| <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"> |
| <html xmlns="http://www.w3.org/1999/xhtml"> |
| <head> |
| <meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/> |
| <title>CMSIS DSP Software Library: arm_mat_mult_q15.c Source File</title> |
| <link href="tabs.css" rel="stylesheet" type="text/css"/> |
| <link href="search/search.css" rel="stylesheet" type="text/css"/> |
| <script type="text/javaScript" src="search/search.js"></script> |
| <link href="doxygen.css" rel="stylesheet" type="text/css"/> |
| </head> |
| <body onload='searchBox.OnSelectItem(0);'> |
| <!-- Generated by Doxygen 1.7.2 --> |
| <script type="text/javascript"><!-- |
| var searchBox = new SearchBox("searchBox", "search",false,'Search'); |
| --></script> |
| <div class="navigation" id="top"> |
| <div class="tabs"> |
| <ul class="tablist"> |
| <li><a href="index.html"><span>Main Page</span></a></li> |
| <li><a href="modules.html"><span>Modules</span></a></li> |
| <li><a href="annotated.html"><span>Data Structures</span></a></li> |
| <li class="current"><a href="files.html"><span>Files</span></a></li> |
| <li><a href="examples.html"><span>Examples</span></a></li> |
| <li id="searchli"> |
| <div id="MSearchBox" class="MSearchBoxInactive"> |
| <span class="left"> |
| <img id="MSearchSelect" src="search/mag_sel.png" |
| onmouseover="return searchBox.OnSearchSelectShow()" |
| onmouseout="return searchBox.OnSearchSelectHide()" |
| alt=""/> |
| <input type="text" id="MSearchField" value="Search" accesskey="S" |
| onfocus="searchBox.OnSearchFieldFocus(true)" |
| onblur="searchBox.OnSearchFieldFocus(false)" |
| onkeyup="searchBox.OnSearchFieldChange(event)"/> |
| </span><span class="right"> |
| <a id="MSearchClose" href="javascript:searchBox.CloseResultsWindow()"><img id="MSearchCloseImg" border="0" src="search/close.png" alt=""/></a> |
| </span> |
| </div> |
| </li> |
| </ul> |
| </div> |
| <div class="tabs2"> |
| <ul class="tablist"> |
| <li><a href="files.html"><span>File List</span></a></li> |
| <li><a href="globals.html"><span>Globals</span></a></li> |
| </ul> |
| </div> |
| <div class="header"> |
| <div class="headertitle"> |
| <h1>arm_mat_mult_q15.c</h1> </div> |
| </div> |
| <div class="contents"> |
| <a href="arm__mat__mult__q15_8c.html">Go to the documentation of this file.</a><div class="fragment"><pre class="fragment"><a name="l00001"></a>00001 <span class="comment">/* ---------------------------------------------------------------------- </span> |
| <a name="l00002"></a>00002 <span class="comment">* Copyright (C) 2010 ARM Limited. All rights reserved. </span> |
| <a name="l00003"></a>00003 <span class="comment">* </span> |
| <a name="l00004"></a>00004 <span class="comment">* $Date: 15. July 2011 </span> |
| <a name="l00005"></a>00005 <span class="comment">* $Revision: V1.0.10 </span> |
| <a name="l00006"></a>00006 <span class="comment">* </span> |
| <a name="l00007"></a>00007 <span class="comment">* Project: CMSIS DSP Library </span> |
| <a name="l00008"></a>00008 <span class="comment">* Title: arm_mat_mult_q15.c </span> |
| <a name="l00009"></a>00009 <span class="comment">* </span> |
| <a name="l00010"></a>00010 <span class="comment">* Description: Q15 matrix multiplication. </span> |
| <a name="l00011"></a>00011 <span class="comment">* </span> |
| <a name="l00012"></a>00012 <span class="comment">* Target Processor: Cortex-M4/Cortex-M3/Cortex-M0</span> |
| <a name="l00013"></a>00013 <span class="comment">* </span> |
| <a name="l00014"></a>00014 <span class="comment">* Version 1.0.10 2011/7/15 </span> |
| <a name="l00015"></a>00015 <span class="comment">* Big Endian support added and Merged M0 and M3/M4 Source code. </span> |
| <a name="l00016"></a>00016 <span class="comment">* </span> |
| <a name="l00017"></a>00017 <span class="comment">* Version 1.0.3 2010/11/29 </span> |
| <a name="l00018"></a>00018 <span class="comment">* Re-organized the CMSIS folders and updated documentation. </span> |
| <a name="l00019"></a>00019 <span class="comment">* </span> |
| <a name="l00020"></a>00020 <span class="comment">* Version 1.0.2 2010/11/11 </span> |
| <a name="l00021"></a>00021 <span class="comment">* Documentation updated. </span> |
| <a name="l00022"></a>00022 <span class="comment">* </span> |
| <a name="l00023"></a>00023 <span class="comment">* Version 1.0.1 2010/10/05 </span> |
| <a name="l00024"></a>00024 <span class="comment">* Production release and review comments incorporated. </span> |
| <a name="l00025"></a>00025 <span class="comment">* </span> |
| <a name="l00026"></a>00026 <span class="comment">* Version 1.0.0 2010/09/20 </span> |
| <a name="l00027"></a>00027 <span class="comment">* Production release and review comments incorporated. </span> |
| <a name="l00028"></a>00028 <span class="comment">* </span> |
| <a name="l00029"></a>00029 <span class="comment">* Version 0.0.5 2010/04/26 </span> |
| <a name="l00030"></a>00030 <span class="comment">* incorporated review comments and updated with latest CMSIS layer </span> |
| <a name="l00031"></a>00031 <span class="comment">* </span> |
| <a name="l00032"></a>00032 <span class="comment">* Version 0.0.3 2010/03/10 </span> |
| <a name="l00033"></a>00033 <span class="comment">* Initial version </span> |
| <a name="l00034"></a>00034 <span class="comment">* -------------------------------------------------------------------- */</span> |
| <a name="l00035"></a>00035 |
| <a name="l00036"></a>00036 <span class="preprocessor">#include "<a class="code" href="arm__math_8h.html">arm_math.h</a>"</span> |
| <a name="l00037"></a>00037 |
| <a name="l00074"></a><a class="code" href="group___matrix_mult.html#ga7521d59196189bb6dde26e8cdfb66e21">00074</a> <a class="code" href="arm__math_8h.html#a5e459c6409dfcd2927bb8a57491d7cf6" title="Error status returned by some functions in the library.">arm_status</a> <a class="code" href="group___matrix_mult.html#ga7521d59196189bb6dde26e8cdfb66e21" title="Q15 matrix multiplication.">arm_mat_mult_q15</a>( |
| <a name="l00075"></a>00075 <span class="keyword">const</span> <a class="code" href="structarm__matrix__instance__q15.html" title="Instance structure for the Q15 matrix structure.">arm_matrix_instance_q15</a> * pSrcA, |
| <a name="l00076"></a>00076 <span class="keyword">const</span> <a class="code" href="structarm__matrix__instance__q15.html" title="Instance structure for the Q15 matrix structure.">arm_matrix_instance_q15</a> * pSrcB, |
| <a name="l00077"></a>00077 <a class="code" href="structarm__matrix__instance__q15.html" title="Instance structure for the Q15 matrix structure.">arm_matrix_instance_q15</a> * pDst, |
| <a name="l00078"></a>00078 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> * pState) |
| <a name="l00079"></a>00079 { |
| <a name="l00080"></a>00080 <a class="code" href="arm__math_8h.html#a5aea1cb12fc02d9d44c8abf217eaa5c6" title="64-bit fractional data type in 1.63 format.">q63_t</a> sum; <span class="comment">/* accumulator */</span> |
| <a name="l00081"></a>00081 |
| <a name="l00082"></a>00082 <span class="preprocessor">#ifndef ARM_MATH_CM0</span> |
| <a name="l00083"></a>00083 <span class="preprocessor"></span> |
| <a name="l00084"></a>00084 <span class="comment">/* Run the below code for Cortex-M4 and Cortex-M3 */</span> |
| <a name="l00085"></a>00085 |
| <a name="l00086"></a>00086 <a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> in; <span class="comment">/* Temporary variable to hold the input value */</span> |
| <a name="l00087"></a>00087 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *pSrcBT = pState; <span class="comment">/* input data matrix pointer for transpose */</span> |
| <a name="l00088"></a>00088 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *pInA = pSrcA-><a class="code" href="structarm__matrix__instance__q15.html#a6da33a5553e634787d0f515cf8d724af">pData</a>; <span class="comment">/* input data matrix pointer A of Q15 type */</span> |
| <a name="l00089"></a>00089 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *pInB = pSrcB-><a class="code" href="structarm__matrix__instance__q15.html#a6da33a5553e634787d0f515cf8d724af">pData</a>; <span class="comment">/* input data matrix pointer B of Q15 type */</span> |
| <a name="l00090"></a>00090 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *px; <span class="comment">/* Temporary output data matrix pointer */</span> |
| <a name="l00091"></a>00091 uint16_t numRowsA = pSrcA-><a class="code" href="structarm__matrix__instance__q15.html#a9bac6ed54be287c4d4f01a1a28be65f5">numRows</a>; <span class="comment">/* number of rows of input matrix A */</span> |
| <a name="l00092"></a>00092 uint16_t numColsB = pSrcB-><a class="code" href="structarm__matrix__instance__q15.html#acbbce67ba058d8e1c867c71d57288c97">numCols</a>; <span class="comment">/* number of columns of input matrix B */</span> |
| <a name="l00093"></a>00093 uint16_t numColsA = pSrcA-><a class="code" href="structarm__matrix__instance__q15.html#acbbce67ba058d8e1c867c71d57288c97">numCols</a>; <span class="comment">/* number of columns of input matrix A */</span> |
| <a name="l00094"></a>00094 uint16_t numRowsB = pSrcB-><a class="code" href="structarm__matrix__instance__q15.html#a9bac6ed54be287c4d4f01a1a28be65f5">numRows</a>; <span class="comment">/* number of rows of input matrix A */</span> |
| <a name="l00095"></a>00095 uint16_t col, i = 0u, row = numRowsB, colCnt; <span class="comment">/* loop counters */</span> |
| <a name="l00096"></a>00096 <a class="code" href="arm__math_8h.html#a5e459c6409dfcd2927bb8a57491d7cf6" title="Error status returned by some functions in the library.">arm_status</a> <a class="code" href="arm__dotproduct__example__f32_8c.html#a88ccb294236ab22b00310c47164c53c3">status</a>; <span class="comment">/* status of matrix multiplication */</span> |
| <a name="l00097"></a>00097 |
| <a name="l00098"></a>00098 <span class="preprocessor">#ifdef ARM_MATH_MATRIX_CHECK</span> |
| <a name="l00099"></a>00099 <span class="preprocessor"></span> |
| <a name="l00100"></a>00100 |
| <a name="l00101"></a>00101 <span class="comment">/* Check for matrix mismatch condition */</span> |
| <a name="l00102"></a>00102 |
| <a name="l00103"></a>00103 <span class="keywordflow">if</span>((pSrcA-><a class="code" href="structarm__matrix__instance__q15.html#acbbce67ba058d8e1c867c71d57288c97">numCols</a> != pSrcB-><a class="code" href="structarm__matrix__instance__q15.html#a9bac6ed54be287c4d4f01a1a28be65f5">numRows</a>) || |
| <a name="l00104"></a>00104 (pSrcA-><a class="code" href="structarm__matrix__instance__q15.html#a9bac6ed54be287c4d4f01a1a28be65f5">numRows</a> != pDst-><a class="code" href="structarm__matrix__instance__q15.html#a9bac6ed54be287c4d4f01a1a28be65f5">numRows</a>) || (pSrcB-><a class="code" href="structarm__matrix__instance__q15.html#acbbce67ba058d8e1c867c71d57288c97">numCols</a> != pDst-><a class="code" href="structarm__matrix__instance__q15.html#acbbce67ba058d8e1c867c71d57288c97">numCols</a>)) |
| <a name="l00105"></a>00105 { |
| <a name="l00106"></a>00106 <span class="comment">/* Set status as ARM_MATH_SIZE_MISMATCH */</span> |
| <a name="l00107"></a>00107 status = <a class="code" href="arm__math_8h.html#a5e459c6409dfcd2927bb8a57491d7cf6a7071b92f1f6bc3c5c312a237ea91105b">ARM_MATH_SIZE_MISMATCH</a>; |
| <a name="l00108"></a>00108 } |
| <a name="l00109"></a>00109 <span class="keywordflow">else</span> |
| <a name="l00110"></a>00110 <span class="preprocessor">#endif </span><span class="comment">/* #ifdef ARM_MATH_MATRIX_CHECK */</span> |
| <a name="l00111"></a>00111 |
| <a name="l00112"></a>00112 { |
| <a name="l00113"></a>00113 <span class="comment">/* Matrix transpose */</span> |
| <a name="l00114"></a>00114 <span class="keywordflow">do</span> |
| <a name="l00115"></a>00115 { |
| <a name="l00116"></a>00116 <span class="comment">/* Apply loop unrolling and exchange the columns with row elements */</span> |
| <a name="l00117"></a>00117 col = numColsB >> 2; |
| <a name="l00118"></a>00118 |
| <a name="l00119"></a>00119 <span class="comment">/* The pointer px is set to starting address of the column being processed */</span> |
| <a name="l00120"></a>00120 px = pSrcBT + i; |
| <a name="l00121"></a>00121 |
| <a name="l00122"></a>00122 <span class="comment">/* First part of the processing with loop unrolling. Compute 4 outputs at a time. </span> |
| <a name="l00123"></a>00123 <span class="comment"> ** a second loop below computes the remaining 1 to 3 samples. */</span> |
| <a name="l00124"></a>00124 <span class="keywordflow">while</span>(col > 0u) |
| <a name="l00125"></a>00125 { |
| <a name="l00126"></a>00126 <span class="comment">/* Read two elements from the row */</span> |
| <a name="l00127"></a>00127 in = *<a class="code" href="arm__math_8h.html#a9de2e0a5785be82866bcb96012282248" title="definition to read/write two 16 bit values.">__SIMD32</a>(pInB)++; |
| <a name="l00128"></a>00128 |
| <a name="l00129"></a>00129 <span class="comment">/* Unpack and store one element in the destination */</span> |
| <a name="l00130"></a>00130 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l00131"></a>00131 <span class="preprocessor"></span> |
| <a name="l00132"></a>00132 *px = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) in; |
| <a name="l00133"></a>00133 |
| <a name="l00134"></a>00134 <span class="preprocessor">#else</span> |
| <a name="l00135"></a>00135 <span class="preprocessor"></span> |
| <a name="l00136"></a>00136 *px = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) ((in & (<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) 0xffff0000) >> 16); |
| <a name="l00137"></a>00137 |
| <a name="l00138"></a>00138 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l00139"></a>00139 |
| <a name="l00140"></a>00140 <span class="comment">/* Update the pointer px to point to the next row of the transposed matrix */</span> |
| <a name="l00141"></a>00141 px += numRowsB; |
| <a name="l00142"></a>00142 |
| <a name="l00143"></a>00143 <span class="comment">/* Unpack and store the second element in the destination */</span> |
| <a name="l00144"></a>00144 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l00145"></a>00145 <span class="preprocessor"></span> |
| <a name="l00146"></a>00146 *px = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) ((in & (<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) 0xffff0000) >> 16); |
| <a name="l00147"></a>00147 |
| <a name="l00148"></a>00148 <span class="preprocessor">#else</span> |
| <a name="l00149"></a>00149 <span class="preprocessor"></span> |
| <a name="l00150"></a>00150 *px = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) in; |
| <a name="l00151"></a>00151 |
| <a name="l00152"></a>00152 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l00153"></a>00153 |
| <a name="l00154"></a>00154 |
| <a name="l00155"></a>00155 <span class="comment">/* Update the pointer px to point to the next row of the transposed matrix */</span> |
| <a name="l00156"></a>00156 px += numRowsB; |
| <a name="l00157"></a>00157 |
| <a name="l00158"></a>00158 <span class="comment">/* Read two elements from the row */</span> |
| <a name="l00159"></a>00159 in = *<a class="code" href="arm__math_8h.html#a9de2e0a5785be82866bcb96012282248" title="definition to read/write two 16 bit values.">__SIMD32</a>(pInB)++; |
| <a name="l00160"></a>00160 |
| <a name="l00161"></a>00161 <span class="comment">/* Unpack and store one element in the destination */</span> |
| <a name="l00162"></a>00162 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l00163"></a>00163 <span class="preprocessor"></span> |
| <a name="l00164"></a>00164 *px = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) in; |
| <a name="l00165"></a>00165 |
| <a name="l00166"></a>00166 <span class="preprocessor">#else</span> |
| <a name="l00167"></a>00167 <span class="preprocessor"></span> |
| <a name="l00168"></a>00168 *px = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) ((in & (<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) 0xffff0000) >> 16); |
| <a name="l00169"></a>00169 |
| <a name="l00170"></a>00170 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l00171"></a>00171 |
| <a name="l00172"></a>00172 <span class="comment">/* Update the pointer px to point to the next row of the transposed matrix */</span> |
| <a name="l00173"></a>00173 px += numRowsB; |
| <a name="l00174"></a>00174 |
| <a name="l00175"></a>00175 <span class="comment">/* Unpack and store the second element in the destination */</span> |
| <a name="l00176"></a>00176 |
| <a name="l00177"></a>00177 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l00178"></a>00178 <span class="preprocessor"></span> |
| <a name="l00179"></a>00179 *px = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) ((in & (<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) 0xffff0000) >> 16); |
| <a name="l00180"></a>00180 |
| <a name="l00181"></a>00181 <span class="preprocessor">#else</span> |
| <a name="l00182"></a>00182 <span class="preprocessor"></span> |
| <a name="l00183"></a>00183 *px = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) in; |
| <a name="l00184"></a>00184 |
| <a name="l00185"></a>00185 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l00186"></a>00186 |
| <a name="l00187"></a>00187 <span class="comment">/* Update the pointer px to point to the next row of the transposed matrix */</span> |
| <a name="l00188"></a>00188 px += numRowsB; |
| <a name="l00189"></a>00189 |
| <a name="l00190"></a>00190 <span class="comment">/* Decrement the column loop counter */</span> |
| <a name="l00191"></a>00191 col--; |
| <a name="l00192"></a>00192 } |
| <a name="l00193"></a>00193 |
| <a name="l00194"></a>00194 <span class="comment">/* If the columns of pSrcB is not a multiple of 4, compute any remaining output samples here. </span> |
| <a name="l00195"></a>00195 <span class="comment"> ** No loop unrolling is used. */</span> |
| <a name="l00196"></a>00196 col = numColsB % 0x4u; |
| <a name="l00197"></a>00197 |
| <a name="l00198"></a>00198 <span class="keywordflow">while</span>(col > 0u) |
| <a name="l00199"></a>00199 { |
| <a name="l00200"></a>00200 <span class="comment">/* Read and store the input element in the destination */</span> |
| <a name="l00201"></a>00201 *px = *pInB++; |
| <a name="l00202"></a>00202 |
| <a name="l00203"></a>00203 <span class="comment">/* Update the pointer px to point to the next row of the transposed matrix */</span> |
| <a name="l00204"></a>00204 px += numRowsB; |
| <a name="l00205"></a>00205 |
| <a name="l00206"></a>00206 <span class="comment">/* Decrement the column loop counter */</span> |
| <a name="l00207"></a>00207 col--; |
| <a name="l00208"></a>00208 } |
| <a name="l00209"></a>00209 |
| <a name="l00210"></a>00210 i++; |
| <a name="l00211"></a>00211 |
| <a name="l00212"></a>00212 <span class="comment">/* Decrement the row loop counter */</span> |
| <a name="l00213"></a>00213 row--; |
| <a name="l00214"></a>00214 |
| <a name="l00215"></a>00215 } <span class="keywordflow">while</span>(row > 0u); |
| <a name="l00216"></a>00216 |
| <a name="l00217"></a>00217 <span class="comment">/* Reset the variables for the usage in the following multiplication process */</span> |
| <a name="l00218"></a>00218 row = numRowsA; |
| <a name="l00219"></a>00219 i = 0u; |
| <a name="l00220"></a>00220 px = pDst-><a class="code" href="structarm__matrix__instance__q15.html#a6da33a5553e634787d0f515cf8d724af">pData</a>; |
| <a name="l00221"></a>00221 |
| <a name="l00222"></a>00222 <span class="comment">/* The following loop performs the dot-product of each row in pSrcA with each column in pSrcB */</span> |
| <a name="l00223"></a>00223 <span class="comment">/* row loop */</span> |
| <a name="l00224"></a>00224 <span class="keywordflow">do</span> |
| <a name="l00225"></a>00225 { |
| <a name="l00226"></a>00226 <span class="comment">/* For every row wise process, the column loop counter is to be initiated */</span> |
| <a name="l00227"></a>00227 col = numColsB; |
| <a name="l00228"></a>00228 |
| <a name="l00229"></a>00229 <span class="comment">/* For every row wise process, the pIn2 pointer is set </span> |
| <a name="l00230"></a>00230 <span class="comment"> ** to the starting address of the transposed pSrcB data */</span> |
| <a name="l00231"></a>00231 pInB = pSrcBT; |
| <a name="l00232"></a>00232 |
| <a name="l00233"></a>00233 <span class="comment">/* column loop */</span> |
| <a name="l00234"></a>00234 <span class="keywordflow">do</span> |
| <a name="l00235"></a>00235 { |
| <a name="l00236"></a>00236 <span class="comment">/* Set the variable sum, that acts as accumulator, to zero */</span> |
| <a name="l00237"></a>00237 sum = 0; |
| <a name="l00238"></a>00238 |
| <a name="l00239"></a>00239 <span class="comment">/* Apply loop unrolling and compute 2 MACs simultaneously. */</span> |
| <a name="l00240"></a>00240 colCnt = numColsA >> 1; |
| <a name="l00241"></a>00241 |
| <a name="l00242"></a>00242 <span class="comment">/* Initiate the pointer pIn1 to point to the starting address of the column being processed */</span> |
| <a name="l00243"></a>00243 pInA = pSrcA-><a class="code" href="structarm__matrix__instance__q15.html#a6da33a5553e634787d0f515cf8d724af">pData</a> + i; |
| <a name="l00244"></a>00244 |
| <a name="l00245"></a>00245 <span class="comment">/* matrix multiplication */</span> |
| <a name="l00246"></a>00246 <span class="keywordflow">while</span>(colCnt > 0u) |
| <a name="l00247"></a>00247 { |
| <a name="l00248"></a>00248 <span class="comment">/* c(m,n) = a(1,1)*b(1,1) + a(1,2) * b(2,1) + .... + a(m,p)*b(p,n) */</span> |
| <a name="l00249"></a>00249 sum = __SMLALD(*<a class="code" href="arm__math_8h.html#a9de2e0a5785be82866bcb96012282248" title="definition to read/write two 16 bit values.">__SIMD32</a>(pInA)++, *<a class="code" href="arm__math_8h.html#a9de2e0a5785be82866bcb96012282248" title="definition to read/write two 16 bit values.">__SIMD32</a>(pInB)++, sum); |
| <a name="l00250"></a>00250 |
| <a name="l00251"></a>00251 <span class="comment">/* Decrement the loop counter */</span> |
| <a name="l00252"></a>00252 colCnt--; |
| <a name="l00253"></a>00253 } |
| <a name="l00254"></a>00254 |
| <a name="l00255"></a>00255 <span class="comment">/* process odd column samples */</span> |
| <a name="l00256"></a>00256 <span class="keywordflow">if</span>((numColsA & 0x1u) > 0u) |
| <a name="l00257"></a>00257 { |
| <a name="l00258"></a>00258 <span class="comment">/* c(m,n) = a(1,1)*b(1,1) + a(1,2) * b(2,1) + .... + a(m,p)*b(p,n) */</span> |
| <a name="l00259"></a>00259 sum += ((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) * pInA * (*pInB++)); |
| <a name="l00260"></a>00260 } |
| <a name="l00261"></a>00261 |
| <a name="l00262"></a>00262 <span class="comment">/* Saturate and store the result in the destination buffer */</span> |
| <a name="l00263"></a>00263 *px = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) (__SSAT((sum >> 15), 16)); |
| <a name="l00264"></a>00264 px++; |
| <a name="l00265"></a>00265 |
| <a name="l00266"></a>00266 <span class="comment">/* Decrement the column loop counter */</span> |
| <a name="l00267"></a>00267 col--; |
| <a name="l00268"></a>00268 |
| <a name="l00269"></a>00269 } <span class="keywordflow">while</span>(col > 0u); |
| <a name="l00270"></a>00270 |
| <a name="l00271"></a>00271 i = i + numColsA; |
| <a name="l00272"></a>00272 |
| <a name="l00273"></a>00273 <span class="comment">/* Decrement the row loop counter */</span> |
| <a name="l00274"></a>00274 row--; |
| <a name="l00275"></a>00275 |
| <a name="l00276"></a>00276 } <span class="keywordflow">while</span>(row > 0u); |
| <a name="l00277"></a>00277 |
| <a name="l00278"></a>00278 <span class="preprocessor">#else</span> |
| <a name="l00279"></a>00279 <span class="preprocessor"></span> |
| <a name="l00280"></a>00280 <span class="comment">/* Run the below code for Cortex-M0 */</span> |
| <a name="l00281"></a>00281 |
| <a name="l00282"></a>00282 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *pIn1 = pSrcA-><a class="code" href="structarm__matrix__instance__q15.html#a6da33a5553e634787d0f515cf8d724af">pData</a>; <span class="comment">/* input data matrix pointer A */</span> |
| <a name="l00283"></a>00283 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *pIn2 = pSrcB-><a class="code" href="structarm__matrix__instance__q15.html#a6da33a5553e634787d0f515cf8d724af">pData</a>; <span class="comment">/* input data matrix pointer B */</span> |
| <a name="l00284"></a>00284 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *pInA = pSrcA-><a class="code" href="structarm__matrix__instance__q15.html#a6da33a5553e634787d0f515cf8d724af">pData</a>; <span class="comment">/* input data matrix pointer A of Q15 type */</span> |
| <a name="l00285"></a>00285 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *pInB = pSrcB-><a class="code" href="structarm__matrix__instance__q15.html#a6da33a5553e634787d0f515cf8d724af">pData</a>; <span class="comment">/* input data matrix pointer B of Q15 type */</span> |
| <a name="l00286"></a>00286 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *pOut = pDst-><a class="code" href="structarm__matrix__instance__q15.html#a6da33a5553e634787d0f515cf8d724af">pData</a>; <span class="comment">/* output data matrix pointer */</span> |
| <a name="l00287"></a>00287 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *px; <span class="comment">/* Temporary output data matrix pointer */</span> |
| <a name="l00288"></a>00288 uint16_t numColsB = pSrcB-><a class="code" href="structarm__matrix__instance__q15.html#acbbce67ba058d8e1c867c71d57288c97">numCols</a>; <span class="comment">/* number of columns of input matrix B */</span> |
| <a name="l00289"></a>00289 uint16_t numColsA = pSrcA-><a class="code" href="structarm__matrix__instance__q15.html#acbbce67ba058d8e1c867c71d57288c97">numCols</a>; <span class="comment">/* number of columns of input matrix A */</span> |
| <a name="l00290"></a>00290 uint16_t numRowsA = pSrcA-><a class="code" href="structarm__matrix__instance__q15.html#a9bac6ed54be287c4d4f01a1a28be65f5">numRows</a>; <span class="comment">/* number of rows of input matrix A */</span> |
| <a name="l00291"></a>00291 uint16_t col, i = 0u, row = numRowsA, colCnt; <span class="comment">/* loop counters */</span> |
| <a name="l00292"></a>00292 <a class="code" href="arm__math_8h.html#a5e459c6409dfcd2927bb8a57491d7cf6" title="Error status returned by some functions in the library.">arm_status</a> <a class="code" href="arm__dotproduct__example__f32_8c.html#a88ccb294236ab22b00310c47164c53c3">status</a>; <span class="comment">/* status of matrix multiplication */</span> |
| <a name="l00293"></a>00293 |
| <a name="l00294"></a>00294 <span class="preprocessor">#ifdef ARM_MATH_MATRIX_CHECK</span> |
| <a name="l00295"></a>00295 <span class="preprocessor"></span> |
| <a name="l00296"></a>00296 <span class="comment">/* Check for matrix mismatch condition */</span> |
| <a name="l00297"></a>00297 <span class="keywordflow">if</span>((pSrcA-><a class="code" href="structarm__matrix__instance__q15.html#acbbce67ba058d8e1c867c71d57288c97">numCols</a> != pSrcB-><a class="code" href="structarm__matrix__instance__q15.html#a9bac6ed54be287c4d4f01a1a28be65f5">numRows</a>) || |
| <a name="l00298"></a>00298 (pSrcA-><a class="code" href="structarm__matrix__instance__q15.html#a9bac6ed54be287c4d4f01a1a28be65f5">numRows</a> != pDst-><a class="code" href="structarm__matrix__instance__q15.html#a9bac6ed54be287c4d4f01a1a28be65f5">numRows</a>) || (pSrcB-><a class="code" href="structarm__matrix__instance__q15.html#acbbce67ba058d8e1c867c71d57288c97">numCols</a> != pDst-><a class="code" href="structarm__matrix__instance__q15.html#acbbce67ba058d8e1c867c71d57288c97">numCols</a>)) |
| <a name="l00299"></a>00299 { |
| <a name="l00300"></a>00300 <span class="comment">/* Set status as ARM_MATH_SIZE_MISMATCH */</span> |
| <a name="l00301"></a>00301 status = <a class="code" href="arm__math_8h.html#a5e459c6409dfcd2927bb8a57491d7cf6a7071b92f1f6bc3c5c312a237ea91105b">ARM_MATH_SIZE_MISMATCH</a>; |
| <a name="l00302"></a>00302 } |
| <a name="l00303"></a>00303 <span class="keywordflow">else</span> |
| <a name="l00304"></a>00304 <span class="preprocessor">#endif </span><span class="comment">/* #ifdef ARM_MATH_MATRIX_CHECK */</span> |
| <a name="l00305"></a>00305 |
| <a name="l00306"></a>00306 { |
| <a name="l00307"></a>00307 <span class="comment">/* The following loop performs the dot-product of each row in pSrcA with each column in pSrcB */</span> |
| <a name="l00308"></a>00308 <span class="comment">/* row loop */</span> |
| <a name="l00309"></a>00309 <span class="keywordflow">do</span> |
| <a name="l00310"></a>00310 { |
| <a name="l00311"></a>00311 <span class="comment">/* Output pointer is set to starting address of the row being processed */</span> |
| <a name="l00312"></a>00312 px = pOut + i; |
| <a name="l00313"></a>00313 |
| <a name="l00314"></a>00314 <span class="comment">/* For every row wise process, the column loop counter is to be initiated */</span> |
| <a name="l00315"></a>00315 col = numColsB; |
| <a name="l00316"></a>00316 |
| <a name="l00317"></a>00317 <span class="comment">/* For every row wise process, the pIn2 pointer is set </span> |
| <a name="l00318"></a>00318 <span class="comment"> ** to the starting address of the pSrcB data */</span> |
| <a name="l00319"></a>00319 pIn2 = pSrcB-><a class="code" href="structarm__matrix__instance__q15.html#a6da33a5553e634787d0f515cf8d724af">pData</a>; |
| <a name="l00320"></a>00320 |
| <a name="l00321"></a>00321 <span class="comment">/* column loop */</span> |
| <a name="l00322"></a>00322 <span class="keywordflow">do</span> |
| <a name="l00323"></a>00323 { |
| <a name="l00324"></a>00324 <span class="comment">/* Set the variable sum, that acts as accumulator, to zero */</span> |
| <a name="l00325"></a>00325 sum = 0; |
| <a name="l00326"></a>00326 |
| <a name="l00327"></a>00327 <span class="comment">/* Initiate the pointer pIn1 to point to the starting address of pSrcA */</span> |
| <a name="l00328"></a>00328 pIn1 = pInA; |
| <a name="l00329"></a>00329 |
| <a name="l00330"></a>00330 <span class="comment">/* Matrix A columns number of MAC operations are to be performed */</span> |
| <a name="l00331"></a>00331 colCnt = numColsA; |
| <a name="l00332"></a>00332 |
| <a name="l00333"></a>00333 <span class="comment">/* matrix multiplication */</span> |
| <a name="l00334"></a>00334 <span class="keywordflow">while</span>(colCnt > 0u) |
| <a name="l00335"></a>00335 { |
| <a name="l00336"></a>00336 <span class="comment">/* c(m,n) = a(1,1)*b(1,1) + a(1,2) * b(2,1) + .... + a(m,p)*b(p,n) */</span> |
| <a name="l00337"></a>00337 <span class="comment">/* Perform the multiply-accumulates */</span> |
| <a name="l00338"></a>00338 sum += (<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) * pIn1++ * *pIn2; |
| <a name="l00339"></a>00339 pIn2 += numColsB; |
| <a name="l00340"></a>00340 |
| <a name="l00341"></a>00341 <span class="comment">/* Decrement the loop counter */</span> |
| <a name="l00342"></a>00342 colCnt--; |
| <a name="l00343"></a>00343 } |
| <a name="l00344"></a>00344 |
| <a name="l00345"></a>00345 <span class="comment">/* Convert the result from 34.30 to 1.15 format and store the saturated value in destination buffer */</span> |
| <a name="l00346"></a>00346 <span class="comment">/* Saturate and store the result in the destination buffer */</span> |
| <a name="l00347"></a>00347 *px++ = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) __SSAT((sum >> 15), 16); |
| <a name="l00348"></a>00348 |
| <a name="l00349"></a>00349 <span class="comment">/* Decrement the column loop counter */</span> |
| <a name="l00350"></a>00350 col--; |
| <a name="l00351"></a>00351 |
| <a name="l00352"></a>00352 <span class="comment">/* Update the pointer pIn2 to point to the starting address of the next column */</span> |
| <a name="l00353"></a>00353 pIn2 = pInB + (numColsB - col); |
| <a name="l00354"></a>00354 |
| <a name="l00355"></a>00355 } <span class="keywordflow">while</span>(col > 0u); |
| <a name="l00356"></a>00356 |
| <a name="l00357"></a>00357 <span class="comment">/* Update the pointer pSrcA to point to the starting address of the next row */</span> |
| <a name="l00358"></a>00358 i = i + numColsB; |
| <a name="l00359"></a>00359 pInA = pInA + numColsA; |
| <a name="l00360"></a>00360 |
| <a name="l00361"></a>00361 <span class="comment">/* Decrement the row loop counter */</span> |
| <a name="l00362"></a>00362 row--; |
| <a name="l00363"></a>00363 |
| <a name="l00364"></a>00364 } <span class="keywordflow">while</span>(row > 0u); |
| <a name="l00365"></a>00365 |
| <a name="l00366"></a>00366 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_CM0 */</span> |
| <a name="l00367"></a>00367 |
| <a name="l00368"></a>00368 <span class="comment">/* set status as ARM_MATH_SUCCESS */</span> |
| <a name="l00369"></a>00369 status = <a class="code" href="arm__math_8h.html#a5e459c6409dfcd2927bb8a57491d7cf6a9f8b2a10bd827fb4600e77d455902eb0">ARM_MATH_SUCCESS</a>; |
| <a name="l00370"></a>00370 } |
| <a name="l00371"></a>00371 |
| <a name="l00372"></a>00372 <span class="comment">/* Return to application */</span> |
| <a name="l00373"></a>00373 <span class="keywordflow">return</span> (status); |
| <a name="l00374"></a>00374 } |
| <a name="l00375"></a>00375 |
| </pre></div></div> |
| </div> |
| <!--- window showing the filter options --> |
| <div id="MSearchSelectWindow" |
| onmouseover="return searchBox.OnSearchSelectShow()" |
| onmouseout="return searchBox.OnSearchSelectHide()" |
| onkeydown="return searchBox.OnSearchSelectKey(event)"> |
| <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>Files</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(3)"><span class="SelectionMark"> </span>Functions</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(4)"><span class="SelectionMark"> </span>Variables</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(5)"><span class="SelectionMark"> </span>Typedefs</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(6)"><span class="SelectionMark"> </span>Enumerations</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(7)"><span class="SelectionMark"> </span>Enumerator</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(8)"><span class="SelectionMark"> </span>Defines</a></div> |
| |
| <!-- iframe showing the search results (closed by default) --> |
| <div id="MSearchResultsWindow"> |
| <iframe src="" frameborder="0" |
| name="MSearchResults" id="MSearchResults"> |
| </iframe> |
| </div> |
| |
| <hr class="footer"/><address class="footer"><small>Generated on Fri Jul 15 2011 13:16:16 for CMSIS DSP Software Library by  |
| <a href="http://www.doxygen.org/index.html"> |
| <img class="footer" src="doxygen.png" alt="doxygen"/></a> 1.7.2 </small></address> |
| </body> |
| </html> |
