, <a class="el" href="helpers_8h_source.xhtml#l00142">Image::ptr</a>, <a class="el" href="_c_l_2_convolution_8cpp_source.xhtml#l00133">arm_compute::test::validation::src</a>, <a class="el" href="helpers_8h_source.xhtml#l00074">TENSOR3D_DECLARATION</a>, <a class="el" href="fixed__point_8h_source.xhtml#l00093">VEC_DATA_TYPE</a>, <a class="el" href="helpers_8h_source.xhtml#l00060">VECTOR_DECLARATION</a>, <a class="el" href="softmax__layer_8cl_source.xhtml#l00079">VECTOR_SIZE</a>, <a class="el" href="helpers_8h_source.xhtml#l00040">VLOAD</a>, and <a class="el" href="helpers_8h_source.xhtml#l00043">VSTORE</a>.</p>
-<div class="fragment"><div class="line"><a name="l00225"></a><span class="lineno"> 225</span> {</div><div class="line"><a name="l00226"></a><span class="lineno"> 226</span>  <span class="comment">// Compute source and destination addresses</span></div><div class="line"><a name="l00227"></a><span class="lineno"> 227</span>  <a class="code" href="struct_image.xhtml">Image</a> <a class="code" href="namespacearm__compute_1_1test_1_1validation.xhtml#a6743f0a130e8311e6f5b1a23df102472">src</a> = <a class="code" href="helpers_8h.xhtml#aebe814363556c244be043b13e7969197">CONVERT_TO_IMAGE_STRUCT</a>(src);</div><div class="line"><a name="l00228"></a><span class="lineno"> 228</span>  <a class="code" href="struct_image.xhtml">Image</a> <a class="code" href="namespacearm__compute_1_1test_1_1validation.xhtml#adbf67dcee294e673cf796f1ed8aeb6a4">dst</a> = <a class="code" href="helpers_8h.xhtml#aebe814363556c244be043b13e7969197">CONVERT_TO_IMAGE_STRUCT</a>(dst);</div><div class="line"><a name="l00229"></a><span class="lineno"> 229</span> </div><div class="line"><a name="l00230"></a><span class="lineno"> 230</span>  <span class="comment">// Load values from Matrix A</span></div><div class="line"><a name="l00231"></a><span class="lineno"> 231</span>  uchar16 a0 = vload16(0, (__global uchar *)(<a class="code" href="helpers_8h.xhtml#a009469e4d9b8fce3b6d5e97d2077827d">offset</a>(&src, 0, 0)));</div><div class="line"><a name="l00232"></a><span class="lineno"> 232</span>  uchar16 a1 = vload16(0, (__global uchar *)(<a class="code" href="helpers_8h.xhtml#a009469e4d9b8fce3b6d5e97d2077827d">offset</a>(&src, 0, 1)));</div><div class="line"><a name="l00233"></a><span class="lineno"> 233</span>  uchar16 a2 = vload16(0, (__global uchar *)(<a class="code" href="helpers_8h.xhtml#a009469e4d9b8fce3b6d5e97d2077827d">offset</a>(&src, 0, 2)));</div><div class="line"><a name="l00234"></a><span class="lineno"> 234</span>  uchar16 a3 = vload16(0, (__global uchar *)(<a class="code" href="helpers_8h.xhtml#a009469e4d9b8fce3b6d5e97d2077827d">offset</a>(&src, 0, 3)));</div><div class="line"><a name="l00235"></a><span class="lineno"> 235</span> </div><div class="line"><a name="l00236"></a><span class="lineno"> 236</span>  uchar16 val0 = (uchar16)((uchar4)(a0.s0, a1.s0, a2.s0, a3.s0), (uchar4)(a0.s1, a1.s1, a2.s1, a3.s1),</div><div class="line"><a name="l00237"></a><span class="lineno"> 237</span>  (uchar4)(a0.s2, a1.s2, a2.s2, a3.s2), (uchar4)(a0.s3, a1.s3, a2.s3, a3.s3));</div><div class="line"><a name="l00238"></a><span class="lineno"> 238</span>  vstore16(val0, 0, ((__global uchar *)dst.<a class="code" href="struct_image.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a>) + 0);</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>  val0 = (uchar16)((uchar4)(a0.s4, a1.s4, a2.s4, a3.s4), (uchar4)(a0.s5, a1.s5, a2.s5, a3.s5),</div><div class="line"><a name="l00241"></a><span class="lineno"> 241</span>  (uchar4)(a0.s6, a1.s6, a2.s6, a3.s6), (uchar4)(a0.s7, a1.s7, a2.s7, a3.s7));</div><div class="line"><a name="l00242"></a><span class="lineno"> 242</span>  vstore16(val0, 0, ((__global uchar *)dst.<a class="code" href="struct_image.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a>) + 16);</div><div class="line"><a name="l00243"></a><span class="lineno"> 243</span> </div><div class="line"><a name="l00244"></a><span class="lineno"> 244</span>  val0 = (uchar16)((uchar4)(a0.s8, a1.s8, a2.s8, a3.s8), (uchar4)(a0.s9, a1.s9, a2.s9, a3.s9),</div><div class="line"><a name="l00245"></a><span class="lineno"> 245</span>  (uchar4)(a0.sA, a1.sA, a2.sA, a3.sA), (uchar4)(a0.sB, a1.sB, a2.sB, a3.sB));</div><div class="line"><a name="l00246"></a><span class="lineno"> 246</span>  vstore16(val0, 0, ((__global uchar *)dst.<a class="code" href="struct_image.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a>) + 32);</div><div class="line"><a name="l00247"></a><span class="lineno"> 247</span> </div><div class="line"><a name="l00248"></a><span class="lineno"> 248</span>  val0 = (uchar16)((uchar4)(a0.sC, a1.sC, a2.sC, a3.sC), (uchar4)(a0.sD, a1.sD, a2.sD, a3.sD),</div><div class="line"><a name="l00249"></a><span class="lineno"> 249</span>  (uchar4)(a0.sE, a1.sE, a2.sE, a3.sE), (uchar4)(a0.sF, a1.sF, a2.sF, a3.sF));</div><div class="line"><a name="l00250"></a><span class="lineno"> 250</span>  vstore16(val0, 0, ((__global uchar *)dst.<a class="code" href="struct_image.xhtml#acf52c23cbd7424606c10a606524e3e32">ptr</a>) + 48);</div><div class="line"><a name="l00251"></a><span class="lineno"> 251</span> }</div><div class="ttc" id="helpers_8h_xhtml_a009469e4d9b8fce3b6d5e97d2077827d"><div class="ttname"><a href="helpers_8h.xhtml#a009469e4d9b8fce3b6d5e97d2077827d">offset</a></div><div class="ttdeci">__global uchar * offset(const Image *img, int x, int y)</div><div class="ttdoc">Get the pointer position of a Image. </div><div class="ttdef"><b>Definition:</b> <a href="helpers_8h_source.xhtml#l00301">helpers.h:301</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#l00102">helpers.h:102</a></div></div>
-<div class="ttc" id="namespacearm__compute_1_1test_1_1validation_xhtml_adbf67dcee294e673cf796f1ed8aeb6a4"><div class="ttname"><a href="namespacearm__compute_1_1test_1_1validation.xhtml#adbf67dcee294e673cf796f1ed8aeb6a4">arm_compute::test::validation::dst</a></div><div class="ttdeci">CLTensor dst</div><div class="ttdef"><b>Definition:</b> <a href="_c_l_2_convolution_8cpp_source.xhtml#l00123">Convolution.cpp:123</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#l00140">helpers.h:140</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#l00142">helpers.h:142</a></div></div>
-<div class="ttc" id="namespacearm__compute_1_1test_1_1validation_xhtml_a6743f0a130e8311e6f5b1a23df102472"><div class="ttname"><a href="namespacearm__compute_1_1test_1_1validation.xhtml#a6743f0a130e8311e6f5b1a23df102472">arm_compute::test::validation::src</a></div><div class="ttdeci">convolution configure & src</div><div class="ttdef"><b>Definition:</b> <a href="_c_l_2_convolution_8cpp_source.xhtml#l00133">Convolution.cpp:133</a></div></div>
-</div><!-- fragment -->
-</div>
-</div>
-<a class="anchor" id="a97a299ffb620aa98fe5e39bb8ab0833f"></a>
-<div class="memitem">
-<div class="memproto">
- <table class="memname">
- <tr>
- <td class="memname">__kernel void gemm_transpose1x16 </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_offset_first_element_in_bytes</em> </td>
- </tr>
- <tr>
- <td></td>
- <td>)</td>
- <td></td><td></td>
- </tr>
- </table>
-</div><div class="memdoc">
-
-<p>This OpenCL kernel computes the "vector" 1x16 transposition of input matrix. </p>
-<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 matrix. Supported data types: U8/S8/QS8 </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_stride_x</td><td>Stride of the source matrix 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 matrix 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 matrix </td></tr>
- <tr><td class="paramdir">[out]</td><td class="paramname">dst_ptr</td><td>Pointer to the destination matrix Supported data types: 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 matrix in X dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_x</td><td>dst_gx_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 matrix in Y dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_y</td><td>dst_gx_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 matrix </td></tr>
- </table>
- </dd>
-</dl>
-
-<p>Definition at line <a class="el" href="gemm_8cl_source.xhtml#l00109">109</a> of file <a class="el" href="gemm_8cl_source.xhtml">gemm.cl</a>.</p>
-
-<p>References <a class="el" href="helpers_8h_source.xhtml#l00102">CONVERT_TO_IMAGE_STRUCT</a>, <a class="el" href="helpers_8h_source.xhtml#l00142">Image::ptr</a>, and <a class="el" href="_c_l_2_convolution_8cpp_source.xhtml#l00133">arm_compute::test::validation::src</a>.</p>
-<div class="fragment"><div class="line"><a name="l00111"></a><span class="lineno"> 111</span> {</div><div class="line"><a name="l00112"></a><span class="lineno"> 112</span>  uint x = get_global_id(0);</div><div class="line"><a name="l00113"></a><span class="lineno"> 113</span>  uint y = get_global_id(1);</div><div class="line"><a name="l00114"></a><span class="lineno"> 114</span> </div><div class="line"><a name="l00115"></a><span class="lineno"> 115</span>  <span class="comment">// Compute address for Matrix B - source</span></div><div class="line"><a name="l00116"></a><span class="lineno"> 116</span>  <a class="code" href="struct_image.xhtml">Image</a> <a class="code" href="namespacearm__compute_1_1test_1_1validation.xhtml#a6743f0a130e8311e6f5b1a23df102472">src</a> = <a class="code" href="helpers_8h.xhtml#aebe814363556c244be043b13e7969197">CONVERT_TO_IMAGE_STRUCT</a>(src);</div><div class="line"><a name="l00117"></a><span class="lineno"> 117</span> </div><div class="line"><a name="l00118"></a><span class="lineno"> 118</span>  <span class="comment">// Compute address for Matrix B transposed - destination. X and Y are swapped</span></div><div class="line"><a name="l00119"></a><span class="lineno"> 119</span>  uint dst_addr_in_bytes = y * 16 + ((x * dst_stride_y + dst_offset_first_element_in_bytes));</div><div class="line"><a name="l00120"></a><span class="lineno"> 120</span> </div><div class="line"><a name="l00121"></a><span class="lineno"> 121</span>  uchar16 b0 = vload16(0, (__global uchar *)<a class="code" href="namespacearm__compute_1_1test_1_1validation.xhtml#a6743f0a130e8311e6f5b1a23df102472">src</a>.ptr);</div><div class="line"><a name="l00122"></a><span class="lineno"> 122</span> </div><div class="line"><a name="l00123"></a><span class="lineno"> 123</span>  vstore16(b0, 0, (__global uchar *)(dst_ptr + dst_addr_in_bytes));</div><div class="line"><a name="l00124"></a><span class="lineno"> 124</span> }</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#l00102">helpers.h:102</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#l00140">helpers.h:140</a></div></div>
-<div class="ttc" id="namespacearm__compute_1_1test_1_1validation_xhtml_a6743f0a130e8311e6f5b1a23df102472"><div class="ttname"><a href="namespacearm__compute_1_1test_1_1validation.xhtml#a6743f0a130e8311e6f5b1a23df102472">arm_compute::test::validation::src</a></div><div class="ttdeci">convolution configure & src</div><div class="ttdef"><b>Definition:</b> <a href="_c_l_2_convolution_8cpp_source.xhtml#l00133">Convolution.cpp:133</a></div></div>
-</div><!-- fragment -->
-</div>
-</div>
-<a class="anchor" id="ac5eba21b9ac015e1454044aeba93546c"></a>
-<div class="memitem">
-<div class="memproto">
- <table class="memname">
- <tr>
- <td class="memname">__kernel void gemm_transpose1x4 </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_offset_first_element_in_bytes</em> </td>
- </tr>
- <tr>
- <td></td>
- <td>)</td>
- <td></td><td></td>
- </tr>
- </table>
-</div><div class="memdoc">
-
-<p>This OpenCL kernel computes the "vector" 1x4 transposition of input matrix. </p>
-<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 matrix. Supported data types: U32/S32/F32 </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_stride_x</td><td>Stride of the source matrix 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 matrix 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 matrix </td></tr>
- <tr><td class="paramdir">[out]</td><td class="paramname">dst_ptr</td><td>Pointer to the destination matrix Supported data types: 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 matrix in X dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_x</td><td>dst_gx_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 matrix in Y dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_y</td><td>dst_gx_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 matrix </td></tr>
- </table>
- </dd>
-</dl>
-
-<p>Definition at line <a class="el" href="gemm_8cl_source.xhtml#l00045">45</a> of file <a class="el" href="gemm_8cl_source.xhtml">gemm.cl</a>.</p>
-
-<p>References <a class="el" href="helpers_8h_source.xhtml#l00102">CONVERT_TO_IMAGE_STRUCT</a>, <a class="el" href="helpers_8h_source.xhtml#l00142">Image::ptr</a>, and <a class="el" href="_c_l_2_convolution_8cpp_source.xhtml#l00133">arm_compute::test::validation::src</a>.</p>
-<div class="fragment"><div class="line"><a name="l00047"></a><span class="lineno"> 47</span> {</div><div class="line"><a name="l00048"></a><span class="lineno"> 48</span>  uint x = get_global_id(0);</div><div class="line"><a name="l00049"></a><span class="lineno"> 49</span>  uint y = get_global_id(1);</div><div class="line"><a name="l00050"></a><span class="lineno"> 50</span> </div><div class="line"><a name="l00051"></a><span class="lineno"> 51</span>  <span class="comment">// Compute address for Matrix B - source</span></div><div class="line"><a name="l00052"></a><span class="lineno"> 52</span>  <a class="code" href="struct_image.xhtml">Image</a> <a class="code" href="namespacearm__compute_1_1test_1_1validation.xhtml#a6743f0a130e8311e6f5b1a23df102472">src</a> = <a class="code" href="helpers_8h.xhtml#aebe814363556c244be043b13e7969197">CONVERT_TO_IMAGE_STRUCT</a>(src);</div><div class="line"><a name="l00053"></a><span class="lineno"> 53</span> </div><div class="line"><a name="l00054"></a><span class="lineno"> 54</span>  <span class="comment">// Compute address for Matrix B transposed - destination. X and Y are swapped</span></div><div class="line"><a name="l00055"></a><span class="lineno"> 55</span>  uint dst_addr_in_bytes = y * 16 + ((x * dst_stride_y + dst_offset_first_element_in_bytes));</div><div class="line"><a name="l00056"></a><span class="lineno"> 56</span> </div><div class="line"><a name="l00057"></a><span class="lineno"> 57</span>  uint4 b0 = vload4(0, (__global uint *)<a class="code" href="namespacearm__compute_1_1test_1_1validation.xhtml#a6743f0a130e8311e6f5b1a23df102472">src</a>.ptr);</div><div class="line"><a name="l00058"></a><span class="lineno"> 58</span> </div><div class="line"><a name="l00059"></a><span class="lineno"> 59</span>  vstore4(b0, 0, (__global uint *)(dst_ptr + dst_addr_in_bytes));</div><div class="line"><a name="l00060"></a><span class="lineno"> 60</span> }</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#l00102">helpers.h:102</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#l00140">helpers.h:140</a></div></div>
-<div class="ttc" id="namespacearm__compute_1_1test_1_1validation_xhtml_a6743f0a130e8311e6f5b1a23df102472"><div class="ttname"><a href="namespacearm__compute_1_1test_1_1validation.xhtml#a6743f0a130e8311e6f5b1a23df102472">arm_compute::test::validation::src</a></div><div class="ttdeci">convolution configure & src</div><div class="ttdef"><b>Definition:</b> <a href="_c_l_2_convolution_8cpp_source.xhtml#l00133">Convolution.cpp:133</a></div></div>
-</div><!-- fragment -->
-</div>
-</div>
-<a class="anchor" id="a75c35ac120f123e9cce56c43ba3f48d2"></a>
-<div class="memitem">
-<div class="memproto">
- <table class="memname">
- <tr>
- <td class="memname">__kernel void gemm_transpose1x8 </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_offset_first_element_in_bytes</em> </td>
- </tr>
- <tr>
- <td></td>
- <td>)</td>
- <td></td><td></td>
- </tr>
- </table>
-</div><div class="memdoc">
-
-<p>This OpenCL kernel computes the "vector" 1x8 transposition of input matrix. </p>
-<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 matrix. Supported data types: U16/S16/QS16/F16 </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">src_stride_x</td><td>Stride of the source matrix 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 matrix 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 matrix </td></tr>
- <tr><td class="paramdir">[out]</td><td class="paramname">dst_ptr</td><td>Pointer to the destination matrix Supported data types: 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 matrix in X dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_x</td><td>dst_gx_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 matrix in Y dimension (in bytes) </td></tr>
- <tr><td class="paramdir">[in]</td><td class="paramname">dst_step_y</td><td>dst_gx_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 matrix </td></tr>
- </table>
- </dd>
-</dl>
-
-<p>Definition at line <a class="el" href="gemm_8cl_source.xhtml#l00077">77</a> of file <a class="el" href="gemm_8cl_source.xhtml">gemm.cl</a>.</p>
-
-<p>References <a class="el" href="helpers_8h_source.xhtml#l00102">CONVERT_TO_IMAGE_STRUCT</a>, <a class="el" href="helpers_8h_source.xhtml#l00142">Image::ptr</a>, and <a class="el" href="_c_l_2_convolution_8cpp_source.xhtml#l00133">arm_compute::test::validation::src</a>.</p>
-<div class="fragment"><div class="line"><a name="l00079"></a><span class="lineno"> 79</span> {</div><div class="line"><a name="l00080"></a><span class="lineno"> 80</span>  uint x = get_global_id(0);</div><div class="line"><a name="l00081"></a><span class="lineno"> 81</span>  uint y = get_global_id(1);</div><div class="line"><a name="l00082"></a><span class="lineno"> 82</span> </div><div class="line"><a name="l00083"></a><span class="lineno"> 83</span>  <span class="comment">// Compute address for Matrix B - source</span></div><div class="line"><a name="l00084"></a><span class="lineno"> 84</span>  <a class="code" href="struct_image.xhtml">Image</a> <a class="code" href="namespacearm__compute_1_1test_1_1validation.xhtml#a6743f0a130e8311e6f5b1a23df102472">src</a> = <a class="code" href="helpers_8h.xhtml#aebe814363556c244be043b13e7969197">CONVERT_TO_IMAGE_STRUCT</a>(src);</div><div class="line"><a name="l00085"></a><span class="lineno"> 85</span> </div><div class="line"><a name="l00086"></a><span class="lineno"> 86</span>  <span class="comment">// Compute address for Matrix B transposed - destination. X and Y are swapped</span></div><div class="line"><a name="l00087"></a><span class="lineno"> 87</span>  uint dst_addr_in_bytes = y * 16 + ((x * dst_stride_y + dst_offset_first_element_in_bytes));</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>  ushort8 b0 = vload8(0, (__global ushort *)<a class="code" href="namespacearm__compute_1_1test_1_1validation.xhtml#a6743f0a130e8311e6f5b1a23df102472">src</a>.ptr);</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>  vstore8(b0, 0, (__global ushort *)(dst_ptr + dst_addr_in_bytes));</div><div class="line"><a name="l00092"></a><span class="lineno"> 92</span> }</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#l00102">helpers.h:102</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#l00140">helpers.h:140</a></div></div>
-<div class="ttc" id="namespacearm__compute_1_1test_1_1validation_xhtml_a6743f0a130e8311e6f5b1a23df102472"><div class="ttname"><a href="namespacearm__compute_1_1test_1_1validation.xhtml#a6743f0a130e8311e6f5b1a23df102472">arm_compute::test::validation::src</a></div><div class="ttdeci">convolution configure & src</div><div class="ttdef"><b>Definition:</b> <a href="_c_l_2_convolution_8cpp_source.xhtml#l00133">Convolution.cpp:133</a></div></div>
-</div><!-- fragment -->
-</div>
-</div>
</div><!-- contents -->
</div><!-- doc-content -->
<!-- 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="gemm_8cl.xhtml">gemm.cl</a></li>
- <li class="footer">Generated on Wed Jan 24 2018 14:30:43 for Compute Library by
+ <li class="footer">Generated on Thu Feb 22 2018 15:45:22 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>
</ul>