blob: 5ec02833ea9e31db2c8820e9fd8e959e775cda38 [file] [log] [blame]
<!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_dct4_q31.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&#160;Page</span></a></li>
<li><a href="modules.html"><span>Modules</span></a></li>
<li><a href="annotated.html"><span>Data&#160;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&#160;List</span></a></li>
<li><a href="globals.html"><span>Globals</span></a></li>
</ul>
</div>
<div class="header">
<div class="headertitle">
<h1>arm_dct4_q31.c</h1> </div>
</div>
<div class="contents">
<a href="arm__dct4__q31_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_dct4_q31.c </span>
<a name="l00009"></a>00009 <span class="comment">* </span>
<a name="l00010"></a>00010 <span class="comment">* Description: Processing function of DCT4 &amp; IDCT4 Q31. </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
<a name="l00030"></a>00030 <span class="preprocessor">#include &quot;<a class="code" href="arm__math_8h.html">arm_math.h</a>&quot;</span>
<a name="l00031"></a>00031
<a name="l00053"></a><a class="code" href="group___d_c_t4___i_d_c_t4.html#gad04d0baab6ed081d8e8afe02538eb80b">00053</a> <span class="keywordtype">void</span> <a class="code" href="group___d_c_t4___i_d_c_t4.html#gad04d0baab6ed081d8e8afe02538eb80b" title="Processing function for the Q31 DCT4/IDCT4.">arm_dct4_q31</a>(
<a name="l00054"></a>00054 <span class="keyword">const</span> <a class="code" href="structarm__dct4__instance__q31.html" title="Instance structure for the Q31 DCT4/IDCT4 function.">arm_dct4_instance_q31</a> * S,
<a name="l00055"></a>00055 <a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> * pState,
<a name="l00056"></a>00056 <a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> * pInlineBuffer)
<a name="l00057"></a>00057 {
<a name="l00058"></a>00058 uint16_t i; <span class="comment">/* Loop counter */</span>
<a name="l00059"></a>00059 <a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *weights = S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a7db236e22673146bb1d2c962f0713f08">pTwiddle</a>; <span class="comment">/* Pointer to the Weights table */</span>
<a name="l00060"></a>00060 <a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *cosFact = S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#af97204d1838925621fc82021a0c2d6c1">pCosFactor</a>; <span class="comment">/* Pointer to the cos factors table */</span>
<a name="l00061"></a>00061 <a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *pS1, *pS2, *pbuff; <span class="comment">/* Temporary pointers for input buffer and pState buffer */</span>
<a name="l00062"></a>00062 <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 */</span>
<a name="l00063"></a>00063
<a name="l00064"></a>00064
<a name="l00065"></a>00065 <span class="comment">/* DCT4 computation involves DCT2 (which is calculated using RFFT) </span>
<a name="l00066"></a>00066 <span class="comment"> * along with some pre-processing and post-processing. </span>
<a name="l00067"></a>00067 <span class="comment"> * Computational procedure is explained as follows: </span>
<a name="l00068"></a>00068 <span class="comment"> * (a) Pre-processing involves multiplying input with cos factor, </span>
<a name="l00069"></a>00069 <span class="comment"> * r(n) = 2 * u(n) * cos(pi*(2*n+1)/(4*n)) </span>
<a name="l00070"></a>00070 <span class="comment"> * where, </span>
<a name="l00071"></a>00071 <span class="comment"> * r(n) -- output of preprocessing </span>
<a name="l00072"></a>00072 <span class="comment"> * u(n) -- input to preprocessing(actual Source buffer) </span>
<a name="l00073"></a>00073 <span class="comment"> * (b) Calculation of DCT2 using FFT is divided into three steps: </span>
<a name="l00074"></a>00074 <span class="comment"> * Step1: Re-ordering of even and odd elements of input. </span>
<a name="l00075"></a>00075 <span class="comment"> * Step2: Calculating FFT of the re-ordered input. </span>
<a name="l00076"></a>00076 <span class="comment"> * Step3: Taking the real part of the product of FFT output and weights. </span>
<a name="l00077"></a>00077 <span class="comment"> * (c) Post-processing - DCT4 can be obtained from DCT2 output using the following equation: </span>
<a name="l00078"></a>00078 <span class="comment"> * Y4(k) = Y2(k) - Y4(k-1) and Y4(-1) = Y4(0) </span>
<a name="l00079"></a>00079 <span class="comment"> * where, </span>
<a name="l00080"></a>00080 <span class="comment"> * Y4 -- DCT4 output, Y2 -- DCT2 output </span>
<a name="l00081"></a>00081 <span class="comment"> * (d) Multiplying the output with the normalizing factor sqrt(2/N). </span>
<a name="l00082"></a>00082 <span class="comment"> */</span>
<a name="l00083"></a>00083
<a name="l00084"></a>00084 <span class="comment">/*-------- Pre-processing ------------*/</span>
<a name="l00085"></a>00085 <span class="comment">/* Multiplying input with cos factor i.e. r(n) = 2 * x(n) * cos(pi*(2*n+1)/(4*n)) */</span>
<a name="l00086"></a>00086 <a class="code" href="group___basic_mult.html#ga3528c0f54a0607acc603f0490d3ca6c6" title="Q31 vector multiplication.">arm_mult_q31</a>(pInlineBuffer, cosFact, pInlineBuffer, S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a46a9f136457350676e2bfd3768ff9d6d">N</a>);
<a name="l00087"></a>00087 <a class="code" href="group__shift.html#ga387dd8b7b87377378280978f16cdb13d" title="Shifts the elements of a Q31 vector a specified number of bits.">arm_shift_q31</a>(pInlineBuffer, 1, pInlineBuffer, S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a46a9f136457350676e2bfd3768ff9d6d">N</a>);
<a name="l00088"></a>00088
<a name="l00089"></a>00089 <span class="comment">/* ---------------------------------------------------------------- </span>
<a name="l00090"></a>00090 <span class="comment"> * Step1: Re-ordering of even and odd elements as </span>
<a name="l00091"></a>00091 <span class="comment"> * pState[i] = pInlineBuffer[2*i] and </span>
<a name="l00092"></a>00092 <span class="comment"> * pState[N-i-1] = pInlineBuffer[2*i+1] where i = 0 to N/2 </span>
<a name="l00093"></a>00093 <span class="comment"> ---------------------------------------------------------------------*/</span>
<a name="l00094"></a>00094
<a name="l00095"></a>00095 <span class="comment">/* pS1 initialized to pState */</span>
<a name="l00096"></a>00096 pS1 = pState;
<a name="l00097"></a>00097
<a name="l00098"></a>00098 <span class="comment">/* pS2 initialized to pState+N-1, so that it points to the end of the state buffer */</span>
<a name="l00099"></a>00099 pS2 = pState + (S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a46a9f136457350676e2bfd3768ff9d6d">N</a> - 1u);
<a name="l00100"></a>00100
<a name="l00101"></a>00101 <span class="comment">/* pbuff initialized to input buffer */</span>
<a name="l00102"></a>00102 pbuff = pInlineBuffer;
<a name="l00103"></a>00103
<a name="l00104"></a>00104 <span class="preprocessor">#ifndef ARM_MATH_CM0</span>
<a name="l00105"></a>00105 <span class="preprocessor"></span>
<a name="l00106"></a>00106 <span class="comment">/* Run the below code for Cortex-M4 and Cortex-M3 */</span>
<a name="l00107"></a>00107
<a name="l00108"></a>00108 <span class="comment">/* Initializing the loop counter to N/2 &gt;&gt; 2 for loop unrolling by 4 */</span>
<a name="l00109"></a>00109 i = S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a32d3268ba4629908dba056599f0a904d">Nby2</a> &gt;&gt; 2u;
<a name="l00110"></a>00110
<a name="l00111"></a>00111 <span class="comment">/* First part of the processing with loop unrolling. Compute 4 outputs at a time. </span>
<a name="l00112"></a>00112 <span class="comment"> ** a second loop below computes the remaining 1 to 3 samples. */</span>
<a name="l00113"></a>00113 <span class="keywordflow">do</span>
<a name="l00114"></a>00114 {
<a name="l00115"></a>00115 <span class="comment">/* Re-ordering of even and odd elements */</span>
<a name="l00116"></a>00116 <span class="comment">/* pState[i] = pInlineBuffer[2*i] */</span>
<a name="l00117"></a>00117 *pS1++ = *pbuff++;
<a name="l00118"></a>00118 <span class="comment">/* pState[N-i-1] = pInlineBuffer[2*i+1] */</span>
<a name="l00119"></a>00119 *pS2-- = *pbuff++;
<a name="l00120"></a>00120
<a name="l00121"></a>00121 *pS1++ = *pbuff++;
<a name="l00122"></a>00122 *pS2-- = *pbuff++;
<a name="l00123"></a>00123
<a name="l00124"></a>00124 *pS1++ = *pbuff++;
<a name="l00125"></a>00125 *pS2-- = *pbuff++;
<a name="l00126"></a>00126
<a name="l00127"></a>00127 *pS1++ = *pbuff++;
<a name="l00128"></a>00128 *pS2-- = *pbuff++;
<a name="l00129"></a>00129
<a name="l00130"></a>00130 <span class="comment">/* Decrement the loop counter */</span>
<a name="l00131"></a>00131 i--;
<a name="l00132"></a>00132 } <span class="keywordflow">while</span>(i &gt; 0u);
<a name="l00133"></a>00133
<a name="l00134"></a>00134 <span class="comment">/* pbuff initialized to input buffer */</span>
<a name="l00135"></a>00135 pbuff = pInlineBuffer;
<a name="l00136"></a>00136
<a name="l00137"></a>00137 <span class="comment">/* pS1 initialized to pState */</span>
<a name="l00138"></a>00138 pS1 = pState;
<a name="l00139"></a>00139
<a name="l00140"></a>00140 <span class="comment">/* Initializing the loop counter to N/4 instead of N for loop unrolling */</span>
<a name="l00141"></a>00141 i = S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a46a9f136457350676e2bfd3768ff9d6d">N</a> &gt;&gt; 2u;
<a name="l00142"></a>00142
<a name="l00143"></a>00143 <span class="comment">/* Processing with loop unrolling 4 times as N is always multiple of 4. </span>
<a name="l00144"></a>00144 <span class="comment"> * Compute 4 outputs at a time */</span>
<a name="l00145"></a>00145 <span class="keywordflow">do</span>
<a name="l00146"></a>00146 {
<a name="l00147"></a>00147 <span class="comment">/* Writing the re-ordered output back to inplace input buffer */</span>
<a name="l00148"></a>00148 *pbuff++ = *pS1++;
<a name="l00149"></a>00149 *pbuff++ = *pS1++;
<a name="l00150"></a>00150 *pbuff++ = *pS1++;
<a name="l00151"></a>00151 *pbuff++ = *pS1++;
<a name="l00152"></a>00152
<a name="l00153"></a>00153 <span class="comment">/* Decrement the loop counter */</span>
<a name="l00154"></a>00154 i--;
<a name="l00155"></a>00155 } <span class="keywordflow">while</span>(i &gt; 0u);
<a name="l00156"></a>00156
<a name="l00157"></a>00157
<a name="l00158"></a>00158 <span class="comment">/* --------------------------------------------------------- </span>
<a name="l00159"></a>00159 <span class="comment"> * Step2: Calculate RFFT for N-point input </span>
<a name="l00160"></a>00160 <span class="comment"> * ---------------------------------------------------------- */</span>
<a name="l00161"></a>00161 <span class="comment">/* pInlineBuffer is real input of length N , pState is the complex output of length 2N */</span>
<a name="l00162"></a>00162 <a class="code" href="group___r_f_f_t___r_i_f_f_t.html#gabaeab5646aeea9844e6d42ca8c73fe3a" title="Processing function for the Q31 RFFT/RIFFT.">arm_rfft_q31</a>(S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#af1487dab5e7963b85dc0fdc6bf492542">pRfft</a>, pInlineBuffer, pState);
<a name="l00163"></a>00163
<a name="l00164"></a>00164 <span class="comment">/*---------------------------------------------------------------------- </span>
<a name="l00165"></a>00165 <span class="comment"> * Step3: Multiply the FFT output with the weights. </span>
<a name="l00166"></a>00166 <span class="comment"> *----------------------------------------------------------------------*/</span>
<a name="l00167"></a>00167 <a class="code" href="group___cmplx_by_cmplx_mult.html#ga1829e50993a90742de225a0ce4213838" title="Q31 complex-by-complex multiplication.">arm_cmplx_mult_cmplx_q31</a>(pState, weights, pState, S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a46a9f136457350676e2bfd3768ff9d6d">N</a>);
<a name="l00168"></a>00168
<a name="l00169"></a>00169 <span class="comment">/* The output of complex multiplication is in 3.29 format. </span>
<a name="l00170"></a>00170 <span class="comment"> * Hence changing the format of N (i.e. 2*N elements) complex numbers to 1.31 format by shifting left by 2 bits. */</span>
<a name="l00171"></a>00171 <a class="code" href="group__shift.html#ga387dd8b7b87377378280978f16cdb13d" title="Shifts the elements of a Q31 vector a specified number of bits.">arm_shift_q31</a>(pState, 2, pState, S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a46a9f136457350676e2bfd3768ff9d6d">N</a> * 2);
<a name="l00172"></a>00172
<a name="l00173"></a>00173 <span class="comment">/* ----------- Post-processing ---------- */</span>
<a name="l00174"></a>00174 <span class="comment">/* DCT-IV can be obtained from DCT-II by the equation, </span>
<a name="l00175"></a>00175 <span class="comment"> * Y4(k) = Y2(k) - Y4(k-1) and Y4(-1) = Y4(0) </span>
<a name="l00176"></a>00176 <span class="comment"> * Hence, Y4(0) = Y2(0)/2 */</span>
<a name="l00177"></a>00177 <span class="comment">/* Getting only real part from the output and Converting to DCT-IV */</span>
<a name="l00178"></a>00178
<a name="l00179"></a>00179 <span class="comment">/* Initializing the loop counter to N &gt;&gt; 2 for loop unrolling by 4 */</span>
<a name="l00180"></a>00180 i = (S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a46a9f136457350676e2bfd3768ff9d6d">N</a> - 1u) &gt;&gt; 2u;
<a name="l00181"></a>00181
<a name="l00182"></a>00182 <span class="comment">/* pbuff initialized to input buffer. */</span>
<a name="l00183"></a>00183 pbuff = pInlineBuffer;
<a name="l00184"></a>00184
<a name="l00185"></a>00185 <span class="comment">/* pS1 initialized to pState */</span>
<a name="l00186"></a>00186 pS1 = pState;
<a name="l00187"></a>00187
<a name="l00188"></a>00188 <span class="comment">/* Calculating Y4(0) from Y2(0) using Y4(0) = Y2(0)/2 */</span>
<a name="l00189"></a>00189 in = *pS1++ &gt;&gt; 1u;
<a name="l00190"></a>00190 <span class="comment">/* input buffer acts as inplace, so output values are stored in the input itself. */</span>
<a name="l00191"></a>00191 *pbuff++ = in;
<a name="l00192"></a>00192
<a name="l00193"></a>00193 <span class="comment">/* pState pointer is incremented twice as the real values are located alternatively in the array */</span>
<a name="l00194"></a>00194 pS1++;
<a name="l00195"></a>00195
<a name="l00196"></a>00196 <span class="comment">/* First part of the processing with loop unrolling. Compute 4 outputs at a time. </span>
<a name="l00197"></a>00197 <span class="comment"> ** a second loop below computes the remaining 1 to 3 samples. */</span>
<a name="l00198"></a>00198 <span class="keywordflow">do</span>
<a name="l00199"></a>00199 {
<a name="l00200"></a>00200 <span class="comment">/* Calculating Y4(1) to Y4(N-1) from Y2 using equation Y4(k) = Y2(k) - Y4(k-1) */</span>
<a name="l00201"></a>00201 <span class="comment">/* pState pointer (pS1) is incremented twice as the real values are located alternatively in the array */</span>
<a name="l00202"></a>00202 in = *pS1++ - in;
<a name="l00203"></a>00203 *pbuff++ = in;
<a name="l00204"></a>00204 <span class="comment">/* points to the next real value */</span>
<a name="l00205"></a>00205 pS1++;
<a name="l00206"></a>00206
<a name="l00207"></a>00207 in = *pS1++ - in;
<a name="l00208"></a>00208 *pbuff++ = in;
<a name="l00209"></a>00209 pS1++;
<a name="l00210"></a>00210
<a name="l00211"></a>00211 in = *pS1++ - in;
<a name="l00212"></a>00212 *pbuff++ = in;
<a name="l00213"></a>00213 pS1++;
<a name="l00214"></a>00214
<a name="l00215"></a>00215 in = *pS1++ - in;
<a name="l00216"></a>00216 *pbuff++ = in;
<a name="l00217"></a>00217 pS1++;
<a name="l00218"></a>00218
<a name="l00219"></a>00219 <span class="comment">/* Decrement the loop counter */</span>
<a name="l00220"></a>00220 i--;
<a name="l00221"></a>00221 } <span class="keywordflow">while</span>(i &gt; 0u);
<a name="l00222"></a>00222
<a name="l00223"></a>00223 <span class="comment">/* If the blockSize is not a multiple of 4, compute any remaining output samples here. </span>
<a name="l00224"></a>00224 <span class="comment"> ** No loop unrolling is used. */</span>
<a name="l00225"></a>00225 i = (S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a46a9f136457350676e2bfd3768ff9d6d">N</a> - 1u) % 0x4u;
<a name="l00226"></a>00226
<a name="l00227"></a>00227 <span class="keywordflow">while</span>(i &gt; 0u)
<a name="l00228"></a>00228 {
<a name="l00229"></a>00229 <span class="comment">/* Calculating Y4(1) to Y4(N-1) from Y2 using equation Y4(k) = Y2(k) - Y4(k-1) */</span>
<a name="l00230"></a>00230 <span class="comment">/* pState pointer (pS1) is incremented twice as the real values are located alternatively in the array */</span>
<a name="l00231"></a>00231 in = *pS1++ - in;
<a name="l00232"></a>00232 *pbuff++ = in;
<a name="l00233"></a>00233 <span class="comment">/* points to the next real value */</span>
<a name="l00234"></a>00234 pS1++;
<a name="l00235"></a>00235
<a name="l00236"></a>00236 <span class="comment">/* Decrement the loop counter */</span>
<a name="l00237"></a>00237 i--;
<a name="l00238"></a>00238 }
<a name="l00239"></a>00239
<a name="l00240"></a>00240
<a name="l00241"></a>00241 <span class="comment">/*------------ Normalizing the output by multiplying with the normalizing factor ----------*/</span>
<a name="l00242"></a>00242
<a name="l00243"></a>00243 <span class="comment">/* Initializing the loop counter to N/4 instead of N for loop unrolling */</span>
<a name="l00244"></a>00244 i = S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a46a9f136457350676e2bfd3768ff9d6d">N</a> &gt;&gt; 2u;
<a name="l00245"></a>00245
<a name="l00246"></a>00246 <span class="comment">/* pbuff initialized to the pInlineBuffer(now contains the output values) */</span>
<a name="l00247"></a>00247 pbuff = pInlineBuffer;
<a name="l00248"></a>00248
<a name="l00249"></a>00249 <span class="comment">/* Processing with loop unrolling 4 times as N is always multiple of 4. Compute 4 outputs at a time */</span>
<a name="l00250"></a>00250 <span class="keywordflow">do</span>
<a name="l00251"></a>00251 {
<a name="l00252"></a>00252 <span class="comment">/* Multiplying pInlineBuffer with the normalizing factor sqrt(2/N) */</span>
<a name="l00253"></a>00253 in = *pbuff;
<a name="l00254"></a>00254 *pbuff++ = ((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) (((<a class="code" href="arm__math_8h.html#a5aea1cb12fc02d9d44c8abf217eaa5c6" title="64-bit fractional data type in 1.63 format.">q63_t</a>) in * S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#ac80ff7b28fca36aeef74dea12e8312dd">normalize</a>) &gt;&gt; 31));
<a name="l00255"></a>00255
<a name="l00256"></a>00256 in = *pbuff;
<a name="l00257"></a>00257 *pbuff++ = ((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) (((<a class="code" href="arm__math_8h.html#a5aea1cb12fc02d9d44c8abf217eaa5c6" title="64-bit fractional data type in 1.63 format.">q63_t</a>) in * S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#ac80ff7b28fca36aeef74dea12e8312dd">normalize</a>) &gt;&gt; 31));
<a name="l00258"></a>00258
<a name="l00259"></a>00259 in = *pbuff;
<a name="l00260"></a>00260 *pbuff++ = ((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) (((<a class="code" href="arm__math_8h.html#a5aea1cb12fc02d9d44c8abf217eaa5c6" title="64-bit fractional data type in 1.63 format.">q63_t</a>) in * S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#ac80ff7b28fca36aeef74dea12e8312dd">normalize</a>) &gt;&gt; 31));
<a name="l00261"></a>00261
<a name="l00262"></a>00262 in = *pbuff;
<a name="l00263"></a>00263 *pbuff++ = ((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) (((<a class="code" href="arm__math_8h.html#a5aea1cb12fc02d9d44c8abf217eaa5c6" title="64-bit fractional data type in 1.63 format.">q63_t</a>) in * S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#ac80ff7b28fca36aeef74dea12e8312dd">normalize</a>) &gt;&gt; 31));
<a name="l00264"></a>00264
<a name="l00265"></a>00265 <span class="comment">/* Decrement the loop counter */</span>
<a name="l00266"></a>00266 i--;
<a name="l00267"></a>00267 } <span class="keywordflow">while</span>(i &gt; 0u);
<a name="l00268"></a>00268
<a name="l00269"></a>00269
<a name="l00270"></a>00270 <span class="preprocessor">#else</span>
<a name="l00271"></a>00271 <span class="preprocessor"></span>
<a name="l00272"></a>00272 <span class="comment">/* Run the below code for Cortex-M0 */</span>
<a name="l00273"></a>00273
<a name="l00274"></a>00274 <span class="comment">/* Initializing the loop counter to N/2 */</span>
<a name="l00275"></a>00275 i = S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a32d3268ba4629908dba056599f0a904d">Nby2</a>;
<a name="l00276"></a>00276
<a name="l00277"></a>00277 <span class="keywordflow">do</span>
<a name="l00278"></a>00278 {
<a name="l00279"></a>00279 <span class="comment">/* Re-ordering of even and odd elements */</span>
<a name="l00280"></a>00280 <span class="comment">/* pState[i] = pInlineBuffer[2*i] */</span>
<a name="l00281"></a>00281 *pS1++ = *pbuff++;
<a name="l00282"></a>00282 <span class="comment">/* pState[N-i-1] = pInlineBuffer[2*i+1] */</span>
<a name="l00283"></a>00283 *pS2-- = *pbuff++;
<a name="l00284"></a>00284
<a name="l00285"></a>00285 <span class="comment">/* Decrement the loop counter */</span>
<a name="l00286"></a>00286 i--;
<a name="l00287"></a>00287 } <span class="keywordflow">while</span>(i &gt; 0u);
<a name="l00288"></a>00288
<a name="l00289"></a>00289 <span class="comment">/* pbuff initialized to input buffer */</span>
<a name="l00290"></a>00290 pbuff = pInlineBuffer;
<a name="l00291"></a>00291
<a name="l00292"></a>00292 <span class="comment">/* pS1 initialized to pState */</span>
<a name="l00293"></a>00293 pS1 = pState;
<a name="l00294"></a>00294
<a name="l00295"></a>00295 <span class="comment">/* Initializing the loop counter */</span>
<a name="l00296"></a>00296 i = S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a46a9f136457350676e2bfd3768ff9d6d">N</a>;
<a name="l00297"></a>00297
<a name="l00298"></a>00298 <span class="keywordflow">do</span>
<a name="l00299"></a>00299 {
<a name="l00300"></a>00300 <span class="comment">/* Writing the re-ordered output back to inplace input buffer */</span>
<a name="l00301"></a>00301 *pbuff++ = *pS1++;
<a name="l00302"></a>00302
<a name="l00303"></a>00303 <span class="comment">/* Decrement the loop counter */</span>
<a name="l00304"></a>00304 i--;
<a name="l00305"></a>00305 } <span class="keywordflow">while</span>(i &gt; 0u);
<a name="l00306"></a>00306
<a name="l00307"></a>00307
<a name="l00308"></a>00308 <span class="comment">/* --------------------------------------------------------- </span>
<a name="l00309"></a>00309 <span class="comment"> * Step2: Calculate RFFT for N-point input </span>
<a name="l00310"></a>00310 <span class="comment"> * ---------------------------------------------------------- */</span>
<a name="l00311"></a>00311 <span class="comment">/* pInlineBuffer is real input of length N , pState is the complex output of length 2N */</span>
<a name="l00312"></a>00312 <a class="code" href="group___r_f_f_t___r_i_f_f_t.html#gabaeab5646aeea9844e6d42ca8c73fe3a" title="Processing function for the Q31 RFFT/RIFFT.">arm_rfft_q31</a>(S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#af1487dab5e7963b85dc0fdc6bf492542">pRfft</a>, pInlineBuffer, pState);
<a name="l00313"></a>00313
<a name="l00314"></a>00314 <span class="comment">/*---------------------------------------------------------------------- </span>
<a name="l00315"></a>00315 <span class="comment"> * Step3: Multiply the FFT output with the weights. </span>
<a name="l00316"></a>00316 <span class="comment"> *----------------------------------------------------------------------*/</span>
<a name="l00317"></a>00317 <a class="code" href="group___cmplx_by_cmplx_mult.html#ga1829e50993a90742de225a0ce4213838" title="Q31 complex-by-complex multiplication.">arm_cmplx_mult_cmplx_q31</a>(pState, weights, pState, S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a46a9f136457350676e2bfd3768ff9d6d">N</a>);
<a name="l00318"></a>00318
<a name="l00319"></a>00319 <span class="comment">/* The output of complex multiplication is in 3.29 format. </span>
<a name="l00320"></a>00320 <span class="comment"> * Hence changing the format of N (i.e. 2*N elements) complex numbers to 1.31 format by shifting left by 2 bits. */</span>
<a name="l00321"></a>00321 <a class="code" href="group__shift.html#ga387dd8b7b87377378280978f16cdb13d" title="Shifts the elements of a Q31 vector a specified number of bits.">arm_shift_q31</a>(pState, 2, pState, S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a46a9f136457350676e2bfd3768ff9d6d">N</a> * 2);
<a name="l00322"></a>00322
<a name="l00323"></a>00323 <span class="comment">/* ----------- Post-processing ---------- */</span>
<a name="l00324"></a>00324 <span class="comment">/* DCT-IV can be obtained from DCT-II by the equation, </span>
<a name="l00325"></a>00325 <span class="comment"> * Y4(k) = Y2(k) - Y4(k-1) and Y4(-1) = Y4(0) </span>
<a name="l00326"></a>00326 <span class="comment"> * Hence, Y4(0) = Y2(0)/2 */</span>
<a name="l00327"></a>00327 <span class="comment">/* Getting only real part from the output and Converting to DCT-IV */</span>
<a name="l00328"></a>00328
<a name="l00329"></a>00329 <span class="comment">/* pbuff initialized to input buffer. */</span>
<a name="l00330"></a>00330 pbuff = pInlineBuffer;
<a name="l00331"></a>00331
<a name="l00332"></a>00332 <span class="comment">/* pS1 initialized to pState */</span>
<a name="l00333"></a>00333 pS1 = pState;
<a name="l00334"></a>00334
<a name="l00335"></a>00335 <span class="comment">/* Calculating Y4(0) from Y2(0) using Y4(0) = Y2(0)/2 */</span>
<a name="l00336"></a>00336 in = *pS1++ &gt;&gt; 1u;
<a name="l00337"></a>00337 <span class="comment">/* input buffer acts as inplace, so output values are stored in the input itself. */</span>
<a name="l00338"></a>00338 *pbuff++ = in;
<a name="l00339"></a>00339
<a name="l00340"></a>00340 <span class="comment">/* pState pointer is incremented twice as the real values are located alternatively in the array */</span>
<a name="l00341"></a>00341 pS1++;
<a name="l00342"></a>00342
<a name="l00343"></a>00343 <span class="comment">/* Initializing the loop counter */</span>
<a name="l00344"></a>00344 i = (S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a46a9f136457350676e2bfd3768ff9d6d">N</a> - 1u);
<a name="l00345"></a>00345
<a name="l00346"></a>00346 <span class="keywordflow">while</span>(i &gt; 0u)
<a name="l00347"></a>00347 {
<a name="l00348"></a>00348 <span class="comment">/* Calculating Y4(1) to Y4(N-1) from Y2 using equation Y4(k) = Y2(k) - Y4(k-1) */</span>
<a name="l00349"></a>00349 <span class="comment">/* pState pointer (pS1) is incremented twice as the real values are located alternatively in the array */</span>
<a name="l00350"></a>00350 in = *pS1++ - in;
<a name="l00351"></a>00351 *pbuff++ = in;
<a name="l00352"></a>00352 <span class="comment">/* points to the next real value */</span>
<a name="l00353"></a>00353 pS1++;
<a name="l00354"></a>00354
<a name="l00355"></a>00355 <span class="comment">/* Decrement the loop counter */</span>
<a name="l00356"></a>00356 i--;
<a name="l00357"></a>00357 }
<a name="l00358"></a>00358
<a name="l00359"></a>00359
<a name="l00360"></a>00360 <span class="comment">/*------------ Normalizing the output by multiplying with the normalizing factor ----------*/</span>
<a name="l00361"></a>00361
<a name="l00362"></a>00362 <span class="comment">/* Initializing the loop counter */</span>
<a name="l00363"></a>00363 i = S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#a46a9f136457350676e2bfd3768ff9d6d">N</a>;
<a name="l00364"></a>00364
<a name="l00365"></a>00365 <span class="comment">/* pbuff initialized to the pInlineBuffer(now contains the output values) */</span>
<a name="l00366"></a>00366 pbuff = pInlineBuffer;
<a name="l00367"></a>00367
<a name="l00368"></a>00368 <span class="keywordflow">do</span>
<a name="l00369"></a>00369 {
<a name="l00370"></a>00370 <span class="comment">/* Multiplying pInlineBuffer with the normalizing factor sqrt(2/N) */</span>
<a name="l00371"></a>00371 in = *pbuff;
<a name="l00372"></a>00372 *pbuff++ = ((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) (((<a class="code" href="arm__math_8h.html#a5aea1cb12fc02d9d44c8abf217eaa5c6" title="64-bit fractional data type in 1.63 format.">q63_t</a>) in * S-&gt;<a class="code" href="structarm__dct4__instance__q31.html#ac80ff7b28fca36aeef74dea12e8312dd">normalize</a>) &gt;&gt; 31));
<a name="l00373"></a>00373
<a name="l00374"></a>00374 <span class="comment">/* Decrement the loop counter */</span>
<a name="l00375"></a>00375 i--;
<a name="l00376"></a>00376 } <span class="keywordflow">while</span>(i &gt; 0u);
<a name="l00377"></a>00377
<a name="l00378"></a>00378 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_CM0 */</span>
<a name="l00379"></a>00379
<a name="l00380"></a>00380 }
<a name="l00381"></a>00381
</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">&#160;</span>All</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(1)"><span class="SelectionMark">&#160;</span>Data Structures</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(2)"><span class="SelectionMark">&#160;</span>Files</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(3)"><span class="SelectionMark">&#160;</span>Functions</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(4)"><span class="SelectionMark">&#160;</span>Variables</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(5)"><span class="SelectionMark">&#160;</span>Typedefs</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(6)"><span class="SelectionMark">&#160;</span>Enumerations</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(7)"><span class="SelectionMark">&#160;</span>Enumerator</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(8)"><span class="SelectionMark">&#160;</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&#160;
<a href="http://www.doxygen.org/index.html">
<img class="footer" src="doxygen.png" alt="doxygen"/></a> 1.7.2 </small></address>
</body>
</html>