| <!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_cfft_radix4_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_cfft_radix4_q15.c</h1> </div> |
| </div> |
| <div class="contents"> |
| <a href="arm__cfft__radix4__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_cfft_radix4_q15.c </span> |
| <a name="l00009"></a>00009 <span class="comment">* </span> |
| <a name="l00010"></a>00010 <span class="comment">* Description: This file has function definition of Radix-4 FFT & IFFT function and </span> |
| <a name="l00011"></a>00011 <span class="comment">* In-place bit reversal using bit reversal table </span> |
| <a name="l00012"></a>00012 <span class="comment">* </span> |
| <a name="l00013"></a>00013 <span class="comment">* Target Processor: Cortex-M4/Cortex-M3/Cortex-M0</span> |
| <a name="l00014"></a>00014 <span class="comment">* </span> |
| <a name="l00015"></a>00015 <span class="comment">* Version 1.0.10 2011/7/15 </span> |
| <a name="l00016"></a>00016 <span class="comment">* Big Endian support added and Merged M0 and M3/M4 Source code. </span> |
| <a name="l00017"></a>00017 <span class="comment">* </span> |
| <a name="l00018"></a>00018 <span class="comment">* Version 1.0.3 2010/11/29 </span> |
| <a name="l00019"></a>00019 <span class="comment">* Re-organized the CMSIS folders and updated documentation. </span> |
| <a name="l00020"></a>00020 <span class="comment">* </span> |
| <a name="l00021"></a>00021 <span class="comment">* Version 1.0.2 2010/11/11 </span> |
| <a name="l00022"></a>00022 <span class="comment">* Documentation updated. </span> |
| <a name="l00023"></a>00023 <span class="comment">* </span> |
| <a name="l00024"></a>00024 <span class="comment">* Version 1.0.1 2010/10/05 </span> |
| <a name="l00025"></a>00025 <span class="comment">* Production release and review comments incorporated. </span> |
| <a name="l00026"></a>00026 <span class="comment">* </span> |
| <a name="l00027"></a>00027 <span class="comment">* Version 1.0.0 2010/09/20 </span> |
| <a name="l00028"></a>00028 <span class="comment">* Production release and review comments incorporated. </span> |
| <a name="l00029"></a>00029 <span class="comment">* </span> |
| <a name="l00030"></a>00030 <span class="comment">* Version 0.0.5 2010/04/26 </span> |
| <a name="l00031"></a>00031 <span class="comment">* incorporated review comments and updated with latest CMSIS layer </span> |
| <a name="l00032"></a>00032 <span class="comment">* </span> |
| <a name="l00033"></a>00033 <span class="comment">* Version 0.0.3 2010/03/10 </span> |
| <a name="l00034"></a>00034 <span class="comment">* Initial version </span> |
| <a name="l00035"></a>00035 <span class="comment">* -------------------------------------------------------------------- */</span> |
| <a name="l00036"></a>00036 |
| <a name="l00037"></a>00037 <span class="preprocessor">#include "<a class="code" href="arm__math_8h.html">arm_math.h</a>"</span> |
| <a name="l00038"></a>00038 |
| <a name="l00066"></a><a class="code" href="group___c_f_f_t___c_i_f_f_t.html#ga8d66cdac41b8bf6cefdb895456eee84a">00066</a> <span class="keywordtype">void</span> <a class="code" href="group___c_f_f_t___c_i_f_f_t.html#ga8d66cdac41b8bf6cefdb895456eee84a" title="Processing function for the Q15 CFFT/CIFFT.">arm_cfft_radix4_q15</a>( |
| <a name="l00067"></a>00067 <span class="keyword">const</span> <a class="code" href="structarm__cfft__radix4__instance__q15.html" title="Instance structure for the Q15 CFFT/CIFFT function.">arm_cfft_radix4_instance_q15</a> * S, |
| <a name="l00068"></a>00068 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> * pSrc) |
| <a name="l00069"></a>00069 { |
| <a name="l00070"></a>00070 <span class="keywordflow">if</span>(S-><a class="code" href="structarm__cfft__radix4__instance__q15.html#a2ecff6ea735cb4d22e922d0fd5736655">ifftFlag</a> == 1u) |
| <a name="l00071"></a>00071 { |
| <a name="l00072"></a>00072 <span class="comment">/* Complex IFFT radix-4 */</span> |
| <a name="l00073"></a>00073 <a class="code" href="arm__cfft__radix4__q15_8c.html#aad04e8439d17dab5617bf1be268bb391" title="Core function for the Q15 CIFFT butterfly process.">arm_radix4_butterfly_inverse_q15</a>(pSrc, S-><a class="code" href="structarm__cfft__radix4__instance__q15.html#a5fc543e7d84ca8cb7cf6648970f21ca6">fftLen</a>, S-><a class="code" href="structarm__cfft__radix4__instance__q15.html#a29dd693537e45421a36891f8439e1fba">pTwiddle</a>, |
| <a name="l00074"></a>00074 S-><a class="code" href="structarm__cfft__radix4__instance__q15.html#af32fdc78bcc27ca385f9b76a0a1f71c3">twidCoefModifier</a>); |
| <a name="l00075"></a>00075 } |
| <a name="l00076"></a>00076 <span class="keywordflow">else</span> |
| <a name="l00077"></a>00077 { |
| <a name="l00078"></a>00078 <span class="comment">/* Complex FFT radix-4 */</span> |
| <a name="l00079"></a>00079 <a class="code" href="arm__cfft__radix4__q15_8c.html#a2d01d2045f280c32036da97d33c52440" title="Core function for the Q15 CFFT butterfly process.">arm_radix4_butterfly_q15</a>(pSrc, S-><a class="code" href="structarm__cfft__radix4__instance__q15.html#a5fc543e7d84ca8cb7cf6648970f21ca6">fftLen</a>, S-><a class="code" href="structarm__cfft__radix4__instance__q15.html#a29dd693537e45421a36891f8439e1fba">pTwiddle</a>, |
| <a name="l00080"></a>00080 S-><a class="code" href="structarm__cfft__radix4__instance__q15.html#af32fdc78bcc27ca385f9b76a0a1f71c3">twidCoefModifier</a>); |
| <a name="l00081"></a>00081 } |
| <a name="l00082"></a>00082 |
| <a name="l00083"></a>00083 <span class="keywordflow">if</span>(S-><a class="code" href="structarm__cfft__radix4__instance__q15.html#a101e3f7b0bd6b5b14cd5214f23df4133">bitReverseFlag</a> == 1u) |
| <a name="l00084"></a>00084 { |
| <a name="l00085"></a>00085 <span class="comment">/* Bit Reversal */</span> |
| <a name="l00086"></a>00086 <a class="code" href="arm__cfft__radix4__q15_8c.html#a12a07b49948c354172ae07358309a4a5" title="In-place bit reversal function.">arm_bitreversal_q15</a>(pSrc, S-><a class="code" href="structarm__cfft__radix4__instance__q15.html#a5fc543e7d84ca8cb7cf6648970f21ca6">fftLen</a>, S-><a class="code" href="structarm__cfft__radix4__instance__q15.html#a6b010e5f02d1130c621e3d2e26b95df1">bitRevFactor</a>, S-><a class="code" href="structarm__cfft__radix4__instance__q15.html#a4acf704ae0cf30b53bf0fbfae8e34a59">pBitRevTable</a>); |
| <a name="l00087"></a>00087 } |
| <a name="l00088"></a>00088 |
| <a name="l00089"></a>00089 } |
| <a name="l00090"></a>00090 |
| <a name="l00095"></a>00095 <span class="comment">/* </span> |
| <a name="l00096"></a>00096 <span class="comment">* Radix-4 FFT algorithm used is : </span> |
| <a name="l00097"></a>00097 <span class="comment">* </span> |
| <a name="l00098"></a>00098 <span class="comment">* Input real and imaginary data: </span> |
| <a name="l00099"></a>00099 <span class="comment">* x(n) = xa + j * ya </span> |
| <a name="l00100"></a>00100 <span class="comment">* x(n+N/4 ) = xb + j * yb </span> |
| <a name="l00101"></a>00101 <span class="comment">* x(n+N/2 ) = xc + j * yc </span> |
| <a name="l00102"></a>00102 <span class="comment">* x(n+3N 4) = xd + j * yd </span> |
| <a name="l00103"></a>00103 <span class="comment">* </span> |
| <a name="l00104"></a>00104 <span class="comment">* </span> |
| <a name="l00105"></a>00105 <span class="comment">* Output real and imaginary data: </span> |
| <a name="l00106"></a>00106 <span class="comment">* x(4r) = xa'+ j * ya' </span> |
| <a name="l00107"></a>00107 <span class="comment">* x(4r+1) = xb'+ j * yb' </span> |
| <a name="l00108"></a>00108 <span class="comment">* x(4r+2) = xc'+ j * yc' </span> |
| <a name="l00109"></a>00109 <span class="comment">* x(4r+3) = xd'+ j * yd' </span> |
| <a name="l00110"></a>00110 <span class="comment">* </span> |
| <a name="l00111"></a>00111 <span class="comment">* </span> |
| <a name="l00112"></a>00112 <span class="comment">* Twiddle factors for radix-4 FFT: </span> |
| <a name="l00113"></a>00113 <span class="comment">* Wn = co1 + j * (- si1) </span> |
| <a name="l00114"></a>00114 <span class="comment">* W2n = co2 + j * (- si2) </span> |
| <a name="l00115"></a>00115 <span class="comment">* W3n = co3 + j * (- si3) </span> |
| <a name="l00116"></a>00116 <span class="comment"> </span> |
| <a name="l00117"></a>00117 <span class="comment">* The real and imaginary output values for the radix-4 butterfly are </span> |
| <a name="l00118"></a>00118 <span class="comment">* xa' = xa + xb + xc + xd </span> |
| <a name="l00119"></a>00119 <span class="comment">* ya' = ya + yb + yc + yd </span> |
| <a name="l00120"></a>00120 <span class="comment">* xb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) </span> |
| <a name="l00121"></a>00121 <span class="comment">* yb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) </span> |
| <a name="l00122"></a>00122 <span class="comment">* xc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) </span> |
| <a name="l00123"></a>00123 <span class="comment">* yc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) </span> |
| <a name="l00124"></a>00124 <span class="comment">* xd' = (xa-yb-xc+yd)* co3 + (ya+xb-yc-xd)* (si3) </span> |
| <a name="l00125"></a>00125 <span class="comment">* yd' = (ya+xb-yc-xd)* co3 - (xa-yb-xc+yd)* (si3) </span> |
| <a name="l00126"></a>00126 <span class="comment">* </span> |
| <a name="l00127"></a>00127 <span class="comment">*/</span> |
| <a name="l00128"></a>00128 |
| <a name="l00138"></a><a class="code" href="arm__math_8h.html#a2d01d2045f280c32036da97d33c52440">00138</a> <span class="keywordtype">void</span> <a class="code" href="arm__cfft__radix4__q15_8c.html#a2d01d2045f280c32036da97d33c52440" title="Core function for the Q15 CFFT butterfly process.">arm_radix4_butterfly_q15</a>( |
| <a name="l00139"></a>00139 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> * pSrc16, |
| <a name="l00140"></a>00140 uint32_t fftLen, |
| <a name="l00141"></a>00141 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> * pCoef16, |
| <a name="l00142"></a>00142 uint32_t twidCoefModifier) |
| <a name="l00143"></a>00143 { |
| <a name="l00144"></a>00144 |
| <a name="l00145"></a>00145 <span class="preprocessor">#ifndef ARM_MATH_CM0</span> |
| <a name="l00146"></a>00146 <span class="preprocessor"></span> |
| <a name="l00147"></a>00147 <span class="comment">/* Run the below code for Cortex-M4 and Cortex-M3 */</span> |
| <a name="l00148"></a>00148 |
| <a name="l00149"></a>00149 <a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> R, S, T, U; |
| <a name="l00150"></a>00150 <a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> C1, C2, C3, out1, out2; |
| <a name="l00151"></a>00151 <a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *pSrc, *pCoeff; |
| <a name="l00152"></a>00152 uint32_t n1, n2, ic, i0, i1, i2, i3, j, k; |
| <a name="l00153"></a>00153 <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="l00154"></a>00154 |
| <a name="l00155"></a>00155 <span class="comment">/* Total process is divided into three stages */</span> |
| <a name="l00156"></a>00156 |
| <a name="l00157"></a>00157 <span class="comment">/* process first stage, middle stages, & last stage */</span> |
| <a name="l00158"></a>00158 |
| <a name="l00159"></a>00159 <span class="comment">/* pointer initializations for SIMD calculations */</span> |
| <a name="l00160"></a>00160 pSrc = (<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) pSrc16; |
| <a name="l00161"></a>00161 pCoeff = (<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) pCoef16; |
| <a name="l00162"></a>00162 |
| <a name="l00163"></a>00163 <span class="comment">/* Initializations for the first stage */</span> |
| <a name="l00164"></a>00164 n2 = fftLen; |
| <a name="l00165"></a>00165 n1 = n2; |
| <a name="l00166"></a>00166 |
| <a name="l00167"></a>00167 <span class="comment">/* n2 = fftLen/4 */</span> |
| <a name="l00168"></a>00168 n2 >>= 2u; |
| <a name="l00169"></a>00169 |
| <a name="l00170"></a>00170 <span class="comment">/* Index for twiddle coefficient */</span> |
| <a name="l00171"></a>00171 ic = 0u; |
| <a name="l00172"></a>00172 |
| <a name="l00173"></a>00173 <span class="comment">/* Index for input read and output write */</span> |
| <a name="l00174"></a>00174 i0 = 0u; |
| <a name="l00175"></a>00175 j = n2; |
| <a name="l00176"></a>00176 |
| <a name="l00177"></a>00177 <span class="comment">/* Input is in 1.15(q15) format */</span> |
| <a name="l00178"></a>00178 |
| <a name="l00179"></a>00179 <span class="comment">/* start of first stage process */</span> |
| <a name="l00180"></a>00180 <span class="keywordflow">do</span> |
| <a name="l00181"></a>00181 { |
| <a name="l00182"></a>00182 <span class="comment">/* Butterfly implementation */</span> |
| <a name="l00183"></a>00183 |
| <a name="l00184"></a>00184 <span class="comment">/* index calculation for the input as, */</span> |
| <a name="l00185"></a>00185 <span class="comment">/* pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */</span> |
| <a name="l00186"></a>00186 i1 = i0 + n2; |
| <a name="l00187"></a>00187 i2 = i1 + n2; |
| <a name="l00188"></a>00188 i3 = i2 + n2; |
| <a name="l00189"></a>00189 |
| <a name="l00190"></a>00190 <span class="comment">/* Reading i0, i0+fftLen/2 inputs */</span> |
| <a name="l00191"></a>00191 <span class="comment">/* Read ya (real), xa(imag) input */</span> |
| <a name="l00192"></a>00192 T = pSrc[i0]; |
| <a name="l00193"></a>00193 in = ((int16_t) (T & 0xFFFF)) >> 2; |
| <a name="l00194"></a>00194 T = ((T >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
| <a name="l00195"></a>00195 <span class="comment">/* Read yc (real), xc(imag) input */</span> |
| <a name="l00196"></a>00196 S = pSrc[i2]; |
| <a name="l00197"></a>00197 in = ((int16_t) (S & 0xFFFF)) >> 2; |
| <a name="l00198"></a>00198 S = ((S >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
| <a name="l00199"></a>00199 <span class="comment">/* R = packed((ya + yc), (xa + xc) ) */</span> |
| <a name="l00200"></a>00200 R = __QADD16(T, S); |
| <a name="l00201"></a>00201 <span class="comment">/* S = packed((ya - yc), (xa - xc) ) */</span> |
| <a name="l00202"></a>00202 S = __QSUB16(T, S); |
| <a name="l00203"></a>00203 |
| <a name="l00204"></a>00204 <span class="comment">/* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */</span> |
| <a name="l00205"></a>00205 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l00206"></a>00206 T = pSrc[i1]; |
| <a name="l00207"></a>00207 in = ((int16_t) (T & 0xFFFF)) >> 2; |
| <a name="l00208"></a>00208 T = ((T >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
| <a name="l00209"></a>00209 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l00210"></a>00210 U = pSrc[i3]; |
| <a name="l00211"></a>00211 in = ((int16_t) (U & 0xFFFF)) >> 2; |
| <a name="l00212"></a>00212 U = ((U >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
| <a name="l00213"></a>00213 <span class="comment">/* T = packed((yb + yd), (xb + xd) ) */</span> |
| <a name="l00214"></a>00214 T = __QADD16(T, U); |
| <a name="l00215"></a>00215 |
| <a name="l00216"></a>00216 <span class="comment">/* writing the butterfly processed i0 sample */</span> |
| <a name="l00217"></a>00217 <span class="comment">/* xa' = xa + xb + xc + xd */</span> |
| <a name="l00218"></a>00218 <span class="comment">/* ya' = ya + yb + yc + yd */</span> |
| <a name="l00219"></a>00219 pSrc[i0] = __SHADD16(R, T); |
| <a name="l00220"></a>00220 |
| <a name="l00221"></a>00221 <span class="comment">/* R = packed((ya + yc) - (yb + yd), (xa + xc)- (xb + xd)) */</span> |
| <a name="l00222"></a>00222 R = __QSUB16(R, T); |
| <a name="l00223"></a>00223 |
| <a name="l00224"></a>00224 <span class="comment">/* co2 & si2 are read from SIMD Coefficient pointer */</span> |
| <a name="l00225"></a>00225 C2 = pCoeff[2u * ic]; |
| <a name="l00226"></a>00226 |
| <a name="l00227"></a>00227 |
| <a name="l00228"></a>00228 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l00229"></a>00229 <span class="preprocessor"></span> |
| <a name="l00230"></a>00230 <span class="comment">/* xc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) */</span> |
| <a name="l00231"></a>00231 out1 = __SMUAD(C2, R) >> 16u; |
| <a name="l00232"></a>00232 <span class="comment">/* yc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */</span> |
| <a name="l00233"></a>00233 out2 = __SMUSDX(C2, R); |
| <a name="l00234"></a>00234 |
| <a name="l00235"></a>00235 <span class="preprocessor">#else</span> |
| <a name="l00236"></a>00236 <span class="preprocessor"></span> |
| <a name="l00237"></a>00237 <span class="comment">/* xc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */</span> |
| <a name="l00238"></a>00238 out1 = __SMUSDX(R, C2) >> 16u; |
| <a name="l00239"></a>00239 <span class="comment">/* yc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) */</span> |
| <a name="l00240"></a>00240 out2 = __SMUAD(C2, R); |
| <a name="l00241"></a>00241 |
| <a name="l00242"></a>00242 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l00243"></a>00243 |
| <a name="l00244"></a>00244 <span class="comment">/* Reading i0+fftLen/4 */</span> |
| <a name="l00245"></a>00245 <span class="comment">/* T = packed(yb, xb) */</span> |
| <a name="l00246"></a>00246 T = pSrc[i1]; |
| <a name="l00247"></a>00247 in = ((int16_t) (T & 0xFFFF)) >> 2; |
| <a name="l00248"></a>00248 T = ((T >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
| <a name="l00249"></a>00249 |
| <a name="l00250"></a>00250 <span class="comment">/* writing the butterfly processed i0 + fftLen/4 sample */</span> |
| <a name="l00251"></a>00251 <span class="comment">/* writing output(xc', yc') in little endian format */</span> |
| <a name="l00252"></a>00252 pSrc[i1] = (<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| <a name="l00253"></a>00253 |
| <a name="l00254"></a>00254 <span class="comment">/* Butterfly calculations */</span> |
| <a name="l00255"></a>00255 <span class="comment">/* U = packed(yd, xd) */</span> |
| <a name="l00256"></a>00256 U = pSrc[i3]; |
| <a name="l00257"></a>00257 in = ((int16_t) (U & 0xFFFF)) >> 2; |
| <a name="l00258"></a>00258 U = ((U >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
| <a name="l00259"></a>00259 <span class="comment">/* T = packed(yb-yd, xb-xd) */</span> |
| <a name="l00260"></a>00260 T = __QSUB16(T, U); |
| <a name="l00261"></a>00261 |
| <a name="l00262"></a>00262 |
| <a name="l00263"></a>00263 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l00264"></a>00264 <span class="preprocessor"></span> |
| <a name="l00265"></a>00265 <span class="comment">/* R = packed((ya-yc) + (xb- xd) , (xa-xc) - (yb-yd)) */</span> |
| <a name="l00266"></a>00266 R = __QASX(S, T); |
| <a name="l00267"></a>00267 <span class="comment">/* S = packed((ya-yc) - (xb- xd), (xa-xc) + (yb-yd)) */</span> |
| <a name="l00268"></a>00268 S = __QSAX(S, T); |
| <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">/* R = packed((ya-yc) + (xb- xd) , (xa-xc) - (yb-yd)) */</span> |
| <a name="l00273"></a>00273 R = __QSAX(S, T); |
| <a name="l00274"></a>00274 <span class="comment">/* S = packed((ya-yc) - (xb- xd), (xa-xc) + (yb-yd)) */</span> |
| <a name="l00275"></a>00275 S = __QASX(S, T); |
| <a name="l00276"></a>00276 |
| <a name="l00277"></a>00277 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l00278"></a>00278 |
| <a name="l00279"></a>00279 <span class="comment">/* co1 & si1 are read from SIMD Coefficient pointer */</span> |
| <a name="l00280"></a>00280 C1 = pCoeff[ic]; |
| <a name="l00281"></a>00281 <span class="comment">/* Butterfly process for the i0+fftLen/2 sample */</span> |
| <a name="l00282"></a>00282 |
| <a name="l00283"></a>00283 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l00284"></a>00284 <span class="preprocessor"></span> |
| <a name="l00285"></a>00285 <span class="comment">/* xb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) */</span> |
| <a name="l00286"></a>00286 out1 = __SMUAD(C1, S) >> 16u; |
| <a name="l00287"></a>00287 <span class="comment">/* yb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) */</span> |
| <a name="l00288"></a>00288 out2 = __SMUSDX(C1, S); |
| <a name="l00289"></a>00289 |
| <a name="l00290"></a>00290 <span class="preprocessor">#else</span> |
| <a name="l00291"></a>00291 <span class="preprocessor"></span> |
| <a name="l00292"></a>00292 <span class="comment">/* xb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) */</span> |
| <a name="l00293"></a>00293 out1 = __SMUSDX(S, C1) >> 16u; |
| <a name="l00294"></a>00294 <span class="comment">/* yb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) */</span> |
| <a name="l00295"></a>00295 out2 = __SMUAD(C1, S); |
| <a name="l00296"></a>00296 |
| <a name="l00297"></a>00297 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l00298"></a>00298 |
| <a name="l00299"></a>00299 <span class="comment">/* writing output(xb', yb') in little endian format */</span> |
| <a name="l00300"></a>00300 pSrc[i2] = ((out2) & 0xFFFF0000) | ((out1) & 0x0000FFFF); |
| <a name="l00301"></a>00301 |
| <a name="l00302"></a>00302 |
| <a name="l00303"></a>00303 <span class="comment">/* co3 & si3 are read from SIMD Coefficient pointer */</span> |
| <a name="l00304"></a>00304 C3 = pCoeff[3u * ic]; |
| <a name="l00305"></a>00305 <span class="comment">/* Butterfly process for the i0+3fftLen/4 sample */</span> |
| <a name="l00306"></a>00306 |
| <a name="l00307"></a>00307 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l00308"></a>00308 <span class="preprocessor"></span> |
| <a name="l00309"></a>00309 <span class="comment">/* xd' = (xa-yb-xc+yd)* co3 + (ya+xb-yc-xd)* (si3) */</span> |
| <a name="l00310"></a>00310 out1 = __SMUAD(C3, R) >> 16u; |
| <a name="l00311"></a>00311 <span class="comment">/* yd' = (ya+xb-yc-xd)* co3 - (xa-yb-xc+yd)* (si3) */</span> |
| <a name="l00312"></a>00312 out2 = __SMUSDX(C3, R); |
| <a name="l00313"></a>00313 |
| <a name="l00314"></a>00314 <span class="preprocessor">#else</span> |
| <a name="l00315"></a>00315 <span class="preprocessor"></span> |
| <a name="l00316"></a>00316 <span class="comment">/* xd' = (ya+xb-yc-xd)* co3 - (xa-yb-xc+yd)* (si3) */</span> |
| <a name="l00317"></a>00317 out1 = __SMUSDX(R, C3) >> 16u; |
| <a name="l00318"></a>00318 <span class="comment">/* yd' = (xa-yb-xc+yd)* co3 + (ya+xb-yc-xd)* (si3) */</span> |
| <a name="l00319"></a>00319 out2 = __SMUAD(C3, R); |
| <a name="l00320"></a>00320 |
| <a name="l00321"></a>00321 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l00322"></a>00322 |
| <a name="l00323"></a>00323 <span class="comment">/* writing output(xd', yd') in little endian format */</span> |
| <a name="l00324"></a>00324 pSrc[i3] = ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| <a name="l00325"></a>00325 |
| <a name="l00326"></a>00326 <span class="comment">/* Twiddle coefficients index modifier */</span> |
| <a name="l00327"></a>00327 ic = ic + twidCoefModifier; |
| <a name="l00328"></a>00328 |
| <a name="l00329"></a>00329 <span class="comment">/* Updating input index */</span> |
| <a name="l00330"></a>00330 i0 = i0 + 1u; |
| <a name="l00331"></a>00331 |
| <a name="l00332"></a>00332 } <span class="keywordflow">while</span>(--j); |
| <a name="l00333"></a>00333 <span class="comment">/* data is in 4.11(q11) format */</span> |
| <a name="l00334"></a>00334 |
| <a name="l00335"></a>00335 <span class="comment">/* end of first stage process */</span> |
| <a name="l00336"></a>00336 |
| <a name="l00337"></a>00337 |
| <a name="l00338"></a>00338 <span class="comment">/* start of middle stage process */</span> |
| <a name="l00339"></a>00339 |
| <a name="l00340"></a>00340 <span class="comment">/* Twiddle coefficients index modifier */</span> |
| <a name="l00341"></a>00341 twidCoefModifier <<= 2u; |
| <a name="l00342"></a>00342 |
| <a name="l00343"></a>00343 <span class="comment">/* Calculation of Middle stage */</span> |
| <a name="l00344"></a>00344 <span class="keywordflow">for</span> (k = fftLen / 4u; k > 4u; k >>= 2u) |
| <a name="l00345"></a>00345 { |
| <a name="l00346"></a>00346 <span class="comment">/* Initializations for the middle stage */</span> |
| <a name="l00347"></a>00347 n1 = n2; |
| <a name="l00348"></a>00348 n2 >>= 2u; |
| <a name="l00349"></a>00349 ic = 0u; |
| <a name="l00350"></a>00350 |
| <a name="l00351"></a>00351 <span class="keywordflow">for</span> (j = 0u; j <= (n2 - 1u); j++) |
| <a name="l00352"></a>00352 { |
| <a name="l00353"></a>00353 <span class="comment">/* index calculation for the coefficients */</span> |
| <a name="l00354"></a>00354 C1 = pCoeff[ic]; |
| <a name="l00355"></a>00355 C2 = pCoeff[2u * ic]; |
| <a name="l00356"></a>00356 C3 = pCoeff[3u * ic]; |
| <a name="l00357"></a>00357 |
| <a name="l00358"></a>00358 <span class="comment">/* Twiddle coefficients index modifier */</span> |
| <a name="l00359"></a>00359 ic = ic + twidCoefModifier; |
| <a name="l00360"></a>00360 |
| <a name="l00361"></a>00361 <span class="comment">/* Butterfly implementation */</span> |
| <a name="l00362"></a>00362 <span class="keywordflow">for</span> (i0 = j; i0 < fftLen; i0 += n1) |
| <a name="l00363"></a>00363 { |
| <a name="l00364"></a>00364 <span class="comment">/* index calculation for the input as, */</span> |
| <a name="l00365"></a>00365 <span class="comment">/* pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */</span> |
| <a name="l00366"></a>00366 i1 = i0 + n2; |
| <a name="l00367"></a>00367 i2 = i1 + n2; |
| <a name="l00368"></a>00368 i3 = i2 + n2; |
| <a name="l00369"></a>00369 |
| <a name="l00370"></a>00370 <span class="comment">/* Reading i0, i0+fftLen/2 inputs */</span> |
| <a name="l00371"></a>00371 <span class="comment">/* Read ya (real), xa(imag) input */</span> |
| <a name="l00372"></a>00372 T = pSrc[i0]; |
| <a name="l00373"></a>00373 |
| <a name="l00374"></a>00374 <span class="comment">/* Read yc (real), xc(imag) input */</span> |
| <a name="l00375"></a>00375 S = pSrc[i2]; |
| <a name="l00376"></a>00376 |
| <a name="l00377"></a>00377 <span class="comment">/* R = packed( (ya + yc), (xa + xc)) */</span> |
| <a name="l00378"></a>00378 R = __QADD16(T, S); |
| <a name="l00379"></a>00379 |
| <a name="l00380"></a>00380 <span class="comment">/* S = packed((ya - yc), (xa - xc)) */</span> |
| <a name="l00381"></a>00381 S = __QSUB16(T, S); |
| <a name="l00382"></a>00382 |
| <a name="l00383"></a>00383 <span class="comment">/* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */</span> |
| <a name="l00384"></a>00384 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l00385"></a>00385 T = pSrc[i1]; |
| <a name="l00386"></a>00386 |
| <a name="l00387"></a>00387 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l00388"></a>00388 U = pSrc[i3]; |
| <a name="l00389"></a>00389 |
| <a name="l00390"></a>00390 |
| <a name="l00391"></a>00391 <span class="comment">/* T = packed( (yb + yd), (xb + xd)) */</span> |
| <a name="l00392"></a>00392 T = __QADD16(T, U); |
| <a name="l00393"></a>00393 |
| <a name="l00394"></a>00394 |
| <a name="l00395"></a>00395 <span class="comment">/* writing the butterfly processed i0 sample */</span> |
| <a name="l00396"></a>00396 |
| <a name="l00397"></a>00397 <span class="comment">/* xa' = xa + xb + xc + xd */</span> |
| <a name="l00398"></a>00398 <span class="comment">/* ya' = ya + yb + yc + yd */</span> |
| <a name="l00399"></a>00399 out1 = __SHADD16(R, T); |
| <a name="l00400"></a>00400 in = ((int16_t) (out1 & 0xFFFF)) >> 1; |
| <a name="l00401"></a>00401 out1 = ((out1 >> 1) & 0xFFFF0000) | (in & 0xFFFF); |
| <a name="l00402"></a>00402 pSrc[i0] = out1; |
| <a name="l00403"></a>00403 |
| <a name="l00404"></a>00404 <span class="comment">/* R = packed( (ya + yc) - (yb + yd), (xa + xc) - (xb + xd)) */</span> |
| <a name="l00405"></a>00405 R = __SHSUB16(R, T); |
| <a name="l00406"></a>00406 |
| <a name="l00407"></a>00407 |
| <a name="l00408"></a>00408 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l00409"></a>00409 <span class="preprocessor"></span> |
| <a name="l00410"></a>00410 <span class="comment">/* (ya-yb+yc-yd)* (si2) + (xa-xb+xc-xd)* co2 */</span> |
| <a name="l00411"></a>00411 out1 = __SMUAD(C2, R) >> 16u; |
| <a name="l00412"></a>00412 |
| <a name="l00413"></a>00413 <span class="comment">/* (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */</span> |
| <a name="l00414"></a>00414 out2 = __SMUSDX(C2, R); |
| <a name="l00415"></a>00415 |
| <a name="l00416"></a>00416 <span class="preprocessor">#else</span> |
| <a name="l00417"></a>00417 <span class="preprocessor"></span> |
| <a name="l00418"></a>00418 <span class="comment">/* (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */</span> |
| <a name="l00419"></a>00419 out1 = __SMUSDX(R, C2) >> 16u; |
| <a name="l00420"></a>00420 |
| <a name="l00421"></a>00421 <span class="comment">/* (ya-yb+yc-yd)* (si2) + (xa-xb+xc-xd)* co2 */</span> |
| <a name="l00422"></a>00422 out2 = __SMUAD(C2, R); |
| <a name="l00423"></a>00423 |
| <a name="l00424"></a>00424 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l00425"></a>00425 |
| <a name="l00426"></a>00426 <span class="comment">/* Reading i0+3fftLen/4 */</span> |
| <a name="l00427"></a>00427 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l00428"></a>00428 T = pSrc[i1]; |
| <a name="l00429"></a>00429 |
| <a name="l00430"></a>00430 <span class="comment">/* writing the butterfly processed i0 + fftLen/4 sample */</span> |
| <a name="l00431"></a>00431 <span class="comment">/* xc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) */</span> |
| <a name="l00432"></a>00432 <span class="comment">/* yc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */</span> |
| <a name="l00433"></a>00433 pSrc[i1] = ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| <a name="l00434"></a>00434 |
| <a name="l00435"></a>00435 <span class="comment">/* Butterfly calculations */</span> |
| <a name="l00436"></a>00436 |
| <a name="l00437"></a>00437 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l00438"></a>00438 U = pSrc[i3]; |
| <a name="l00439"></a>00439 |
| <a name="l00440"></a>00440 <span class="comment">/* T = packed(yb-yd, xb-xd) */</span> |
| <a name="l00441"></a>00441 T = __QSUB16(T, U); |
| <a name="l00442"></a>00442 |
| <a name="l00443"></a>00443 |
| <a name="l00444"></a>00444 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l00445"></a>00445 <span class="preprocessor"></span> |
| <a name="l00446"></a>00446 <span class="comment">/* R = packed((ya-yc) + (xb- xd) , (xa-xc) - (yb-yd)) */</span> |
| <a name="l00447"></a>00447 R = __SHASX(S, T); |
| <a name="l00448"></a>00448 |
| <a name="l00449"></a>00449 <span class="comment">/* S = packed((ya-yc) - (xb- xd), (xa-xc) + (yb-yd)) */</span> |
| <a name="l00450"></a>00450 S = __SHSAX(S, T); |
| <a name="l00451"></a>00451 |
| <a name="l00452"></a>00452 |
| <a name="l00453"></a>00453 <span class="comment">/* Butterfly process for the i0+fftLen/2 sample */</span> |
| <a name="l00454"></a>00454 out1 = __SMUAD(C1, S) >> 16u; |
| <a name="l00455"></a>00455 out2 = __SMUSDX(C1, S); |
| <a name="l00456"></a>00456 |
| <a name="l00457"></a>00457 <span class="preprocessor">#else</span> |
| <a name="l00458"></a>00458 <span class="preprocessor"></span> |
| <a name="l00459"></a>00459 <span class="comment">/* R = packed((ya-yc) + (xb- xd) , (xa-xc) - (yb-yd)) */</span> |
| <a name="l00460"></a>00460 R = __SHSAX(S, T); |
| <a name="l00461"></a>00461 |
| <a name="l00462"></a>00462 <span class="comment">/* S = packed((ya-yc) - (xb- xd), (xa-xc) + (yb-yd)) */</span> |
| <a name="l00463"></a>00463 S = __SHASX(S, T); |
| <a name="l00464"></a>00464 |
| <a name="l00465"></a>00465 |
| <a name="l00466"></a>00466 <span class="comment">/* Butterfly process for the i0+fftLen/2 sample */</span> |
| <a name="l00467"></a>00467 out1 = __SMUSDX(S, C1) >> 16u; |
| <a name="l00468"></a>00468 out2 = __SMUAD(C1, S); |
| <a name="l00469"></a>00469 |
| <a name="l00470"></a>00470 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l00471"></a>00471 |
| <a name="l00472"></a>00472 <span class="comment">/* xb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) */</span> |
| <a name="l00473"></a>00473 <span class="comment">/* yb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) */</span> |
| <a name="l00474"></a>00474 pSrc[i2] = ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| <a name="l00475"></a>00475 |
| <a name="l00476"></a>00476 <span class="comment">/* Butterfly process for the i0+3fftLen/4 sample */</span> |
| <a name="l00477"></a>00477 |
| <a name="l00478"></a>00478 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l00479"></a>00479 <span class="preprocessor"></span> |
| <a name="l00480"></a>00480 out1 = __SMUAD(C3, R) >> 16u; |
| <a name="l00481"></a>00481 out2 = __SMUSDX(C3, R); |
| <a name="l00482"></a>00482 |
| <a name="l00483"></a>00483 <span class="preprocessor">#else</span> |
| <a name="l00484"></a>00484 <span class="preprocessor"></span> |
| <a name="l00485"></a>00485 out1 = __SMUSDX(R, C3) >> 16u; |
| <a name="l00486"></a>00486 out2 = __SMUAD(C3, R); |
| <a name="l00487"></a>00487 |
| <a name="l00488"></a>00488 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l00489"></a>00489 |
| <a name="l00490"></a>00490 <span class="comment">/* xd' = (xa-yb-xc+yd)* co3 + (ya+xb-yc-xd)* (si3) */</span> |
| <a name="l00491"></a>00491 <span class="comment">/* yd' = (ya+xb-yc-xd)* co3 - (xa-yb-xc+yd)* (si3) */</span> |
| <a name="l00492"></a>00492 pSrc[i3] = ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| <a name="l00493"></a>00493 } |
| <a name="l00494"></a>00494 } |
| <a name="l00495"></a>00495 <span class="comment">/* Twiddle coefficients index modifier */</span> |
| <a name="l00496"></a>00496 twidCoefModifier <<= 2u; |
| <a name="l00497"></a>00497 } |
| <a name="l00498"></a>00498 <span class="comment">/* end of middle stage process */</span> |
| <a name="l00499"></a>00499 |
| <a name="l00500"></a>00500 |
| <a name="l00501"></a>00501 <span class="comment">/* data is in 10.6(q6) format for the 1024 point */</span> |
| <a name="l00502"></a>00502 <span class="comment">/* data is in 8.8(q8) format for the 256 point */</span> |
| <a name="l00503"></a>00503 <span class="comment">/* data is in 6.10(q10) format for the 64 point */</span> |
| <a name="l00504"></a>00504 <span class="comment">/* data is in 4.12(q12) format for the 16 point */</span> |
| <a name="l00505"></a>00505 |
| <a name="l00506"></a>00506 <span class="comment">/* Initializations for the last stage */</span> |
| <a name="l00507"></a>00507 n1 = n2; |
| <a name="l00508"></a>00508 n2 >>= 2u; |
| <a name="l00509"></a>00509 |
| <a name="l00510"></a>00510 <span class="comment">/* start of last stage process */</span> |
| <a name="l00511"></a>00511 |
| <a name="l00512"></a>00512 <span class="comment">/* Butterfly implementation */</span> |
| <a name="l00513"></a>00513 <span class="keywordflow">for</span> (i0 = 0u; i0 <= (fftLen - n1); i0 += n1) |
| <a name="l00514"></a>00514 { |
| <a name="l00515"></a>00515 <span class="comment">/* index calculation for the input as, */</span> |
| <a name="l00516"></a>00516 <span class="comment">/* pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */</span> |
| <a name="l00517"></a>00517 i1 = i0 + n2; |
| <a name="l00518"></a>00518 i2 = i1 + n2; |
| <a name="l00519"></a>00519 i3 = i2 + n2; |
| <a name="l00520"></a>00520 |
| <a name="l00521"></a>00521 <span class="comment">/* Reading i0, i0+fftLen/2 inputs */</span> |
| <a name="l00522"></a>00522 <span class="comment">/* Read ya (real), xa(imag) input */</span> |
| <a name="l00523"></a>00523 T = pSrc[i0]; |
| <a name="l00524"></a>00524 <span class="comment">/* Read yc (real), xc(imag) input */</span> |
| <a name="l00525"></a>00525 S = pSrc[i2]; |
| <a name="l00526"></a>00526 |
| <a name="l00527"></a>00527 <span class="comment">/* R = packed((ya + yc), (xa + xc)) */</span> |
| <a name="l00528"></a>00528 R = __QADD16(T, S); |
| <a name="l00529"></a>00529 <span class="comment">/* S = packed((ya - yc), (xa - xc)) */</span> |
| <a name="l00530"></a>00530 S = __QSUB16(T, S); |
| <a name="l00531"></a>00531 |
| <a name="l00532"></a>00532 <span class="comment">/* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */</span> |
| <a name="l00533"></a>00533 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l00534"></a>00534 T = pSrc[i1]; |
| <a name="l00535"></a>00535 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l00536"></a>00536 U = pSrc[i3]; |
| <a name="l00537"></a>00537 |
| <a name="l00538"></a>00538 <span class="comment">/* T = packed((yb + yd), (xb + xd)) */</span> |
| <a name="l00539"></a>00539 T = __QADD16(T, U); |
| <a name="l00540"></a>00540 |
| <a name="l00541"></a>00541 <span class="comment">/* writing the butterfly processed i0 sample */</span> |
| <a name="l00542"></a>00542 <span class="comment">/* xa' = xa + xb + xc + xd */</span> |
| <a name="l00543"></a>00543 <span class="comment">/* ya' = ya + yb + yc + yd */</span> |
| <a name="l00544"></a>00544 pSrc[i0] = __SHADD16(R, T); |
| <a name="l00545"></a>00545 |
| <a name="l00546"></a>00546 <span class="comment">/* R = packed((ya + yc) - (yb + yd), (xa + xc) - (xb + xd)) */</span> |
| <a name="l00547"></a>00547 R = __SHSUB16(R, T); |
| <a name="l00548"></a>00548 |
| <a name="l00549"></a>00549 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l00550"></a>00550 T = pSrc[i1]; |
| <a name="l00551"></a>00551 |
| <a name="l00552"></a>00552 <span class="comment">/* writing the butterfly processed i0 + fftLen/4 sample */</span> |
| <a name="l00553"></a>00553 <span class="comment">/* xc' = (xa-xb+xc-xd) */</span> |
| <a name="l00554"></a>00554 <span class="comment">/* yc' = (ya-yb+yc-yd) */</span> |
| <a name="l00555"></a>00555 pSrc[i1] = R; |
| <a name="l00556"></a>00556 |
| <a name="l00557"></a>00557 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l00558"></a>00558 U = pSrc[i3]; |
| <a name="l00559"></a>00559 <span class="comment">/* T = packed( (yb - yd), (xb - xd)) */</span> |
| <a name="l00560"></a>00560 T = __QSUB16(T, U); |
| <a name="l00561"></a>00561 |
| <a name="l00562"></a>00562 |
| <a name="l00563"></a>00563 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l00564"></a>00564 <span class="preprocessor"></span> |
| <a name="l00565"></a>00565 <span class="comment">/* writing the butterfly processed i0 + fftLen/2 sample */</span> |
| <a name="l00566"></a>00566 <span class="comment">/* xb' = (xa+yb-xc-yd) */</span> |
| <a name="l00567"></a>00567 <span class="comment">/* yb' = (ya-xb-yc+xd) */</span> |
| <a name="l00568"></a>00568 pSrc[i2] = __SHSAX(S, T); |
| <a name="l00569"></a>00569 |
| <a name="l00570"></a>00570 <span class="comment">/* writing the butterfly processed i0 + 3fftLen/4 sample */</span> |
| <a name="l00571"></a>00571 <span class="comment">/* xd' = (xa-yb-xc+yd) */</span> |
| <a name="l00572"></a>00572 <span class="comment">/* yd' = (ya+xb-yc-xd) */</span> |
| <a name="l00573"></a>00573 pSrc[i3] = __SHASX(S, T); |
| <a name="l00574"></a>00574 |
| <a name="l00575"></a>00575 <span class="preprocessor">#else</span> |
| <a name="l00576"></a>00576 <span class="preprocessor"></span> |
| <a name="l00577"></a>00577 <span class="comment">/* writing the butterfly processed i0 + fftLen/2 sample */</span> |
| <a name="l00578"></a>00578 <span class="comment">/* xb' = (xa+yb-xc-yd) */</span> |
| <a name="l00579"></a>00579 <span class="comment">/* yb' = (ya-xb-yc+xd) */</span> |
| <a name="l00580"></a>00580 pSrc[i2] = __SHASX(S, T); |
| <a name="l00581"></a>00581 |
| <a name="l00582"></a>00582 <span class="comment">/* writing the butterfly processed i0 + 3fftLen/4 sample */</span> |
| <a name="l00583"></a>00583 <span class="comment">/* xd' = (xa-yb-xc+yd) */</span> |
| <a name="l00584"></a>00584 <span class="comment">/* yd' = (ya+xb-yc-xd) */</span> |
| <a name="l00585"></a>00585 pSrc[i3] = __SHSAX(S, T); |
| <a name="l00586"></a>00586 |
| <a name="l00587"></a>00587 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l00588"></a>00588 |
| <a name="l00589"></a>00589 } |
| <a name="l00590"></a>00590 |
| <a name="l00591"></a>00591 <span class="comment">/* end of last stage process */</span> |
| <a name="l00592"></a>00592 |
| <a name="l00593"></a>00593 <span class="comment">/* output is in 11.5(q5) format for the 1024 point */</span> |
| <a name="l00594"></a>00594 <span class="comment">/* output is in 9.7(q7) format for the 256 point */</span> |
| <a name="l00595"></a>00595 <span class="comment">/* output is in 7.9(q9) format for the 64 point */</span> |
| <a name="l00596"></a>00596 <span class="comment">/* output is in 5.11(q11) format for the 16 point */</span> |
| <a name="l00597"></a>00597 |
| <a name="l00598"></a>00598 |
| <a name="l00599"></a>00599 <span class="preprocessor">#else</span> |
| <a name="l00600"></a>00600 <span class="preprocessor"></span> |
| <a name="l00601"></a>00601 <span class="comment">/* Run the below code for Cortex-M0 */</span> |
| <a name="l00602"></a>00602 |
| <a name="l00603"></a>00603 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> R0, R1, S0, S1, T0, T1, U0, U1; |
| <a name="l00604"></a>00604 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> Co1, Si1, Co2, Si2, Co3, Si3, out1, out2; |
| <a name="l00605"></a>00605 uint32_t n1, n2, ic, i0, i1, i2, i3, j, k; |
| <a name="l00606"></a>00606 |
| <a name="l00607"></a>00607 <span class="comment">/* Total process is divided into three stages */</span> |
| <a name="l00608"></a>00608 |
| <a name="l00609"></a>00609 <span class="comment">/* process first stage, middle stages, & last stage */</span> |
| <a name="l00610"></a>00610 |
| <a name="l00611"></a>00611 <span class="comment">/* Initializations for the first stage */</span> |
| <a name="l00612"></a>00612 n2 = fftLen; |
| <a name="l00613"></a>00613 n1 = n2; |
| <a name="l00614"></a>00614 |
| <a name="l00615"></a>00615 <span class="comment">/* n2 = fftLen/4 */</span> |
| <a name="l00616"></a>00616 n2 >>= 2u; |
| <a name="l00617"></a>00617 |
| <a name="l00618"></a>00618 <span class="comment">/* Index for twiddle coefficient */</span> |
| <a name="l00619"></a>00619 ic = 0u; |
| <a name="l00620"></a>00620 |
| <a name="l00621"></a>00621 <span class="comment">/* Index for input read and output write */</span> |
| <a name="l00622"></a>00622 i0 = 0u; |
| <a name="l00623"></a>00623 j = n2; |
| <a name="l00624"></a>00624 |
| <a name="l00625"></a>00625 <span class="comment">/* Input is in 1.15(q15) format */</span> |
| <a name="l00626"></a>00626 |
| <a name="l00627"></a>00627 <span class="comment">/* start of first stage process */</span> |
| <a name="l00628"></a>00628 <span class="keywordflow">do</span> |
| <a name="l00629"></a>00629 { |
| <a name="l00630"></a>00630 <span class="comment">/* Butterfly implementation */</span> |
| <a name="l00631"></a>00631 |
| <a name="l00632"></a>00632 <span class="comment">/* index calculation for the input as, */</span> |
| <a name="l00633"></a>00633 <span class="comment">/* pSrc16[i0 + 0], pSrc16[i0 + fftLen/4], pSrc16[i0 + fftLen/2], pSrc16[i0 + 3fftLen/4] */</span> |
| <a name="l00634"></a>00634 i1 = i0 + n2; |
| <a name="l00635"></a>00635 i2 = i1 + n2; |
| <a name="l00636"></a>00636 i3 = i2 + n2; |
| <a name="l00637"></a>00637 |
| <a name="l00638"></a>00638 <span class="comment">/* Reading i0, i0+fftLen/2 inputs */</span> |
| <a name="l00639"></a>00639 |
| <a name="l00640"></a>00640 <span class="comment">/* input is down scale by 4 to avoid overflow */</span> |
| <a name="l00641"></a>00641 <span class="comment">/* Read ya (real), xa(imag) input */</span> |
| <a name="l00642"></a>00642 T0 = pSrc16[i0 * 2u] >> 2u; |
| <a name="l00643"></a>00643 T1 = pSrc16[(i0 * 2u) + 1u] >> 2u; |
| <a name="l00644"></a>00644 |
| <a name="l00645"></a>00645 <span class="comment">/* input is down scale by 4 to avoid overflow */</span> |
| <a name="l00646"></a>00646 <span class="comment">/* Read yc (real), xc(imag) input */</span> |
| <a name="l00647"></a>00647 S0 = pSrc16[i2 * 2u] >> 2u; |
| <a name="l00648"></a>00648 S1 = pSrc16[(i2 * 2u) + 1u] >> 2u; |
| <a name="l00649"></a>00649 |
| <a name="l00650"></a>00650 <span class="comment">/* R0 = (ya + yc) */</span> |
| <a name="l00651"></a>00651 R0 = __SSAT(T0 + S0, 16u); |
| <a name="l00652"></a>00652 <span class="comment">/* R1 = (xa + xc) */</span> |
| <a name="l00653"></a>00653 R1 = __SSAT(T1 + S1, 16u); |
| <a name="l00654"></a>00654 |
| <a name="l00655"></a>00655 <span class="comment">/* S0 = (ya - yc) */</span> |
| <a name="l00656"></a>00656 S0 = __SSAT(T0 - S0, 16); |
| <a name="l00657"></a>00657 <span class="comment">/* S1 = (xa - xc) */</span> |
| <a name="l00658"></a>00658 S1 = __SSAT(T1 - S1, 16); |
| <a name="l00659"></a>00659 |
| <a name="l00660"></a>00660 <span class="comment">/* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */</span> |
| <a name="l00661"></a>00661 <span class="comment">/* input is down scale by 4 to avoid overflow */</span> |
| <a name="l00662"></a>00662 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l00663"></a>00663 T0 = pSrc16[i1 * 2u] >> 2u; |
| <a name="l00664"></a>00664 T1 = pSrc16[(i1 * 2u) + 1u] >> 2u; |
| <a name="l00665"></a>00665 |
| <a name="l00666"></a>00666 <span class="comment">/* input is down scale by 4 to avoid overflow */</span> |
| <a name="l00667"></a>00667 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l00668"></a>00668 U0 = pSrc16[i3 * 2u] >> 2u; |
| <a name="l00669"></a>00669 U1 = pSrc16[(i3 * 2u) + 1] >> 2u; |
| <a name="l00670"></a>00670 |
| <a name="l00671"></a>00671 <span class="comment">/* T0 = (yb + yd) */</span> |
| <a name="l00672"></a>00672 T0 = __SSAT(T0 + U0, 16u); |
| <a name="l00673"></a>00673 <span class="comment">/* T1 = (xb + xd) */</span> |
| <a name="l00674"></a>00674 T1 = __SSAT(T1 + U1, 16u); |
| <a name="l00675"></a>00675 |
| <a name="l00676"></a>00676 <span class="comment">/* writing the butterfly processed i0 sample */</span> |
| <a name="l00677"></a>00677 <span class="comment">/* ya' = ya + yb + yc + yd */</span> |
| <a name="l00678"></a>00678 <span class="comment">/* xa' = xa + xb + xc + xd */</span> |
| <a name="l00679"></a>00679 pSrc16[i0 * 2u] = (R0 >> 1u) + (T0 >> 1u); |
| <a name="l00680"></a>00680 pSrc16[(i0 * 2u) + 1u] = (R1 >> 1u) + (T1 >> 1u); |
| <a name="l00681"></a>00681 |
| <a name="l00682"></a>00682 <span class="comment">/* R0 = (ya + yc) - (yb + yd) */</span> |
| <a name="l00683"></a>00683 <span class="comment">/* R1 = (xa + xc) - (xb + xd) */</span> |
| <a name="l00684"></a>00684 R0 = __SSAT(R0 - T0, 16u); |
| <a name="l00685"></a>00685 R1 = __SSAT(R1 - T1, 16u); |
| <a name="l00686"></a>00686 |
| <a name="l00687"></a>00687 <span class="comment">/* co2 & si2 are read from Coefficient pointer */</span> |
| <a name="l00688"></a>00688 Co2 = pCoef16[2u * ic * 2u]; |
| <a name="l00689"></a>00689 Si2 = pCoef16[(2u * ic * 2u) + 1]; |
| <a name="l00690"></a>00690 |
| <a name="l00691"></a>00691 <span class="comment">/* xc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) */</span> |
| <a name="l00692"></a>00692 out1 = (short) ((Co2 * R0 + Si2 * R1) >> 16u); |
| <a name="l00693"></a>00693 <span class="comment">/* yc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */</span> |
| <a name="l00694"></a>00694 out2 = (short) ((-Si2 * R0 + Co2 * R1) >> 16u); |
| <a name="l00695"></a>00695 |
| <a name="l00696"></a>00696 <span class="comment">/* Reading i0+fftLen/4 */</span> |
| <a name="l00697"></a>00697 <span class="comment">/* input is down scale by 4 to avoid overflow */</span> |
| <a name="l00698"></a>00698 <span class="comment">/* T0 = yb, T1 = xb */</span> |
| <a name="l00699"></a>00699 T0 = pSrc16[i1 * 2u] >> 2; |
| <a name="l00700"></a>00700 T1 = pSrc16[(i1 * 2u) + 1] >> 2; |
| <a name="l00701"></a>00701 |
| <a name="l00702"></a>00702 <span class="comment">/* writing the butterfly processed i0 + fftLen/4 sample */</span> |
| <a name="l00703"></a>00703 <span class="comment">/* writing output(xc', yc') in little endian format */</span> |
| <a name="l00704"></a>00704 pSrc16[i1 * 2u] = out1; |
| <a name="l00705"></a>00705 pSrc16[(i1 * 2u) + 1] = out2; |
| <a name="l00706"></a>00706 |
| <a name="l00707"></a>00707 <span class="comment">/* Butterfly calculations */</span> |
| <a name="l00708"></a>00708 <span class="comment">/* input is down scale by 4 to avoid overflow */</span> |
| <a name="l00709"></a>00709 <span class="comment">/* U0 = yd, U1 = xd */</span> |
| <a name="l00710"></a>00710 U0 = pSrc16[i3 * 2u] >> 2; |
| <a name="l00711"></a>00711 U1 = pSrc16[(i3 * 2u) + 1] >> 2; |
| <a name="l00712"></a>00712 <span class="comment">/* T0 = yb-yd */</span> |
| <a name="l00713"></a>00713 T0 = __SSAT(T0 - U0, 16); |
| <a name="l00714"></a>00714 <span class="comment">/* T1 = xb-xd */</span> |
| <a name="l00715"></a>00715 T1 = __SSAT(T1 - U1, 16); |
| <a name="l00716"></a>00716 |
| <a name="l00717"></a>00717 <span class="comment">/* R1 = (ya-yc) + (xb- xd), R0 = (xa-xc) - (yb-yd)) */</span> |
| <a name="l00718"></a>00718 R0 = (short) __SSAT((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) (S0 - T1), 16); |
| <a name="l00719"></a>00719 R1 = (short) __SSAT((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) (S1 + T0), 16); |
| <a name="l00720"></a>00720 |
| <a name="l00721"></a>00721 <span class="comment">/* S1 = (ya-yc) - (xb- xd), S0 = (xa-xc) + (yb-yd)) */</span> |
| <a name="l00722"></a>00722 S0 = (short) __SSAT(((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) S0 + T1), 16u); |
| <a name="l00723"></a>00723 S1 = (short) __SSAT(((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) S1 - T0), 16u); |
| <a name="l00724"></a>00724 |
| <a name="l00725"></a>00725 <span class="comment">/* co1 & si1 are read from Coefficient pointer */</span> |
| <a name="l00726"></a>00726 Co1 = pCoef16[ic * 2u]; |
| <a name="l00727"></a>00727 Si1 = pCoef16[(ic * 2u) + 1]; |
| <a name="l00728"></a>00728 <span class="comment">/* Butterfly process for the i0+fftLen/2 sample */</span> |
| <a name="l00729"></a>00729 <span class="comment">/* xb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) */</span> |
| <a name="l00730"></a>00730 out1 = (short) ((Si1 * S1 + Co1 * S0) >> 16); |
| <a name="l00731"></a>00731 <span class="comment">/* yb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) */</span> |
| <a name="l00732"></a>00732 out2 = (short) ((-Si1 * S0 + Co1 * S1) >> 16); |
| <a name="l00733"></a>00733 |
| <a name="l00734"></a>00734 <span class="comment">/* writing output(xb', yb') in little endian format */</span> |
| <a name="l00735"></a>00735 pSrc16[i2 * 2u] = out1; |
| <a name="l00736"></a>00736 pSrc16[(i2 * 2u) + 1] = out2; |
| <a name="l00737"></a>00737 |
| <a name="l00738"></a>00738 <span class="comment">/* Co3 & si3 are read from Coefficient pointer */</span> |
| <a name="l00739"></a>00739 Co3 = pCoef16[3u * (ic * 2u)]; |
| <a name="l00740"></a>00740 Si3 = pCoef16[(3u * (ic * 2u)) + 1]; |
| <a name="l00741"></a>00741 <span class="comment">/* Butterfly process for the i0+3fftLen/4 sample */</span> |
| <a name="l00742"></a>00742 <span class="comment">/* xd' = (xa-yb-xc+yd)* Co3 + (ya+xb-yc-xd)* (si3) */</span> |
| <a name="l00743"></a>00743 out1 = (short) ((Si3 * R1 + Co3 * R0) >> 16u); |
| <a name="l00744"></a>00744 <span class="comment">/* yd' = (ya+xb-yc-xd)* Co3 - (xa-yb-xc+yd)* (si3) */</span> |
| <a name="l00745"></a>00745 out2 = (short) ((-Si3 * R0 + Co3 * R1) >> 16u); |
| <a name="l00746"></a>00746 <span class="comment">/* writing output(xd', yd') in little endian format */</span> |
| <a name="l00747"></a>00747 pSrc16[i3 * 2u] = out1; |
| <a name="l00748"></a>00748 pSrc16[(i3 * 2u) + 1] = out2; |
| <a name="l00749"></a>00749 |
| <a name="l00750"></a>00750 <span class="comment">/* Twiddle coefficients index modifier */</span> |
| <a name="l00751"></a>00751 ic = ic + twidCoefModifier; |
| <a name="l00752"></a>00752 |
| <a name="l00753"></a>00753 <span class="comment">/* Updating input index */</span> |
| <a name="l00754"></a>00754 i0 = i0 + 1u; |
| <a name="l00755"></a>00755 |
| <a name="l00756"></a>00756 } <span class="keywordflow">while</span>(--j); |
| <a name="l00757"></a>00757 <span class="comment">/* data is in 4.11(q11) format */</span> |
| <a name="l00758"></a>00758 |
| <a name="l00759"></a>00759 <span class="comment">/* end of first stage process */</span> |
| <a name="l00760"></a>00760 |
| <a name="l00761"></a>00761 |
| <a name="l00762"></a>00762 <span class="comment">/* start of middle stage process */</span> |
| <a name="l00763"></a>00763 |
| <a name="l00764"></a>00764 <span class="comment">/* Twiddle coefficients index modifier */</span> |
| <a name="l00765"></a>00765 twidCoefModifier <<= 2u; |
| <a name="l00766"></a>00766 |
| <a name="l00767"></a>00767 <span class="comment">/* Calculation of Middle stage */</span> |
| <a name="l00768"></a>00768 <span class="keywordflow">for</span> (k = fftLen / 4u; k > 4u; k >>= 2u) |
| <a name="l00769"></a>00769 { |
| <a name="l00770"></a>00770 <span class="comment">/* Initializations for the middle stage */</span> |
| <a name="l00771"></a>00771 n1 = n2; |
| <a name="l00772"></a>00772 n2 >>= 2u; |
| <a name="l00773"></a>00773 ic = 0u; |
| <a name="l00774"></a>00774 |
| <a name="l00775"></a>00775 <span class="keywordflow">for</span> (j = 0u; j <= (n2 - 1u); j++) |
| <a name="l00776"></a>00776 { |
| <a name="l00777"></a>00777 <span class="comment">/* index calculation for the coefficients */</span> |
| <a name="l00778"></a>00778 Co1 = pCoef16[ic * 2u]; |
| <a name="l00779"></a>00779 Si1 = pCoef16[(ic * 2u) + 1u]; |
| <a name="l00780"></a>00780 Co2 = pCoef16[2u * (ic * 2u)]; |
| <a name="l00781"></a>00781 Si2 = pCoef16[(2u * (ic * 2u)) + 1u]; |
| <a name="l00782"></a>00782 Co3 = pCoef16[3u * (ic * 2u)]; |
| <a name="l00783"></a>00783 Si3 = pCoef16[(3u * (ic * 2u)) + 1u]; |
| <a name="l00784"></a>00784 |
| <a name="l00785"></a>00785 <span class="comment">/* Twiddle coefficients index modifier */</span> |
| <a name="l00786"></a>00786 ic = ic + twidCoefModifier; |
| <a name="l00787"></a>00787 |
| <a name="l00788"></a>00788 <span class="comment">/* Butterfly implementation */</span> |
| <a name="l00789"></a>00789 <span class="keywordflow">for</span> (i0 = j; i0 < fftLen; i0 += n1) |
| <a name="l00790"></a>00790 { |
| <a name="l00791"></a>00791 <span class="comment">/* index calculation for the input as, */</span> |
| <a name="l00792"></a>00792 <span class="comment">/* pSrc16[i0 + 0], pSrc16[i0 + fftLen/4], pSrc16[i0 + fftLen/2], pSrc16[i0 + 3fftLen/4] */</span> |
| <a name="l00793"></a>00793 i1 = i0 + n2; |
| <a name="l00794"></a>00794 i2 = i1 + n2; |
| <a name="l00795"></a>00795 i3 = i2 + n2; |
| <a name="l00796"></a>00796 |
| <a name="l00797"></a>00797 <span class="comment">/* Reading i0, i0+fftLen/2 inputs */</span> |
| <a name="l00798"></a>00798 <span class="comment">/* Read ya (real), xa(imag) input */</span> |
| <a name="l00799"></a>00799 T0 = pSrc16[i0 * 2u]; |
| <a name="l00800"></a>00800 T1 = pSrc16[(i0 * 2u) + 1u]; |
| <a name="l00801"></a>00801 |
| <a name="l00802"></a>00802 <span class="comment">/* Read yc (real), xc(imag) input */</span> |
| <a name="l00803"></a>00803 S0 = pSrc16[i2 * 2u]; |
| <a name="l00804"></a>00804 S1 = pSrc16[(i2 * 2u) + 1u]; |
| <a name="l00805"></a>00805 |
| <a name="l00806"></a>00806 <span class="comment">/* R0 = (ya + yc), R1 = (xa + xc) */</span> |
| <a name="l00807"></a>00807 R0 = __SSAT(T0 + S0, 16); |
| <a name="l00808"></a>00808 R1 = __SSAT(T1 + S1, 16); |
| <a name="l00809"></a>00809 |
| <a name="l00810"></a>00810 <span class="comment">/* S0 = (ya - yc), S1 =(xa - xc) */</span> |
| <a name="l00811"></a>00811 S0 = __SSAT(T0 - S0, 16); |
| <a name="l00812"></a>00812 S1 = __SSAT(T1 - S1, 16); |
| <a name="l00813"></a>00813 |
| <a name="l00814"></a>00814 <span class="comment">/* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */</span> |
| <a name="l00815"></a>00815 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l00816"></a>00816 T0 = pSrc16[i1 * 2u]; |
| <a name="l00817"></a>00817 T1 = pSrc16[(i1 * 2u) + 1u]; |
| <a name="l00818"></a>00818 |
| <a name="l00819"></a>00819 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l00820"></a>00820 U0 = pSrc16[i3 * 2u]; |
| <a name="l00821"></a>00821 U1 = pSrc16[(i3 * 2u) + 1u]; |
| <a name="l00822"></a>00822 |
| <a name="l00823"></a>00823 |
| <a name="l00824"></a>00824 <span class="comment">/* T0 = (yb + yd), T1 = (xb + xd) */</span> |
| <a name="l00825"></a>00825 T0 = __SSAT(T0 + U0, 16); |
| <a name="l00826"></a>00826 T1 = __SSAT(T1 + U1, 16); |
| <a name="l00827"></a>00827 |
| <a name="l00828"></a>00828 <span class="comment">/* writing the butterfly processed i0 sample */</span> |
| <a name="l00829"></a>00829 |
| <a name="l00830"></a>00830 <span class="comment">/* xa' = xa + xb + xc + xd */</span> |
| <a name="l00831"></a>00831 <span class="comment">/* ya' = ya + yb + yc + yd */</span> |
| <a name="l00832"></a>00832 out1 = ((R0 >> 1u) + (T0 >> 1u)) >> 1u; |
| <a name="l00833"></a>00833 out2 = ((R1 >> 1u) + (T1 >> 1u)) >> 1u; |
| <a name="l00834"></a>00834 |
| <a name="l00835"></a>00835 pSrc16[i0 * 2u] = out1; |
| <a name="l00836"></a>00836 pSrc16[(2u * i0) + 1u] = out2; |
| <a name="l00837"></a>00837 |
| <a name="l00838"></a>00838 <span class="comment">/* R0 = (ya + yc) - (yb + yd), R1 = (xa + xc) - (xb + xd) */</span> |
| <a name="l00839"></a>00839 R0 = (R0 >> 1u) - (T0 >> 1u); |
| <a name="l00840"></a>00840 R1 = (R1 >> 1u) - (T1 >> 1u); |
| <a name="l00841"></a>00841 |
| <a name="l00842"></a>00842 <span class="comment">/* (ya-yb+yc-yd)* (si2) + (xa-xb+xc-xd)* co2 */</span> |
| <a name="l00843"></a>00843 out1 = (short) ((Co2 * R0 + Si2 * R1) >> 16u); |
| <a name="l00844"></a>00844 |
| <a name="l00845"></a>00845 <span class="comment">/* (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */</span> |
| <a name="l00846"></a>00846 out2 = (short) ((-Si2 * R0 + Co2 * R1) >> 16u); |
| <a name="l00847"></a>00847 |
| <a name="l00848"></a>00848 <span class="comment">/* Reading i0+3fftLen/4 */</span> |
| <a name="l00849"></a>00849 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l00850"></a>00850 T0 = pSrc16[i1 * 2u]; |
| <a name="l00851"></a>00851 T1 = pSrc16[(i1 * 2u) + 1u]; |
| <a name="l00852"></a>00852 |
| <a name="l00853"></a>00853 <span class="comment">/* writing the butterfly processed i0 + fftLen/4 sample */</span> |
| <a name="l00854"></a>00854 <span class="comment">/* xc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) */</span> |
| <a name="l00855"></a>00855 <span class="comment">/* yc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */</span> |
| <a name="l00856"></a>00856 pSrc16[i1 * 2u] = out1; |
| <a name="l00857"></a>00857 pSrc16[(i1 * 2u) + 1u] = out2; |
| <a name="l00858"></a>00858 |
| <a name="l00859"></a>00859 <span class="comment">/* Butterfly calculations */</span> |
| <a name="l00860"></a>00860 |
| <a name="l00861"></a>00861 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l00862"></a>00862 U0 = pSrc16[i3 * 2u]; |
| <a name="l00863"></a>00863 U1 = pSrc16[(i3 * 2u) + 1u]; |
| <a name="l00864"></a>00864 |
| <a name="l00865"></a>00865 <span class="comment">/* T0 = yb-yd, T1 = xb-xd */</span> |
| <a name="l00866"></a>00866 T0 = __SSAT(T0 - U0, 16); |
| <a name="l00867"></a>00867 T1 = __SSAT(T1 - U1, 16); |
| <a name="l00868"></a>00868 |
| <a name="l00869"></a>00869 <span class="comment">/* R0 = (ya-yc) + (xb- xd), R1 = (xa-xc) - (yb-yd)) */</span> |
| <a name="l00870"></a>00870 R0 = (S0 >> 1u) - (T1 >> 1u); |
| <a name="l00871"></a>00871 R1 = (S1 >> 1u) + (T0 >> 1u); |
| <a name="l00872"></a>00872 |
| <a name="l00873"></a>00873 <span class="comment">/* S0 = (ya-yc) - (xb- xd), S1 = (xa-xc) + (yb-yd)) */</span> |
| <a name="l00874"></a>00874 S0 = (S0 >> 1u) + (T1 >> 1u); |
| <a name="l00875"></a>00875 S1 = (S1 >> 1u) - (T0 >> 1u); |
| <a name="l00876"></a>00876 |
| <a name="l00877"></a>00877 <span class="comment">/* Butterfly process for the i0+fftLen/2 sample */</span> |
| <a name="l00878"></a>00878 out1 = (short) ((Co1 * S0 + Si1 * S1) >> 16u); |
| <a name="l00879"></a>00879 |
| <a name="l00880"></a>00880 out2 = (short) ((-Si1 * S0 + Co1 * S1) >> 16u); |
| <a name="l00881"></a>00881 |
| <a name="l00882"></a>00882 <span class="comment">/* xb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) */</span> |
| <a name="l00883"></a>00883 <span class="comment">/* yb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) */</span> |
| <a name="l00884"></a>00884 pSrc16[i2 * 2u] = out1; |
| <a name="l00885"></a>00885 pSrc16[(i2 * 2u) + 1u] = out2; |
| <a name="l00886"></a>00886 |
| <a name="l00887"></a>00887 <span class="comment">/* Butterfly process for the i0+3fftLen/4 sample */</span> |
| <a name="l00888"></a>00888 out1 = (short) ((Si3 * R1 + Co3 * R0) >> 16u); |
| <a name="l00889"></a>00889 |
| <a name="l00890"></a>00890 out2 = (short) ((-Si3 * R0 + Co3 * R1) >> 16u); |
| <a name="l00891"></a>00891 <span class="comment">/* xd' = (xa-yb-xc+yd)* Co3 + (ya+xb-yc-xd)* (si3) */</span> |
| <a name="l00892"></a>00892 <span class="comment">/* yd' = (ya+xb-yc-xd)* Co3 - (xa-yb-xc+yd)* (si3) */</span> |
| <a name="l00893"></a>00893 pSrc16[i3 * 2u] = out1; |
| <a name="l00894"></a>00894 pSrc16[(i3 * 2u) + 1u] = out2; |
| <a name="l00895"></a>00895 } |
| <a name="l00896"></a>00896 } |
| <a name="l00897"></a>00897 <span class="comment">/* Twiddle coefficients index modifier */</span> |
| <a name="l00898"></a>00898 twidCoefModifier <<= 2u; |
| <a name="l00899"></a>00899 } |
| <a name="l00900"></a>00900 <span class="comment">/* end of middle stage process */</span> |
| <a name="l00901"></a>00901 |
| <a name="l00902"></a>00902 |
| <a name="l00903"></a>00903 <span class="comment">/* data is in 10.6(q6) format for the 1024 point */</span> |
| <a name="l00904"></a>00904 <span class="comment">/* data is in 8.8(q8) format for the 256 point */</span> |
| <a name="l00905"></a>00905 <span class="comment">/* data is in 6.10(q10) format for the 64 point */</span> |
| <a name="l00906"></a>00906 <span class="comment">/* data is in 4.12(q12) format for the 16 point */</span> |
| <a name="l00907"></a>00907 |
| <a name="l00908"></a>00908 <span class="comment">/* Initializations for the last stage */</span> |
| <a name="l00909"></a>00909 n1 = n2; |
| <a name="l00910"></a>00910 n2 >>= 2u; |
| <a name="l00911"></a>00911 |
| <a name="l00912"></a>00912 <span class="comment">/* start of last stage process */</span> |
| <a name="l00913"></a>00913 |
| <a name="l00914"></a>00914 <span class="comment">/* Butterfly implementation */</span> |
| <a name="l00915"></a>00915 <span class="keywordflow">for</span> (i0 = 0u; i0 <= (fftLen - n1); i0 += n1) |
| <a name="l00916"></a>00916 { |
| <a name="l00917"></a>00917 <span class="comment">/* index calculation for the input as, */</span> |
| <a name="l00918"></a>00918 <span class="comment">/* pSrc16[i0 + 0], pSrc16[i0 + fftLen/4], pSrc16[i0 + fftLen/2], pSrc16[i0 + 3fftLen/4] */</span> |
| <a name="l00919"></a>00919 i1 = i0 + n2; |
| <a name="l00920"></a>00920 i2 = i1 + n2; |
| <a name="l00921"></a>00921 i3 = i2 + n2; |
| <a name="l00922"></a>00922 |
| <a name="l00923"></a>00923 <span class="comment">/* Reading i0, i0+fftLen/2 inputs */</span> |
| <a name="l00924"></a>00924 <span class="comment">/* Read ya (real), xa(imag) input */</span> |
| <a name="l00925"></a>00925 T0 = pSrc16[i0 * 2u]; |
| <a name="l00926"></a>00926 T1 = pSrc16[(i0 * 2u) + 1u]; |
| <a name="l00927"></a>00927 |
| <a name="l00928"></a>00928 <span class="comment">/* Read yc (real), xc(imag) input */</span> |
| <a name="l00929"></a>00929 S0 = pSrc16[i2 * 2u]; |
| <a name="l00930"></a>00930 S1 = pSrc16[(i2 * 2u) + 1u]; |
| <a name="l00931"></a>00931 |
| <a name="l00932"></a>00932 <span class="comment">/* R0 = (ya + yc), R1 = (xa + xc) */</span> |
| <a name="l00933"></a>00933 R0 = __SSAT(T0 + S0, 16u); |
| <a name="l00934"></a>00934 R1 = __SSAT(T1 + S1, 16u); |
| <a name="l00935"></a>00935 |
| <a name="l00936"></a>00936 <span class="comment">/* S0 = (ya - yc), S1 = (xa - xc) */</span> |
| <a name="l00937"></a>00937 S0 = __SSAT(T0 - S0, 16u); |
| <a name="l00938"></a>00938 S1 = __SSAT(T1 - S1, 16u); |
| <a name="l00939"></a>00939 |
| <a name="l00940"></a>00940 <span class="comment">/* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */</span> |
| <a name="l00941"></a>00941 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l00942"></a>00942 T0 = pSrc16[i1 * 2u]; |
| <a name="l00943"></a>00943 T1 = pSrc16[(i1 * 2u) + 1u]; |
| <a name="l00944"></a>00944 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l00945"></a>00945 U0 = pSrc16[i3 * 2u]; |
| <a name="l00946"></a>00946 U1 = pSrc16[(i3 * 2u) + 1u]; |
| <a name="l00947"></a>00947 |
| <a name="l00948"></a>00948 <span class="comment">/* T0 = (yb + yd), T1 = (xb + xd)) */</span> |
| <a name="l00949"></a>00949 T0 = __SSAT(T0 + U0, 16u); |
| <a name="l00950"></a>00950 T1 = __SSAT(T1 + U1, 16u); |
| <a name="l00951"></a>00951 |
| <a name="l00952"></a>00952 <span class="comment">/* writing the butterfly processed i0 sample */</span> |
| <a name="l00953"></a>00953 <span class="comment">/* xa' = xa + xb + xc + xd */</span> |
| <a name="l00954"></a>00954 <span class="comment">/* ya' = ya + yb + yc + yd */</span> |
| <a name="l00955"></a>00955 pSrc16[i0 * 2u] = (R0 >> 1u) + (T0 >> 1u); |
| <a name="l00956"></a>00956 pSrc16[(i0 * 2u) + 1u] = (R1 >> 1u) + (T1 >> 1u); |
| <a name="l00957"></a>00957 |
| <a name="l00958"></a>00958 <span class="comment">/* R0 = (ya + yc) - (yb + yd), R1 = (xa + xc) - (xb + xd) */</span> |
| <a name="l00959"></a>00959 R0 = (R0 >> 1u) - (T0 >> 1u); |
| <a name="l00960"></a>00960 R1 = (R1 >> 1u) - (T1 >> 1u); |
| <a name="l00961"></a>00961 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l00962"></a>00962 T0 = pSrc16[i1 * 2u]; |
| <a name="l00963"></a>00963 T1 = pSrc16[(i1 * 2u) + 1u]; |
| <a name="l00964"></a>00964 |
| <a name="l00965"></a>00965 <span class="comment">/* writing the butterfly processed i0 + fftLen/4 sample */</span> |
| <a name="l00966"></a>00966 <span class="comment">/* xc' = (xa-xb+xc-xd) */</span> |
| <a name="l00967"></a>00967 <span class="comment">/* yc' = (ya-yb+yc-yd) */</span> |
| <a name="l00968"></a>00968 pSrc16[i1 * 2u] = R0; |
| <a name="l00969"></a>00969 pSrc16[(i1 * 2u) + 1u] = R1; |
| <a name="l00970"></a>00970 |
| <a name="l00971"></a>00971 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l00972"></a>00972 U0 = pSrc16[i3 * 2u]; |
| <a name="l00973"></a>00973 U1 = pSrc16[(i3 * 2u) + 1u]; |
| <a name="l00974"></a>00974 <span class="comment">/* T0 = (yb - yd), T1 = (xb - xd) */</span> |
| <a name="l00975"></a>00975 T0 = __SSAT(T0 - U0, 16u); |
| <a name="l00976"></a>00976 T1 = __SSAT(T1 - U1, 16u); |
| <a name="l00977"></a>00977 |
| <a name="l00978"></a>00978 <span class="comment">/* writing the butterfly processed i0 + fftLen/2 sample */</span> |
| <a name="l00979"></a>00979 <span class="comment">/* xb' = (xa+yb-xc-yd) */</span> |
| <a name="l00980"></a>00980 <span class="comment">/* yb' = (ya-xb-yc+xd) */</span> |
| <a name="l00981"></a>00981 pSrc16[i2 * 2u] = (S0 >> 1u) + (T1 >> 1u); |
| <a name="l00982"></a>00982 pSrc16[(i2 * 2u) + 1u] = (S1 >> 1u) - (T0 >> 1u); |
| <a name="l00983"></a>00983 |
| <a name="l00984"></a>00984 <span class="comment">/* writing the butterfly processed i0 + 3fftLen/4 sample */</span> |
| <a name="l00985"></a>00985 <span class="comment">/* xd' = (xa-yb-xc+yd) */</span> |
| <a name="l00986"></a>00986 <span class="comment">/* yd' = (ya+xb-yc-xd) */</span> |
| <a name="l00987"></a>00987 pSrc16[i3 * 2u] = (S0 >> 1u) - (T1 >> 1u); |
| <a name="l00988"></a>00988 pSrc16[(i3 * 2u) + 1u] = (S1 >> 1u) + (T0 >> 1u); |
| <a name="l00989"></a>00989 |
| <a name="l00990"></a>00990 } |
| <a name="l00991"></a>00991 |
| <a name="l00992"></a>00992 <span class="comment">/* end of last stage process */</span> |
| <a name="l00993"></a>00993 |
| <a name="l00994"></a>00994 <span class="comment">/* output is in 11.5(q5) format for the 1024 point */</span> |
| <a name="l00995"></a>00995 <span class="comment">/* output is in 9.7(q7) format for the 256 point */</span> |
| <a name="l00996"></a>00996 <span class="comment">/* output is in 7.9(q9) format for the 64 point */</span> |
| <a name="l00997"></a>00997 <span class="comment">/* output is in 5.11(q11) format for the 16 point */</span> |
| <a name="l00998"></a>00998 |
| <a name="l00999"></a>00999 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_CM0 */</span> |
| <a name="l01000"></a>01000 |
| <a name="l01001"></a>01001 } |
| <a name="l01002"></a>01002 |
| <a name="l01003"></a>01003 |
| <a name="l01013"></a>01013 <span class="comment">/* </span> |
| <a name="l01014"></a>01014 <span class="comment">* Radix-4 IFFT algorithm used is : </span> |
| <a name="l01015"></a>01015 <span class="comment">* </span> |
| <a name="l01016"></a>01016 <span class="comment">* CIFFT uses same twiddle coefficients as CFFT function </span> |
| <a name="l01017"></a>01017 <span class="comment">* x[k] = x[n] + (j)k * x[n + fftLen/4] + (-1)k * x[n+fftLen/2] + (-j)k * x[n+3*fftLen/4] </span> |
| <a name="l01018"></a>01018 <span class="comment">* </span> |
| <a name="l01019"></a>01019 <span class="comment">* </span> |
| <a name="l01020"></a>01020 <span class="comment">* IFFT is implemented with following changes in equations from FFT </span> |
| <a name="l01021"></a>01021 <span class="comment">* </span> |
| <a name="l01022"></a>01022 <span class="comment">* Input real and imaginary data: </span> |
| <a name="l01023"></a>01023 <span class="comment">* x(n) = xa + j * ya </span> |
| <a name="l01024"></a>01024 <span class="comment">* x(n+N/4 ) = xb + j * yb </span> |
| <a name="l01025"></a>01025 <span class="comment">* x(n+N/2 ) = xc + j * yc </span> |
| <a name="l01026"></a>01026 <span class="comment">* x(n+3N 4) = xd + j * yd </span> |
| <a name="l01027"></a>01027 <span class="comment">* </span> |
| <a name="l01028"></a>01028 <span class="comment">* </span> |
| <a name="l01029"></a>01029 <span class="comment">* Output real and imaginary data: </span> |
| <a name="l01030"></a>01030 <span class="comment">* x(4r) = xa'+ j * ya' </span> |
| <a name="l01031"></a>01031 <span class="comment">* x(4r+1) = xb'+ j * yb' </span> |
| <a name="l01032"></a>01032 <span class="comment">* x(4r+2) = xc'+ j * yc' </span> |
| <a name="l01033"></a>01033 <span class="comment">* x(4r+3) = xd'+ j * yd' </span> |
| <a name="l01034"></a>01034 <span class="comment">* </span> |
| <a name="l01035"></a>01035 <span class="comment">* </span> |
| <a name="l01036"></a>01036 <span class="comment">* Twiddle factors for radix-4 IFFT: </span> |
| <a name="l01037"></a>01037 <span class="comment">* Wn = co1 + j * (si1) </span> |
| <a name="l01038"></a>01038 <span class="comment">* W2n = co2 + j * (si2) </span> |
| <a name="l01039"></a>01039 <span class="comment">* W3n = co3 + j * (si3) </span> |
| <a name="l01040"></a>01040 <span class="comment"> </span> |
| <a name="l01041"></a>01041 <span class="comment">* The real and imaginary output values for the radix-4 butterfly are </span> |
| <a name="l01042"></a>01042 <span class="comment">* xa' = xa + xb + xc + xd </span> |
| <a name="l01043"></a>01043 <span class="comment">* ya' = ya + yb + yc + yd </span> |
| <a name="l01044"></a>01044 <span class="comment">* xb' = (xa-yb-xc+yd)* co1 - (ya+xb-yc-xd)* (si1) </span> |
| <a name="l01045"></a>01045 <span class="comment">* yb' = (ya+xb-yc-xd)* co1 + (xa-yb-xc+yd)* (si1) </span> |
| <a name="l01046"></a>01046 <span class="comment">* xc' = (xa-xb+xc-xd)* co2 - (ya-yb+yc-yd)* (si2) </span> |
| <a name="l01047"></a>01047 <span class="comment">* yc' = (ya-yb+yc-yd)* co2 + (xa-xb+xc-xd)* (si2) </span> |
| <a name="l01048"></a>01048 <span class="comment">* xd' = (xa+yb-xc-yd)* co3 - (ya-xb-yc+xd)* (si3) </span> |
| <a name="l01049"></a>01049 <span class="comment">* yd' = (ya-xb-yc+xd)* co3 + (xa+yb-xc-yd)* (si3) </span> |
| <a name="l01050"></a>01050 <span class="comment">* </span> |
| <a name="l01051"></a>01051 <span class="comment">*/</span> |
| <a name="l01052"></a>01052 |
| <a name="l01053"></a><a class="code" href="arm__math_8h.html#aad04e8439d17dab5617bf1be268bb391">01053</a> <span class="keywordtype">void</span> <a class="code" href="arm__cfft__radix4__q15_8c.html#aad04e8439d17dab5617bf1be268bb391" title="Core function for the Q15 CIFFT butterfly process.">arm_radix4_butterfly_inverse_q15</a>( |
| <a name="l01054"></a>01054 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> * pSrc16, |
| <a name="l01055"></a>01055 uint32_t fftLen, |
| <a name="l01056"></a>01056 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> * pCoef16, |
| <a name="l01057"></a>01057 uint32_t twidCoefModifier) |
| <a name="l01058"></a>01058 { |
| <a name="l01059"></a>01059 |
| <a name="l01060"></a>01060 <span class="preprocessor">#ifndef ARM_MATH_CM0</span> |
| <a name="l01061"></a>01061 <span class="preprocessor"></span> |
| <a name="l01062"></a>01062 <span class="comment">/* Run the below code for Cortex-M4 and Cortex-M3 */</span> |
| <a name="l01063"></a>01063 |
| <a name="l01064"></a>01064 <a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> R, S, T, U; |
| <a name="l01065"></a>01065 <a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> C1, C2, C3, out1, out2; |
| <a name="l01066"></a>01066 <a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *pSrc, *pCoeff; |
| <a name="l01067"></a>01067 uint32_t n1, n2, ic, i0, i1, i2, i3, j, k; |
| <a name="l01068"></a>01068 <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="l01069"></a>01069 |
| <a name="l01070"></a>01070 <span class="comment">/* Total process is divided into three stages */</span> |
| <a name="l01071"></a>01071 |
| <a name="l01072"></a>01072 <span class="comment">/* process first stage, middle stages, & last stage */</span> |
| <a name="l01073"></a>01073 |
| <a name="l01074"></a>01074 <span class="comment">/* pointer initializations for SIMD calculations */</span> |
| <a name="l01075"></a>01075 pSrc = (<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) pSrc16; |
| <a name="l01076"></a>01076 pCoeff = (<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) pCoef16; |
| <a name="l01077"></a>01077 |
| <a name="l01078"></a>01078 <span class="comment">/* Initializations for the first stage */</span> |
| <a name="l01079"></a>01079 n2 = fftLen; |
| <a name="l01080"></a>01080 n1 = n2; |
| <a name="l01081"></a>01081 |
| <a name="l01082"></a>01082 <span class="comment">/* n2 = fftLen/4 */</span> |
| <a name="l01083"></a>01083 n2 >>= 2u; |
| <a name="l01084"></a>01084 |
| <a name="l01085"></a>01085 <span class="comment">/* Index for twiddle coefficient */</span> |
| <a name="l01086"></a>01086 ic = 0u; |
| <a name="l01087"></a>01087 |
| <a name="l01088"></a>01088 <span class="comment">/* Index for input read and output write */</span> |
| <a name="l01089"></a>01089 i0 = 0u; |
| <a name="l01090"></a>01090 |
| <a name="l01091"></a>01091 j = n2; |
| <a name="l01092"></a>01092 |
| <a name="l01093"></a>01093 <span class="comment">/* Input is in 1.15(q15) format */</span> |
| <a name="l01094"></a>01094 |
| <a name="l01095"></a>01095 <span class="comment">/* Start of first stage process */</span> |
| <a name="l01096"></a>01096 <span class="keywordflow">do</span> |
| <a name="l01097"></a>01097 { |
| <a name="l01098"></a>01098 <span class="comment">/* Butterfly implementation */</span> |
| <a name="l01099"></a>01099 |
| <a name="l01100"></a>01100 <span class="comment">/* index calculation for the input as, */</span> |
| <a name="l01101"></a>01101 <span class="comment">/* pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */</span> |
| <a name="l01102"></a>01102 i1 = i0 + n2; |
| <a name="l01103"></a>01103 i2 = i1 + n2; |
| <a name="l01104"></a>01104 i3 = i2 + n2; |
| <a name="l01105"></a>01105 |
| <a name="l01106"></a>01106 <span class="comment">/* Reading i0, i0+fftLen/2 inputs */</span> |
| <a name="l01107"></a>01107 <span class="comment">/* Read ya (real), xa(imag) input */</span> |
| <a name="l01108"></a>01108 T = pSrc[i0]; |
| <a name="l01109"></a>01109 in = ((int16_t) (T & 0xFFFF)) >> 2; |
| <a name="l01110"></a>01110 T = ((T >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
| <a name="l01111"></a>01111 <span class="comment">/* Read yc (real), xc(imag) input */</span> |
| <a name="l01112"></a>01112 S = pSrc[i2]; |
| <a name="l01113"></a>01113 in = ((int16_t) (S & 0xFFFF)) >> 2; |
| <a name="l01114"></a>01114 S = ((S >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
| <a name="l01115"></a>01115 |
| <a name="l01116"></a>01116 <span class="comment">/* R = packed((ya + yc), (xa + xc) ) */</span> |
| <a name="l01117"></a>01117 R = __QADD16(T, S); |
| <a name="l01118"></a>01118 <span class="comment">/* S = packed((ya - yc), (xa - xc) ) */</span> |
| <a name="l01119"></a>01119 S = __QSUB16(T, S); |
| <a name="l01120"></a>01120 |
| <a name="l01121"></a>01121 <span class="comment">/* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */</span> |
| <a name="l01122"></a>01122 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l01123"></a>01123 T = pSrc[i1]; |
| <a name="l01124"></a>01124 in = ((int16_t) (T & 0xFFFF)) >> 2; |
| <a name="l01125"></a>01125 T = ((T >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
| <a name="l01126"></a>01126 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l01127"></a>01127 U = pSrc[i3]; |
| <a name="l01128"></a>01128 in = ((int16_t) (U & 0xFFFF)) >> 2; |
| <a name="l01129"></a>01129 U = ((U >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
| <a name="l01130"></a>01130 |
| <a name="l01131"></a>01131 <span class="comment">/* T = packed((yb + yd), (xb + xd) ) */</span> |
| <a name="l01132"></a>01132 T = __QADD16(T, U); |
| <a name="l01133"></a>01133 |
| <a name="l01134"></a>01134 <span class="comment">/* writing the butterfly processed i0 sample */</span> |
| <a name="l01135"></a>01135 <span class="comment">/* xa' = xa + xb + xc + xd */</span> |
| <a name="l01136"></a>01136 <span class="comment">/* ya' = ya + yb + yc + yd */</span> |
| <a name="l01137"></a>01137 pSrc[i0] = __SHADD16(R, T); |
| <a name="l01138"></a>01138 |
| <a name="l01139"></a>01139 <span class="comment">/* R = packed((ya + yc) - (yb + yd), (xa + xc)- (xb + xd)) */</span> |
| <a name="l01140"></a>01140 R = __QSUB16(R, T); |
| <a name="l01141"></a>01141 <span class="comment">/* co2 & si2 are read from SIMD Coefficient pointer */</span> |
| <a name="l01142"></a>01142 C2 = pCoeff[2u * ic]; |
| <a name="l01143"></a>01143 |
| <a name="l01144"></a>01144 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l01145"></a>01145 <span class="preprocessor"></span> |
| <a name="l01146"></a>01146 <span class="comment">/* xc' = (xa-xb+xc-xd)* co2 - (ya-yb+yc-yd)* (si2) */</span> |
| <a name="l01147"></a>01147 out1 = __SMUSD(C2, R) >> 16u; |
| <a name="l01148"></a>01148 <span class="comment">/* yc' = (ya-yb+yc-yd)* co2 + (xa-xb+xc-xd)* (si2) */</span> |
| <a name="l01149"></a>01149 out2 = __SMUADX(C2, R); |
| <a name="l01150"></a>01150 |
| <a name="l01151"></a>01151 <span class="preprocessor">#else</span> |
| <a name="l01152"></a>01152 <span class="preprocessor"></span> |
| <a name="l01153"></a>01153 <span class="comment">/* xc' = (ya-yb+yc-yd)* co2 + (xa-xb+xc-xd)* (si2) */</span> |
| <a name="l01154"></a>01154 out1 = __SMUADX(C2, R) >> 16u; |
| <a name="l01155"></a>01155 <span class="comment">/* yc' = (xa-xb+xc-xd)* co2 - (ya-yb+yc-yd)* (si2) */</span> |
| <a name="l01156"></a>01156 out2 = __SMUSD(-C2, R); |
| <a name="l01157"></a>01157 |
| <a name="l01158"></a>01158 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l01159"></a>01159 |
| <a name="l01160"></a>01160 <span class="comment">/* Reading i0+fftLen/4 */</span> |
| <a name="l01161"></a>01161 <span class="comment">/* T = packed(yb, xb) */</span> |
| <a name="l01162"></a>01162 T = pSrc[i1]; |
| <a name="l01163"></a>01163 in = ((int16_t) (T & 0xFFFF)) >> 2; |
| <a name="l01164"></a>01164 T = ((T >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
| <a name="l01165"></a>01165 |
| <a name="l01166"></a>01166 <span class="comment">/* writing the butterfly processed i0 + fftLen/4 sample */</span> |
| <a name="l01167"></a>01167 <span class="comment">/* writing output(xc', yc') in little endian format */</span> |
| <a name="l01168"></a>01168 pSrc[i1] = (<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| <a name="l01169"></a>01169 |
| <a name="l01170"></a>01170 <span class="comment">/* Butterfly calculations */</span> |
| <a name="l01171"></a>01171 <span class="comment">/* U = packed(yd, xd) */</span> |
| <a name="l01172"></a>01172 U = pSrc[i3]; |
| <a name="l01173"></a>01173 in = ((int16_t) (U & 0xFFFF)) >> 2; |
| <a name="l01174"></a>01174 U = ((U >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
| <a name="l01175"></a>01175 |
| <a name="l01176"></a>01176 <span class="comment">/* T = packed(yb-yd, xb-xd) */</span> |
| <a name="l01177"></a>01177 T = __QSUB16(T, U); |
| <a name="l01178"></a>01178 |
| <a name="l01179"></a>01179 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l01180"></a>01180 <span class="preprocessor"></span> |
| <a name="l01181"></a>01181 <span class="comment">/* R = packed((ya-yc) - (xb- xd) , (xa-xc) + (yb-yd)) */</span> |
| <a name="l01182"></a>01182 R = __QSAX(S, T); |
| <a name="l01183"></a>01183 <span class="comment">/* S = packed((ya-yc) + (xb- xd), (xa-xc) - (yb-yd)) */</span> |
| <a name="l01184"></a>01184 S = __QASX(S, T); |
| <a name="l01185"></a>01185 |
| <a name="l01186"></a>01186 <span class="preprocessor">#else</span> |
| <a name="l01187"></a>01187 <span class="preprocessor"></span> |
| <a name="l01188"></a>01188 <span class="comment">/* R = packed((ya-yc) - (xb- xd) , (xa-xc) + (yb-yd)) */</span> |
| <a name="l01189"></a>01189 R = __QASX(S, T); |
| <a name="l01190"></a>01190 <span class="comment">/* S = packed((ya-yc) + (xb- xd), (xa-xc) - (yb-yd)) */</span> |
| <a name="l01191"></a>01191 S = __QSAX(S, T); |
| <a name="l01192"></a>01192 |
| <a name="l01193"></a>01193 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l01194"></a>01194 |
| <a name="l01195"></a>01195 <span class="comment">/* co1 & si1 are read from SIMD Coefficient pointer */</span> |
| <a name="l01196"></a>01196 C1 = pCoeff[ic]; |
| <a name="l01197"></a>01197 <span class="comment">/* Butterfly process for the i0+fftLen/2 sample */</span> |
| <a name="l01198"></a>01198 |
| <a name="l01199"></a>01199 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l01200"></a>01200 <span class="preprocessor"></span> |
| <a name="l01201"></a>01201 <span class="comment">/* xb' = (xa-yb-xc+yd)* co1 - (ya+xb-yc-xd)* (si1) */</span> |
| <a name="l01202"></a>01202 out1 = __SMUSD(C1, S) >> 16u; |
| <a name="l01203"></a>01203 <span class="comment">/* yb' = (ya+xb-yc-xd)* co1 + (xa-yb-xc+yd)* (si1) */</span> |
| <a name="l01204"></a>01204 out2 = __SMUADX(C1, S); |
| <a name="l01205"></a>01205 |
| <a name="l01206"></a>01206 <span class="preprocessor">#else</span> |
| <a name="l01207"></a>01207 <span class="preprocessor"></span> |
| <a name="l01208"></a>01208 <span class="comment">/* xb' = (ya+xb-yc-xd)* co1 + (xa-yb-xc+yd)* (si1) */</span> |
| <a name="l01209"></a>01209 out1 = __SMUADX(C1, S) >> 16u; |
| <a name="l01210"></a>01210 <span class="comment">/* yb' = (xa-yb-xc+yd)* co1 - (ya+xb-yc-xd)* (si1) */</span> |
| <a name="l01211"></a>01211 out2 = __SMUSD(-C1, S); |
| <a name="l01212"></a>01212 |
| <a name="l01213"></a>01213 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l01214"></a>01214 |
| <a name="l01215"></a>01215 <span class="comment">/* writing output(xb', yb') in little endian format */</span> |
| <a name="l01216"></a>01216 pSrc[i2] = ((out2) & 0xFFFF0000) | ((out1) & 0x0000FFFF); |
| <a name="l01217"></a>01217 |
| <a name="l01218"></a>01218 <span class="comment">/* co3 & si3 are read from SIMD Coefficient pointer */</span> |
| <a name="l01219"></a>01219 C3 = pCoeff[3u * ic]; |
| <a name="l01220"></a>01220 <span class="comment">/* Butterfly process for the i0+3fftLen/4 sample */</span> |
| <a name="l01221"></a>01221 |
| <a name="l01222"></a>01222 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l01223"></a>01223 <span class="preprocessor"></span> |
| <a name="l01224"></a>01224 <span class="comment">/* xd' = (xa+yb-xc-yd)* co3 - (ya-xb-yc+xd)* (si3) */</span> |
| <a name="l01225"></a>01225 out1 = __SMUSD(C3, R) >> 16u; |
| <a name="l01226"></a>01226 <span class="comment">/* yd' = (ya-xb-yc+xd)* co3 + (xa+yb-xc-yd)* (si3) */</span> |
| <a name="l01227"></a>01227 out2 = __SMUADX(C3, R); |
| <a name="l01228"></a>01228 |
| <a name="l01229"></a>01229 <span class="preprocessor">#else</span> |
| <a name="l01230"></a>01230 <span class="preprocessor"></span> |
| <a name="l01231"></a>01231 <span class="comment">/* xd' = (ya-xb-yc+xd)* co3 + (xa+yb-xc-yd)* (si3) */</span> |
| <a name="l01232"></a>01232 out1 = __SMUADX(C3, R) >> 16u; |
| <a name="l01233"></a>01233 <span class="comment">/* yd' = (xa+yb-xc-yd)* co3 - (ya-xb-yc+xd)* (si3) */</span> |
| <a name="l01234"></a>01234 out2 = __SMUSD(-C3, R); |
| <a name="l01235"></a>01235 |
| <a name="l01236"></a>01236 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l01237"></a>01237 |
| <a name="l01238"></a>01238 <span class="comment">/* writing output(xd', yd') in little endian format */</span> |
| <a name="l01239"></a>01239 pSrc[i3] = ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| <a name="l01240"></a>01240 |
| <a name="l01241"></a>01241 <span class="comment">/* Twiddle coefficients index modifier */</span> |
| <a name="l01242"></a>01242 ic = ic + twidCoefModifier; |
| <a name="l01243"></a>01243 |
| <a name="l01244"></a>01244 <span class="comment">/* Updating input index */</span> |
| <a name="l01245"></a>01245 i0 = i0 + 1u; |
| <a name="l01246"></a>01246 |
| <a name="l01247"></a>01247 } <span class="keywordflow">while</span>(--j); |
| <a name="l01248"></a>01248 |
| <a name="l01249"></a>01249 <span class="comment">/* End of first stage process */</span> |
| <a name="l01250"></a>01250 |
| <a name="l01251"></a>01251 <span class="comment">/* data is in 4.11(q11) format */</span> |
| <a name="l01252"></a>01252 |
| <a name="l01253"></a>01253 |
| <a name="l01254"></a>01254 <span class="comment">/* Start of Middle stage process */</span> |
| <a name="l01255"></a>01255 |
| <a name="l01256"></a>01256 <span class="comment">/* Twiddle coefficients index modifier */</span> |
| <a name="l01257"></a>01257 twidCoefModifier <<= 2u; |
| <a name="l01258"></a>01258 |
| <a name="l01259"></a>01259 <span class="comment">/* Calculation of Middle stage */</span> |
| <a name="l01260"></a>01260 <span class="keywordflow">for</span> (k = fftLen / 4u; k > 4u; k >>= 2u) |
| <a name="l01261"></a>01261 { |
| <a name="l01262"></a>01262 <span class="comment">/* Initializations for the middle stage */</span> |
| <a name="l01263"></a>01263 n1 = n2; |
| <a name="l01264"></a>01264 n2 >>= 2u; |
| <a name="l01265"></a>01265 ic = 0u; |
| <a name="l01266"></a>01266 |
| <a name="l01267"></a>01267 <span class="keywordflow">for</span> (j = 0u; j <= (n2 - 1u); j++) |
| <a name="l01268"></a>01268 { |
| <a name="l01269"></a>01269 <span class="comment">/* index calculation for the coefficients */</span> |
| <a name="l01270"></a>01270 C1 = pCoeff[ic]; |
| <a name="l01271"></a>01271 C2 = pCoeff[2u * ic]; |
| <a name="l01272"></a>01272 C3 = pCoeff[3u * ic]; |
| <a name="l01273"></a>01273 |
| <a name="l01274"></a>01274 <span class="comment">/* Twiddle coefficients index modifier */</span> |
| <a name="l01275"></a>01275 ic = ic + twidCoefModifier; |
| <a name="l01276"></a>01276 |
| <a name="l01277"></a>01277 <span class="comment">/* Butterfly implementation */</span> |
| <a name="l01278"></a>01278 <span class="keywordflow">for</span> (i0 = j; i0 < fftLen; i0 += n1) |
| <a name="l01279"></a>01279 { |
| <a name="l01280"></a>01280 <span class="comment">/* index calculation for the input as, */</span> |
| <a name="l01281"></a>01281 <span class="comment">/* pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */</span> |
| <a name="l01282"></a>01282 i1 = i0 + n2; |
| <a name="l01283"></a>01283 i2 = i1 + n2; |
| <a name="l01284"></a>01284 i3 = i2 + n2; |
| <a name="l01285"></a>01285 |
| <a name="l01286"></a>01286 <span class="comment">/* Reading i0, i0+fftLen/2 inputs */</span> |
| <a name="l01287"></a>01287 <span class="comment">/* Read ya (real), xa(imag) input */</span> |
| <a name="l01288"></a>01288 T = pSrc[i0]; |
| <a name="l01289"></a>01289 |
| <a name="l01290"></a>01290 <span class="comment">/* Read yc (real), xc(imag) input */</span> |
| <a name="l01291"></a>01291 S = pSrc[i2]; |
| <a name="l01292"></a>01292 |
| <a name="l01293"></a>01293 |
| <a name="l01294"></a>01294 <span class="comment">/* R = packed( (ya + yc), (xa + xc)) */</span> |
| <a name="l01295"></a>01295 R = __QADD16(T, S); |
| <a name="l01296"></a>01296 <span class="comment">/* S = packed((ya - yc), (xa - xc)) */</span> |
| <a name="l01297"></a>01297 S = __QSUB16(T, S); |
| <a name="l01298"></a>01298 |
| <a name="l01299"></a>01299 <span class="comment">/* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */</span> |
| <a name="l01300"></a>01300 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l01301"></a>01301 T = pSrc[i1]; |
| <a name="l01302"></a>01302 |
| <a name="l01303"></a>01303 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l01304"></a>01304 U = pSrc[i3]; |
| <a name="l01305"></a>01305 |
| <a name="l01306"></a>01306 |
| <a name="l01307"></a>01307 <span class="comment">/* T = packed( (yb + yd), (xb + xd)) */</span> |
| <a name="l01308"></a>01308 T = __QADD16(T, U); |
| <a name="l01309"></a>01309 |
| <a name="l01310"></a>01310 <span class="comment">/* writing the butterfly processed i0 sample */</span> |
| <a name="l01311"></a>01311 <span class="comment">/* xa' = xa + xb + xc + xd */</span> |
| <a name="l01312"></a>01312 <span class="comment">/* ya' = ya + yb + yc + yd */</span> |
| <a name="l01313"></a>01313 out1 = __SHADD16(R, T); |
| <a name="l01314"></a>01314 in = ((int16_t) (out1 & 0xFFFF)) >> 1; |
| <a name="l01315"></a>01315 out1 = ((out1 >> 1) & 0xFFFF0000) | (in & 0xFFFF); |
| <a name="l01316"></a>01316 pSrc[i0] = out1; |
| <a name="l01317"></a>01317 |
| <a name="l01318"></a>01318 |
| <a name="l01319"></a>01319 |
| <a name="l01320"></a>01320 <span class="comment">/* R = packed( (ya + yc) - (yb + yd), (xa + xc) - (xb + xd)) */</span> |
| <a name="l01321"></a>01321 R = __SHSUB16(R, T); |
| <a name="l01322"></a>01322 |
| <a name="l01323"></a>01323 |
| <a name="l01324"></a>01324 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l01325"></a>01325 <span class="preprocessor"></span> |
| <a name="l01326"></a>01326 <span class="comment">/* (ya-yb+yc-yd)* (si2) - (xa-xb+xc-xd)* co2 */</span> |
| <a name="l01327"></a>01327 out1 = __SMUSD(C2, R) >> 16u; |
| <a name="l01328"></a>01328 <span class="comment">/* (ya-yb+yc-yd)* co2 + (xa-xb+xc-xd)* (si2) */</span> |
| <a name="l01329"></a>01329 out2 = __SMUADX(C2, R); |
| <a name="l01330"></a>01330 |
| <a name="l01331"></a>01331 <span class="preprocessor">#else</span> |
| <a name="l01332"></a>01332 <span class="preprocessor"></span> |
| <a name="l01333"></a>01333 <span class="comment">/* (ya-yb+yc-yd)* co2 + (xa-xb+xc-xd)* (si2) */</span> |
| <a name="l01334"></a>01334 out1 = __SMUADX(R, C2) >> 16u; |
| <a name="l01335"></a>01335 <span class="comment">/* (ya-yb+yc-yd)* (si2) - (xa-xb+xc-xd)* co2 */</span> |
| <a name="l01336"></a>01336 out2 = __SMUSD(-C2, R); |
| <a name="l01337"></a>01337 |
| <a name="l01338"></a>01338 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l01339"></a>01339 |
| <a name="l01340"></a>01340 <span class="comment">/* Reading i0+3fftLen/4 */</span> |
| <a name="l01341"></a>01341 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l01342"></a>01342 T = pSrc[i1]; |
| <a name="l01343"></a>01343 |
| <a name="l01344"></a>01344 <span class="comment">/* writing the butterfly processed i0 + fftLen/4 sample */</span> |
| <a name="l01345"></a>01345 <span class="comment">/* xc' = (xa-xb+xc-xd)* co2 - (ya-yb+yc-yd)* (si2) */</span> |
| <a name="l01346"></a>01346 <span class="comment">/* yc' = (ya-yb+yc-yd)* co2 + (xa-xb+xc-xd)* (si2) */</span> |
| <a name="l01347"></a>01347 pSrc[i1] = ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| <a name="l01348"></a>01348 |
| <a name="l01349"></a>01349 <span class="comment">/* Butterfly calculations */</span> |
| <a name="l01350"></a>01350 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l01351"></a>01351 U = pSrc[i3]; |
| <a name="l01352"></a>01352 |
| <a name="l01353"></a>01353 <span class="comment">/* T = packed(yb-yd, xb-xd) */</span> |
| <a name="l01354"></a>01354 T = __QSUB16(T, U); |
| <a name="l01355"></a>01355 |
| <a name="l01356"></a>01356 |
| <a name="l01357"></a>01357 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l01358"></a>01358 <span class="preprocessor"></span> |
| <a name="l01359"></a>01359 <span class="comment">/* R = packed((ya-yc) - (xb- xd) , (xa-xc) + (yb-yd)) */</span> |
| <a name="l01360"></a>01360 R = __SHSAX(S, T); |
| <a name="l01361"></a>01361 |
| <a name="l01362"></a>01362 <span class="comment">/* S = packed((ya-yc) + (xb- xd), (xa-xc) - (yb-yd)) */</span> |
| <a name="l01363"></a>01363 S = __SHASX(S, T); |
| <a name="l01364"></a>01364 <span class="comment">/* Butterfly process for the i0+fftLen/2 sample */</span> |
| <a name="l01365"></a>01365 out1 = __SMUSD(C1, S) >> 16u; |
| <a name="l01366"></a>01366 out2 = __SMUADX(C1, S); |
| <a name="l01367"></a>01367 |
| <a name="l01368"></a>01368 <span class="preprocessor">#else</span> |
| <a name="l01369"></a>01369 <span class="preprocessor"></span> |
| <a name="l01370"></a>01370 <span class="comment">/* R = packed((ya-yc) - (xb- xd) , (xa-xc) + (yb-yd)) */</span> |
| <a name="l01371"></a>01371 R = __SHASX(S, T); |
| <a name="l01372"></a>01372 |
| <a name="l01373"></a>01373 <span class="comment">/* S = packed((ya-yc) + (xb- xd), (xa-xc) - (yb-yd)) */</span> |
| <a name="l01374"></a>01374 S = __SHSAX(S, T); |
| <a name="l01375"></a>01375 <span class="comment">/* Butterfly process for the i0+fftLen/2 sample */</span> |
| <a name="l01376"></a>01376 out1 = __SMUADX(S, C1) >> 16u; |
| <a name="l01377"></a>01377 out2 = __SMUSD(-C1, S); |
| <a name="l01378"></a>01378 |
| <a name="l01379"></a>01379 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l01380"></a>01380 |
| <a name="l01381"></a>01381 <span class="comment">/* xb' = (xa-yb-xc+yd)* co1 - (ya+xb-yc-xd)* (si1) */</span> |
| <a name="l01382"></a>01382 <span class="comment">/* yb' = (ya+xb-yc-xd)* co1 + (xa-yb-xc+yd)* (si1) */</span> |
| <a name="l01383"></a>01383 pSrc[i2] = ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| <a name="l01384"></a>01384 |
| <a name="l01385"></a>01385 <span class="comment">/* Butterfly process for the i0+3fftLen/4 sample */</span> |
| <a name="l01386"></a>01386 |
| <a name="l01387"></a>01387 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l01388"></a>01388 <span class="preprocessor"></span> |
| <a name="l01389"></a>01389 out1 = __SMUSD(C3, R) >> 16u; |
| <a name="l01390"></a>01390 out2 = __SMUADX(C3, R); |
| <a name="l01391"></a>01391 |
| <a name="l01392"></a>01392 <span class="preprocessor">#else</span> |
| <a name="l01393"></a>01393 <span class="preprocessor"></span> |
| <a name="l01394"></a>01394 out1 = __SMUADX(C3, R) >> 16u; |
| <a name="l01395"></a>01395 out2 = __SMUSD(-C3, R); |
| <a name="l01396"></a>01396 |
| <a name="l01397"></a>01397 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l01398"></a>01398 |
| <a name="l01399"></a>01399 <span class="comment">/* xd' = (xa+yb-xc-yd)* co3 - (ya-xb-yc+xd)* (si3) */</span> |
| <a name="l01400"></a>01400 <span class="comment">/* yd' = (ya-xb-yc+xd)* co3 + (xa+yb-xc-yd)* (si3) */</span> |
| <a name="l01401"></a>01401 pSrc[i3] = ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| <a name="l01402"></a>01402 |
| <a name="l01403"></a>01403 |
| <a name="l01404"></a>01404 } |
| <a name="l01405"></a>01405 } |
| <a name="l01406"></a>01406 <span class="comment">/* Twiddle coefficients index modifier */</span> |
| <a name="l01407"></a>01407 twidCoefModifier <<= 2u; |
| <a name="l01408"></a>01408 } |
| <a name="l01409"></a>01409 <span class="comment">/* End of Middle stages process */</span> |
| <a name="l01410"></a>01410 |
| <a name="l01411"></a>01411 |
| <a name="l01412"></a>01412 <span class="comment">/* data is in 10.6(q6) format for the 1024 point */</span> |
| <a name="l01413"></a>01413 <span class="comment">/* data is in 8.8(q8) format for the 256 point */</span> |
| <a name="l01414"></a>01414 <span class="comment">/* data is in 6.10(q10) format for the 64 point */</span> |
| <a name="l01415"></a>01415 <span class="comment">/* data is in 4.12(q12) format for the 16 point */</span> |
| <a name="l01416"></a>01416 |
| <a name="l01417"></a>01417 <span class="comment">/* start of last stage process */</span> |
| <a name="l01418"></a>01418 |
| <a name="l01419"></a>01419 |
| <a name="l01420"></a>01420 <span class="comment">/* Initializations for the last stage */</span> |
| <a name="l01421"></a>01421 n1 = n2; |
| <a name="l01422"></a>01422 n2 >>= 2u; |
| <a name="l01423"></a>01423 |
| <a name="l01424"></a>01424 <span class="comment">/* Butterfly implementation */</span> |
| <a name="l01425"></a>01425 <span class="keywordflow">for</span> (i0 = 0u; i0 <= (fftLen - n1); i0 += n1) |
| <a name="l01426"></a>01426 { |
| <a name="l01427"></a>01427 <span class="comment">/* index calculation for the input as, */</span> |
| <a name="l01428"></a>01428 <span class="comment">/* pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */</span> |
| <a name="l01429"></a>01429 i1 = i0 + n2; |
| <a name="l01430"></a>01430 i2 = i1 + n2; |
| <a name="l01431"></a>01431 i3 = i2 + n2; |
| <a name="l01432"></a>01432 |
| <a name="l01433"></a>01433 <span class="comment">/* Reading i0, i0+fftLen/2 inputs */</span> |
| <a name="l01434"></a>01434 <span class="comment">/* Read ya (real), xa(imag) input */</span> |
| <a name="l01435"></a>01435 T = pSrc[i0]; |
| <a name="l01436"></a>01436 <span class="comment">/* Read yc (real), xc(imag) input */</span> |
| <a name="l01437"></a>01437 S = pSrc[i2]; |
| <a name="l01438"></a>01438 |
| <a name="l01439"></a>01439 <span class="comment">/* R = packed((ya + yc), (xa + xc)) */</span> |
| <a name="l01440"></a>01440 R = __QADD16(T, S); |
| <a name="l01441"></a>01441 <span class="comment">/* S = packed((ya - yc), (xa - xc)) */</span> |
| <a name="l01442"></a>01442 S = __QSUB16(T, S); |
| <a name="l01443"></a>01443 |
| <a name="l01444"></a>01444 <span class="comment">/* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */</span> |
| <a name="l01445"></a>01445 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l01446"></a>01446 T = pSrc[i1]; |
| <a name="l01447"></a>01447 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l01448"></a>01448 U = pSrc[i3]; |
| <a name="l01449"></a>01449 |
| <a name="l01450"></a>01450 <span class="comment">/* T = packed((yb + yd), (xb + xd)) */</span> |
| <a name="l01451"></a>01451 T = __QADD16(T, U); |
| <a name="l01452"></a>01452 |
| <a name="l01453"></a>01453 <span class="comment">/* writing the butterfly processed i0 sample */</span> |
| <a name="l01454"></a>01454 <span class="comment">/* xa' = xa + xb + xc + xd */</span> |
| <a name="l01455"></a>01455 <span class="comment">/* ya' = ya + yb + yc + yd */</span> |
| <a name="l01456"></a>01456 pSrc[i0] = __SHADD16(R, T); |
| <a name="l01457"></a>01457 |
| <a name="l01458"></a>01458 <span class="comment">/* R = packed((ya + yc) - (yb + yd), (xa + xc) - (xb + xd)) */</span> |
| <a name="l01459"></a>01459 R = __SHSUB16(R, T); |
| <a name="l01460"></a>01460 |
| <a name="l01461"></a>01461 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l01462"></a>01462 T = pSrc[i1]; |
| <a name="l01463"></a>01463 |
| <a name="l01464"></a>01464 <span class="comment">/* writing the butterfly processed i0 + fftLen/4 sample */</span> |
| <a name="l01465"></a>01465 <span class="comment">/* xc' = (xa-xb+xc-xd) */</span> |
| <a name="l01466"></a>01466 <span class="comment">/* yc' = (ya-yb+yc-yd) */</span> |
| <a name="l01467"></a>01467 pSrc[i1] = R; |
| <a name="l01468"></a>01468 |
| <a name="l01469"></a>01469 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l01470"></a>01470 U = pSrc[i3]; |
| <a name="l01471"></a>01471 <span class="comment">/* T = packed( (yb - yd), (xb - xd)) */</span> |
| <a name="l01472"></a>01472 T = __QSUB16(T, U); |
| <a name="l01473"></a>01473 |
| <a name="l01474"></a>01474 |
| <a name="l01475"></a>01475 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span> |
| <a name="l01476"></a>01476 <span class="preprocessor"></span> |
| <a name="l01477"></a>01477 <span class="comment">/* writing the butterfly processed i0 + fftLen/2 sample */</span> |
| <a name="l01478"></a>01478 <span class="comment">/* xb' = (xa-yb-xc+yd) */</span> |
| <a name="l01479"></a>01479 <span class="comment">/* yb' = (ya+xb-yc-xd) */</span> |
| <a name="l01480"></a>01480 pSrc[i2] = __SHASX(S, T); |
| <a name="l01481"></a>01481 |
| <a name="l01482"></a>01482 <span class="comment">/* writing the butterfly processed i0 + 3fftLen/4 sample */</span> |
| <a name="l01483"></a>01483 <span class="comment">/* xd' = (xa+yb-xc-yd) */</span> |
| <a name="l01484"></a>01484 <span class="comment">/* yd' = (ya-xb-yc+xd) */</span> |
| <a name="l01485"></a>01485 pSrc[i3] = __SHSAX(S, T); |
| <a name="l01486"></a>01486 |
| <a name="l01487"></a>01487 |
| <a name="l01488"></a>01488 <span class="preprocessor">#else</span> |
| <a name="l01489"></a>01489 <span class="preprocessor"></span> |
| <a name="l01490"></a>01490 <span class="comment">/* writing the butterfly processed i0 + fftLen/2 sample */</span> |
| <a name="l01491"></a>01491 <span class="comment">/* xb' = (xa-yb-xc+yd) */</span> |
| <a name="l01492"></a>01492 <span class="comment">/* yb' = (ya+xb-yc-xd) */</span> |
| <a name="l01493"></a>01493 pSrc[i2] = __SHSAX(S, T); |
| <a name="l01494"></a>01494 |
| <a name="l01495"></a>01495 <span class="comment">/* writing the butterfly processed i0 + 3fftLen/4 sample */</span> |
| <a name="l01496"></a>01496 <span class="comment">/* xd' = (xa+yb-xc-yd) */</span> |
| <a name="l01497"></a>01497 <span class="comment">/* yd' = (ya-xb-yc+xd) */</span> |
| <a name="l01498"></a>01498 pSrc[i3] = __SHASX(S, T); |
| <a name="l01499"></a>01499 |
| <a name="l01500"></a>01500 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span> |
| <a name="l01501"></a>01501 |
| <a name="l01502"></a>01502 } |
| <a name="l01503"></a>01503 <span class="comment">/* end of last stage process */</span> |
| <a name="l01504"></a>01504 |
| <a name="l01505"></a>01505 <span class="comment">/* output is in 11.5(q5) format for the 1024 point */</span> |
| <a name="l01506"></a>01506 <span class="comment">/* output is in 9.7(q7) format for the 256 point */</span> |
| <a name="l01507"></a>01507 <span class="comment">/* output is in 7.9(q9) format for the 64 point */</span> |
| <a name="l01508"></a>01508 <span class="comment">/* output is in 5.11(q11) format for the 16 point */</span> |
| <a name="l01509"></a>01509 |
| <a name="l01510"></a>01510 |
| <a name="l01511"></a>01511 <span class="preprocessor">#else</span> |
| <a name="l01512"></a>01512 <span class="preprocessor"></span> |
| <a name="l01513"></a>01513 <span class="comment">/* Run the below code for Cortex-M0 */</span> |
| <a name="l01514"></a>01514 |
| <a name="l01515"></a>01515 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> R0, R1, S0, S1, T0, T1, U0, U1; |
| <a name="l01516"></a>01516 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> Co1, Si1, Co2, Si2, Co3, Si3, out1, out2; |
| <a name="l01517"></a>01517 uint32_t n1, n2, ic, i0, i1, i2, i3, j, k; |
| <a name="l01518"></a>01518 |
| <a name="l01519"></a>01519 <span class="comment">/* Total process is divided into three stages */</span> |
| <a name="l01520"></a>01520 |
| <a name="l01521"></a>01521 <span class="comment">/* process first stage, middle stages, & last stage */</span> |
| <a name="l01522"></a>01522 |
| <a name="l01523"></a>01523 <span class="comment">/* Initializations for the first stage */</span> |
| <a name="l01524"></a>01524 n2 = fftLen; |
| <a name="l01525"></a>01525 n1 = n2; |
| <a name="l01526"></a>01526 |
| <a name="l01527"></a>01527 <span class="comment">/* n2 = fftLen/4 */</span> |
| <a name="l01528"></a>01528 n2 >>= 2u; |
| <a name="l01529"></a>01529 |
| <a name="l01530"></a>01530 <span class="comment">/* Index for twiddle coefficient */</span> |
| <a name="l01531"></a>01531 ic = 0u; |
| <a name="l01532"></a>01532 |
| <a name="l01533"></a>01533 <span class="comment">/* Index for input read and output write */</span> |
| <a name="l01534"></a>01534 i0 = 0u; |
| <a name="l01535"></a>01535 |
| <a name="l01536"></a>01536 j = n2; |
| <a name="l01537"></a>01537 |
| <a name="l01538"></a>01538 <span class="comment">/* Input is in 1.15(q15) format */</span> |
| <a name="l01539"></a>01539 |
| <a name="l01540"></a>01540 <span class="comment">/* Start of first stage process */</span> |
| <a name="l01541"></a>01541 <span class="keywordflow">do</span> |
| <a name="l01542"></a>01542 { |
| <a name="l01543"></a>01543 <span class="comment">/* Butterfly implementation */</span> |
| <a name="l01544"></a>01544 |
| <a name="l01545"></a>01545 <span class="comment">/* index calculation for the input as, */</span> |
| <a name="l01546"></a>01546 <span class="comment">/* pSrc16[i0 + 0], pSrc16[i0 + fftLen/4], pSrc16[i0 + fftLen/2], pSrc16[i0 + 3fftLen/4] */</span> |
| <a name="l01547"></a>01547 i1 = i0 + n2; |
| <a name="l01548"></a>01548 i2 = i1 + n2; |
| <a name="l01549"></a>01549 i3 = i2 + n2; |
| <a name="l01550"></a>01550 |
| <a name="l01551"></a>01551 <span class="comment">/* Reading i0, i0+fftLen/2 inputs */</span> |
| <a name="l01552"></a>01552 <span class="comment">/* input is down scale by 4 to avoid overflow */</span> |
| <a name="l01553"></a>01553 <span class="comment">/* Read ya (real), xa(imag) input */</span> |
| <a name="l01554"></a>01554 T0 = pSrc16[i0 * 2u] >> 2u; |
| <a name="l01555"></a>01555 T1 = pSrc16[(i0 * 2u) + 1u] >> 2u; |
| <a name="l01556"></a>01556 <span class="comment">/* input is down scale by 4 to avoid overflow */</span> |
| <a name="l01557"></a>01557 <span class="comment">/* Read yc (real), xc(imag) input */</span> |
| <a name="l01558"></a>01558 S0 = pSrc16[i2 * 2u] >> 2u; |
| <a name="l01559"></a>01559 S1 = pSrc16[(i2 * 2u) + 1u] >> 2u; |
| <a name="l01560"></a>01560 |
| <a name="l01561"></a>01561 <span class="comment">/* R0 = (ya + yc), R1 = (xa + xc) */</span> |
| <a name="l01562"></a>01562 R0 = __SSAT(T0 + S0, 16u); |
| <a name="l01563"></a>01563 R1 = __SSAT(T1 + S1, 16u); |
| <a name="l01564"></a>01564 <span class="comment">/* S0 = (ya - yc), S1 = (xa - xc) */</span> |
| <a name="l01565"></a>01565 S0 = __SSAT(T0 - S0, 16u); |
| <a name="l01566"></a>01566 S1 = __SSAT(T1 - S1, 16u); |
| <a name="l01567"></a>01567 |
| <a name="l01568"></a>01568 <span class="comment">/* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */</span> |
| <a name="l01569"></a>01569 <span class="comment">/* input is down scale by 4 to avoid overflow */</span> |
| <a name="l01570"></a>01570 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l01571"></a>01571 T0 = pSrc16[i1 * 2u] >> 2u; |
| <a name="l01572"></a>01572 T1 = pSrc16[(i1 * 2u) + 1u] >> 2u; |
| <a name="l01573"></a>01573 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l01574"></a>01574 <span class="comment">/* input is down scale by 4 to avoid overflow */</span> |
| <a name="l01575"></a>01575 U0 = pSrc16[i3 * 2u] >> 2u; |
| <a name="l01576"></a>01576 U1 = pSrc16[(i3 * 2u) + 1u] >> 2u; |
| <a name="l01577"></a>01577 |
| <a name="l01578"></a>01578 <span class="comment">/* T0 = (yb + yd), T1 = (xb + xd) */</span> |
| <a name="l01579"></a>01579 T0 = __SSAT(T0 + U0, 16u); |
| <a name="l01580"></a>01580 T1 = __SSAT(T1 + U1, 16u); |
| <a name="l01581"></a>01581 |
| <a name="l01582"></a>01582 <span class="comment">/* writing the butterfly processed i0 sample */</span> |
| <a name="l01583"></a>01583 <span class="comment">/* xa' = xa + xb + xc + xd */</span> |
| <a name="l01584"></a>01584 <span class="comment">/* ya' = ya + yb + yc + yd */</span> |
| <a name="l01585"></a>01585 pSrc16[i0 * 2u] = (R0 >> 1u) + (T0 >> 1u); |
| <a name="l01586"></a>01586 pSrc16[(i0 * 2u) + 1u] = (R1 >> 1u) + (T1 >> 1u); |
| <a name="l01587"></a>01587 |
| <a name="l01588"></a>01588 <span class="comment">/* R0 = (ya + yc) - (yb + yd), R1 = (xa + xc)- (xb + xd) */</span> |
| <a name="l01589"></a>01589 R0 = __SSAT(R0 - T0, 16u); |
| <a name="l01590"></a>01590 R1 = __SSAT(R1 - T1, 16u); |
| <a name="l01591"></a>01591 <span class="comment">/* co2 & si2 are read from Coefficient pointer */</span> |
| <a name="l01592"></a>01592 Co2 = pCoef16[2u * ic * 2u]; |
| <a name="l01593"></a>01593 Si2 = pCoef16[(2u * ic * 2u) + 1u]; |
| <a name="l01594"></a>01594 <span class="comment">/* xc' = (xa-xb+xc-xd)* co2 - (ya-yb+yc-yd)* (si2) */</span> |
| <a name="l01595"></a>01595 out1 = (short) ((Co2 * R0 - Si2 * R1) >> 16u); |
| <a name="l01596"></a>01596 <span class="comment">/* yc' = (ya-yb+yc-yd)* co2 + (xa-xb+xc-xd)* (si2) */</span> |
| <a name="l01597"></a>01597 out2 = (short) ((Si2 * R0 + Co2 * R1) >> 16u); |
| <a name="l01598"></a>01598 |
| <a name="l01599"></a>01599 <span class="comment">/* Reading i0+fftLen/4 */</span> |
| <a name="l01600"></a>01600 <span class="comment">/* input is down scale by 4 to avoid overflow */</span> |
| <a name="l01601"></a>01601 <span class="comment">/* T0 = yb, T1 = xb */</span> |
| <a name="l01602"></a>01602 T0 = pSrc16[i1 * 2u] >> 2u; |
| <a name="l01603"></a>01603 T1 = pSrc16[(i1 * 2u) + 1u] >> 2u; |
| <a name="l01604"></a>01604 |
| <a name="l01605"></a>01605 <span class="comment">/* writing the butterfly processed i0 + fftLen/4 sample */</span> |
| <a name="l01606"></a>01606 <span class="comment">/* writing output(xc', yc') in little endian format */</span> |
| <a name="l01607"></a>01607 pSrc16[i1 * 2u] = out1; |
| <a name="l01608"></a>01608 pSrc16[(i1 * 2u) + 1u] = out2; |
| <a name="l01609"></a>01609 |
| <a name="l01610"></a>01610 <span class="comment">/* Butterfly calculations */</span> |
| <a name="l01611"></a>01611 <span class="comment">/* input is down scale by 4 to avoid overflow */</span> |
| <a name="l01612"></a>01612 <span class="comment">/* U0 = yd, U1 = xd) */</span> |
| <a name="l01613"></a>01613 U0 = pSrc16[i3 * 2u] >> 2u; |
| <a name="l01614"></a>01614 U1 = pSrc16[(i3 * 2u) + 1u] >> 2u; |
| <a name="l01615"></a>01615 |
| <a name="l01616"></a>01616 <span class="comment">/* T0 = yb-yd, T1 = xb-xd) */</span> |
| <a name="l01617"></a>01617 T0 = __SSAT(T0 - U0, 16u); |
| <a name="l01618"></a>01618 T1 = __SSAT(T1 - U1, 16u); |
| <a name="l01619"></a>01619 <span class="comment">/* R0 = (ya-yc) - (xb- xd) , R1 = (xa-xc) + (yb-yd) */</span> |
| <a name="l01620"></a>01620 R0 = (short) __SSAT((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) (S0 + T1), 16); |
| <a name="l01621"></a>01621 R1 = (short) __SSAT((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) (S1 - T0), 16); |
| <a name="l01622"></a>01622 <span class="comment">/* S = (ya-yc) + (xb- xd), S1 = (xa-xc) - (yb-yd) */</span> |
| <a name="l01623"></a>01623 S0 = (short) __SSAT((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) (S0 - T1), 16); |
| <a name="l01624"></a>01624 S1 = (short) __SSAT((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) (S1 + T0), 16); |
| <a name="l01625"></a>01625 |
| <a name="l01626"></a>01626 <span class="comment">/* co1 & si1 are read from Coefficient pointer */</span> |
| <a name="l01627"></a>01627 Co1 = pCoef16[ic * 2u]; |
| <a name="l01628"></a>01628 Si1 = pCoef16[(ic * 2u) + 1u]; |
| <a name="l01629"></a>01629 <span class="comment">/* Butterfly process for the i0+fftLen/2 sample */</span> |
| <a name="l01630"></a>01630 <span class="comment">/* xb' = (xa-yb-xc+yd)* co1 - (ya+xb-yc-xd)* (si1) */</span> |
| <a name="l01631"></a>01631 out1 = (short) ((Co1 * S0 - Si1 * S1) >> 16u); |
| <a name="l01632"></a>01632 <span class="comment">/* yb' = (ya+xb-yc-xd)* co1 + (xa-yb-xc+yd)* (si1) */</span> |
| <a name="l01633"></a>01633 out2 = (short) ((Si1 * S0 + Co1 * S1) >> 16u); |
| <a name="l01634"></a>01634 <span class="comment">/* writing output(xb', yb') in little endian format */</span> |
| <a name="l01635"></a>01635 pSrc16[i2 * 2u] = out1; |
| <a name="l01636"></a>01636 pSrc16[(i2 * 2u) + 1u] = out2; |
| <a name="l01637"></a>01637 |
| <a name="l01638"></a>01638 <span class="comment">/* Co3 & si3 are read from Coefficient pointer */</span> |
| <a name="l01639"></a>01639 Co3 = pCoef16[3u * ic * 2u]; |
| <a name="l01640"></a>01640 Si3 = pCoef16[(3u * ic * 2u) + 1u]; |
| <a name="l01641"></a>01641 <span class="comment">/* Butterfly process for the i0+3fftLen/4 sample */</span> |
| <a name="l01642"></a>01642 <span class="comment">/* xd' = (xa+yb-xc-yd)* Co3 - (ya-xb-yc+xd)* (si3) */</span> |
| <a name="l01643"></a>01643 out1 = (short) ((Co3 * R0 - Si3 * R1) >> 16u); |
| <a name="l01644"></a>01644 <span class="comment">/* yd' = (ya-xb-yc+xd)* Co3 + (xa+yb-xc-yd)* (si3) */</span> |
| <a name="l01645"></a>01645 out2 = (short) ((Si3 * R0 + Co3 * R1) >> 16u); |
| <a name="l01646"></a>01646 <span class="comment">/* writing output(xd', yd') in little endian format */</span> |
| <a name="l01647"></a>01647 pSrc16[i3 * 2u] = out1; |
| <a name="l01648"></a>01648 pSrc16[(i3 * 2u) + 1u] = out2; |
| <a name="l01649"></a>01649 |
| <a name="l01650"></a>01650 <span class="comment">/* Twiddle coefficients index modifier */</span> |
| <a name="l01651"></a>01651 ic = ic + twidCoefModifier; |
| <a name="l01652"></a>01652 |
| <a name="l01653"></a>01653 <span class="comment">/* Updating input index */</span> |
| <a name="l01654"></a>01654 i0 = i0 + 1u; |
| <a name="l01655"></a>01655 |
| <a name="l01656"></a>01656 } <span class="keywordflow">while</span>(--j); |
| <a name="l01657"></a>01657 |
| <a name="l01658"></a>01658 <span class="comment">/* End of first stage process */</span> |
| <a name="l01659"></a>01659 |
| <a name="l01660"></a>01660 <span class="comment">/* data is in 4.11(q11) format */</span> |
| <a name="l01661"></a>01661 |
| <a name="l01662"></a>01662 |
| <a name="l01663"></a>01663 <span class="comment">/* Start of Middle stage process */</span> |
| <a name="l01664"></a>01664 |
| <a name="l01665"></a>01665 <span class="comment">/* Twiddle coefficients index modifier */</span> |
| <a name="l01666"></a>01666 twidCoefModifier <<= 2u; |
| <a name="l01667"></a>01667 |
| <a name="l01668"></a>01668 <span class="comment">/* Calculation of Middle stage */</span> |
| <a name="l01669"></a>01669 <span class="keywordflow">for</span> (k = fftLen / 4u; k > 4u; k >>= 2u) |
| <a name="l01670"></a>01670 { |
| <a name="l01671"></a>01671 <span class="comment">/* Initializations for the middle stage */</span> |
| <a name="l01672"></a>01672 n1 = n2; |
| <a name="l01673"></a>01673 n2 >>= 2u; |
| <a name="l01674"></a>01674 ic = 0u; |
| <a name="l01675"></a>01675 |
| <a name="l01676"></a>01676 <span class="keywordflow">for</span> (j = 0u; j <= (n2 - 1u); j++) |
| <a name="l01677"></a>01677 { |
| <a name="l01678"></a>01678 <span class="comment">/* index calculation for the coefficients */</span> |
| <a name="l01679"></a>01679 Co1 = pCoef16[ic * 2u]; |
| <a name="l01680"></a>01680 Si1 = pCoef16[(ic * 2u) + 1u]; |
| <a name="l01681"></a>01681 Co2 = pCoef16[2u * ic * 2u]; |
| <a name="l01682"></a>01682 Si2 = pCoef16[2u * ic * 2u + 1u]; |
| <a name="l01683"></a>01683 Co3 = pCoef16[3u * ic * 2u]; |
| <a name="l01684"></a>01684 Si3 = pCoef16[(3u * ic * 2u) + 1u]; |
| <a name="l01685"></a>01685 |
| <a name="l01686"></a>01686 <span class="comment">/* Twiddle coefficients index modifier */</span> |
| <a name="l01687"></a>01687 ic = ic + twidCoefModifier; |
| <a name="l01688"></a>01688 |
| <a name="l01689"></a>01689 <span class="comment">/* Butterfly implementation */</span> |
| <a name="l01690"></a>01690 <span class="keywordflow">for</span> (i0 = j; i0 < fftLen; i0 += n1) |
| <a name="l01691"></a>01691 { |
| <a name="l01692"></a>01692 <span class="comment">/* index calculation for the input as, */</span> |
| <a name="l01693"></a>01693 <span class="comment">/* pSrc16[i0 + 0], pSrc16[i0 + fftLen/4], pSrc16[i0 + fftLen/2], pSrc16[i0 + 3fftLen/4] */</span> |
| <a name="l01694"></a>01694 i1 = i0 + n2; |
| <a name="l01695"></a>01695 i2 = i1 + n2; |
| <a name="l01696"></a>01696 i3 = i2 + n2; |
| <a name="l01697"></a>01697 |
| <a name="l01698"></a>01698 <span class="comment">/* Reading i0, i0+fftLen/2 inputs */</span> |
| <a name="l01699"></a>01699 <span class="comment">/* Read ya (real), xa(imag) input */</span> |
| <a name="l01700"></a>01700 T0 = pSrc16[i0 * 2u]; |
| <a name="l01701"></a>01701 T1 = pSrc16[(i0 * 2u) + 1u]; |
| <a name="l01702"></a>01702 |
| <a name="l01703"></a>01703 <span class="comment">/* Read yc (real), xc(imag) input */</span> |
| <a name="l01704"></a>01704 S0 = pSrc16[i2 * 2u]; |
| <a name="l01705"></a>01705 S1 = pSrc16[(i2 * 2u) + 1u]; |
| <a name="l01706"></a>01706 |
| <a name="l01707"></a>01707 |
| <a name="l01708"></a>01708 <span class="comment">/* R0 = (ya + yc), R1 = (xa + xc) */</span> |
| <a name="l01709"></a>01709 R0 = __SSAT(T0 + S0, 16u); |
| <a name="l01710"></a>01710 R1 = __SSAT(T1 + S1, 16u); |
| <a name="l01711"></a>01711 <span class="comment">/* S0 = (ya - yc), S1 = (xa - xc) */</span> |
| <a name="l01712"></a>01712 S0 = __SSAT(T0 - S0, 16u); |
| <a name="l01713"></a>01713 S1 = __SSAT(T1 - S1, 16u); |
| <a name="l01714"></a>01714 |
| <a name="l01715"></a>01715 <span class="comment">/* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */</span> |
| <a name="l01716"></a>01716 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l01717"></a>01717 T0 = pSrc16[i1 * 2u]; |
| <a name="l01718"></a>01718 T1 = pSrc16[(i1 * 2u) + 1u]; |
| <a name="l01719"></a>01719 |
| <a name="l01720"></a>01720 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l01721"></a>01721 U0 = pSrc16[i3 * 2u]; |
| <a name="l01722"></a>01722 U1 = pSrc16[(i3 * 2u) + 1u]; |
| <a name="l01723"></a>01723 |
| <a name="l01724"></a>01724 <span class="comment">/* T0 = (yb + yd), T1 = (xb + xd) */</span> |
| <a name="l01725"></a>01725 T0 = __SSAT(T0 + U0, 16u); |
| <a name="l01726"></a>01726 T1 = __SSAT(T1 + U1, 16u); |
| <a name="l01727"></a>01727 |
| <a name="l01728"></a>01728 <span class="comment">/* writing the butterfly processed i0 sample */</span> |
| <a name="l01729"></a>01729 <span class="comment">/* xa' = xa + xb + xc + xd */</span> |
| <a name="l01730"></a>01730 <span class="comment">/* ya' = ya + yb + yc + yd */</span> |
| <a name="l01731"></a>01731 pSrc16[i0 * 2u] = ((R0 >> 1u) + (T0 >> 1u)) >> 1u; |
| <a name="l01732"></a>01732 pSrc16[(i0 * 2u) + 1u] = ((R1 >> 1u) + (T1 >> 1u)) >> 1u; |
| <a name="l01733"></a>01733 |
| <a name="l01734"></a>01734 <span class="comment">/* R0 = (ya + yc) - (yb + yd), R1 = (xa + xc) - (xb + xd) */</span> |
| <a name="l01735"></a>01735 R0 = (R0 >> 1u) - (T0 >> 1u); |
| <a name="l01736"></a>01736 R1 = (R1 >> 1u) - (T1 >> 1u); |
| <a name="l01737"></a>01737 |
| <a name="l01738"></a>01738 <span class="comment">/* (ya-yb+yc-yd)* (si2) - (xa-xb+xc-xd)* co2 */</span> |
| <a name="l01739"></a>01739 out1 = (short) ((Co2 * R0 - Si2 * R1) >> 16); |
| <a name="l01740"></a>01740 <span class="comment">/* (ya-yb+yc-yd)* co2 + (xa-xb+xc-xd)* (si2) */</span> |
| <a name="l01741"></a>01741 out2 = (short) ((Si2 * R0 + Co2 * R1) >> 16); |
| <a name="l01742"></a>01742 |
| <a name="l01743"></a>01743 <span class="comment">/* Reading i0+3fftLen/4 */</span> |
| <a name="l01744"></a>01744 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l01745"></a>01745 T0 = pSrc16[i1 * 2u]; |
| <a name="l01746"></a>01746 T1 = pSrc16[(i1 * 2u) + 1u]; |
| <a name="l01747"></a>01747 |
| <a name="l01748"></a>01748 <span class="comment">/* writing the butterfly processed i0 + fftLen/4 sample */</span> |
| <a name="l01749"></a>01749 <span class="comment">/* xc' = (xa-xb+xc-xd)* co2 - (ya-yb+yc-yd)* (si2) */</span> |
| <a name="l01750"></a>01750 <span class="comment">/* yc' = (ya-yb+yc-yd)* co2 + (xa-xb+xc-xd)* (si2) */</span> |
| <a name="l01751"></a>01751 pSrc16[i1 * 2u] = out1; |
| <a name="l01752"></a>01752 pSrc16[(i1 * 2u) + 1u] = out2; |
| <a name="l01753"></a>01753 |
| <a name="l01754"></a>01754 <span class="comment">/* Butterfly calculations */</span> |
| <a name="l01755"></a>01755 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l01756"></a>01756 U0 = pSrc16[i3 * 2u]; |
| <a name="l01757"></a>01757 U1 = pSrc16[(i3 * 2u) + 1u]; |
| <a name="l01758"></a>01758 |
| <a name="l01759"></a>01759 <span class="comment">/* T0 = yb-yd, T1 = xb-xd) */</span> |
| <a name="l01760"></a>01760 T0 = __SSAT(T0 - U0, 16u); |
| <a name="l01761"></a>01761 T1 = __SSAT(T1 - U1, 16u); |
| <a name="l01762"></a>01762 |
| <a name="l01763"></a>01763 <span class="comment">/* R0 = (ya-yc) - (xb- xd) , R1 = (xa-xc) + (yb-yd) */</span> |
| <a name="l01764"></a>01764 R0 = (S0 >> 1u) + (T1 >> 1u); |
| <a name="l01765"></a>01765 R1 = (S1 >> 1u) - (T0 >> 1u); |
| <a name="l01766"></a>01766 |
| <a name="l01767"></a>01767 <span class="comment">/* S1 = (ya-yc) + (xb- xd), S1 = (xa-xc) - (yb-yd) */</span> |
| <a name="l01768"></a>01768 S0 = (S0 >> 1u) - (T1 >> 1u); |
| <a name="l01769"></a>01769 S1 = (S1 >> 1u) + (T0 >> 1u); |
| <a name="l01770"></a>01770 |
| <a name="l01771"></a>01771 <span class="comment">/* Butterfly process for the i0+fftLen/2 sample */</span> |
| <a name="l01772"></a>01772 out1 = (short) ((Co1 * S0 - Si1 * S1) >> 16u); |
| <a name="l01773"></a>01773 out2 = (short) ((Si1 * S0 + Co1 * S1) >> 16u); |
| <a name="l01774"></a>01774 <span class="comment">/* xb' = (xa-yb-xc+yd)* co1 - (ya+xb-yc-xd)* (si1) */</span> |
| <a name="l01775"></a>01775 <span class="comment">/* yb' = (ya+xb-yc-xd)* co1 + (xa-yb-xc+yd)* (si1) */</span> |
| <a name="l01776"></a>01776 pSrc16[i2 * 2u] = out1; |
| <a name="l01777"></a>01777 pSrc16[(i2 * 2u) + 1u] = out2; |
| <a name="l01778"></a>01778 |
| <a name="l01779"></a>01779 <span class="comment">/* Butterfly process for the i0+3fftLen/4 sample */</span> |
| <a name="l01780"></a>01780 out1 = (short) ((Co3 * R0 - Si3 * R1) >> 16u); |
| <a name="l01781"></a>01781 |
| <a name="l01782"></a>01782 out2 = (short) ((Si3 * R0 + Co3 * R1) >> 16u); |
| <a name="l01783"></a>01783 <span class="comment">/* xd' = (xa+yb-xc-yd)* Co3 - (ya-xb-yc+xd)* (si3) */</span> |
| <a name="l01784"></a>01784 <span class="comment">/* yd' = (ya-xb-yc+xd)* Co3 + (xa+yb-xc-yd)* (si3) */</span> |
| <a name="l01785"></a>01785 pSrc16[i3 * 2u] = out1; |
| <a name="l01786"></a>01786 pSrc16[(i3 * 2u) + 1u] = out2; |
| <a name="l01787"></a>01787 |
| <a name="l01788"></a>01788 |
| <a name="l01789"></a>01789 } |
| <a name="l01790"></a>01790 } |
| <a name="l01791"></a>01791 <span class="comment">/* Twiddle coefficients index modifier */</span> |
| <a name="l01792"></a>01792 twidCoefModifier <<= 2u; |
| <a name="l01793"></a>01793 } |
| <a name="l01794"></a>01794 <span class="comment">/* End of Middle stages process */</span> |
| <a name="l01795"></a>01795 |
| <a name="l01796"></a>01796 |
| <a name="l01797"></a>01797 <span class="comment">/* data is in 10.6(q6) format for the 1024 point */</span> |
| <a name="l01798"></a>01798 <span class="comment">/* data is in 8.8(q8) format for the 256 point */</span> |
| <a name="l01799"></a>01799 <span class="comment">/* data is in 6.10(q10) format for the 64 point */</span> |
| <a name="l01800"></a>01800 <span class="comment">/* data is in 4.12(q12) format for the 16 point */</span> |
| <a name="l01801"></a>01801 |
| <a name="l01802"></a>01802 <span class="comment">/* start of last stage process */</span> |
| <a name="l01803"></a>01803 |
| <a name="l01804"></a>01804 |
| <a name="l01805"></a>01805 <span class="comment">/* Initializations for the last stage */</span> |
| <a name="l01806"></a>01806 n1 = n2; |
| <a name="l01807"></a>01807 n2 >>= 2u; |
| <a name="l01808"></a>01808 |
| <a name="l01809"></a>01809 <span class="comment">/* Butterfly implementation */</span> |
| <a name="l01810"></a>01810 <span class="keywordflow">for</span> (i0 = 0u; i0 <= (fftLen - n1); i0 += n1) |
| <a name="l01811"></a>01811 { |
| <a name="l01812"></a>01812 <span class="comment">/* index calculation for the input as, */</span> |
| <a name="l01813"></a>01813 <span class="comment">/* pSrc16[i0 + 0], pSrc16[i0 + fftLen/4], pSrc16[i0 + fftLen/2], pSrc16[i0 + 3fftLen/4] */</span> |
| <a name="l01814"></a>01814 i1 = i0 + n2; |
| <a name="l01815"></a>01815 i2 = i1 + n2; |
| <a name="l01816"></a>01816 i3 = i2 + n2; |
| <a name="l01817"></a>01817 |
| <a name="l01818"></a>01818 <span class="comment">/* Reading i0, i0+fftLen/2 inputs */</span> |
| <a name="l01819"></a>01819 <span class="comment">/* Read ya (real), xa(imag) input */</span> |
| <a name="l01820"></a>01820 T0 = pSrc16[i0 * 2u]; |
| <a name="l01821"></a>01821 T1 = pSrc16[(i0 * 2u) + 1u]; |
| <a name="l01822"></a>01822 <span class="comment">/* Read yc (real), xc(imag) input */</span> |
| <a name="l01823"></a>01823 S0 = pSrc16[i2 * 2u]; |
| <a name="l01824"></a>01824 S1 = pSrc16[(i2 * 2u) + 1u]; |
| <a name="l01825"></a>01825 |
| <a name="l01826"></a>01826 <span class="comment">/* R0 = (ya + yc), R1 = (xa + xc) */</span> |
| <a name="l01827"></a>01827 R0 = __SSAT(T0 + S0, 16u); |
| <a name="l01828"></a>01828 R1 = __SSAT(T1 + S1, 16u); |
| <a name="l01829"></a>01829 <span class="comment">/* S0 = (ya - yc), S1 = (xa - xc) */</span> |
| <a name="l01830"></a>01830 S0 = __SSAT(T0 - S0, 16u); |
| <a name="l01831"></a>01831 S1 = __SSAT(T1 - S1, 16u); |
| <a name="l01832"></a>01832 |
| <a name="l01833"></a>01833 <span class="comment">/* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */</span> |
| <a name="l01834"></a>01834 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l01835"></a>01835 T0 = pSrc16[i1 * 2u]; |
| <a name="l01836"></a>01836 T1 = pSrc16[(i1 * 2u) + 1u]; |
| <a name="l01837"></a>01837 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l01838"></a>01838 U0 = pSrc16[i3 * 2u]; |
| <a name="l01839"></a>01839 U1 = pSrc16[(i3 * 2u) + 1u]; |
| <a name="l01840"></a>01840 |
| <a name="l01841"></a>01841 <span class="comment">/* T0 = (yb + yd), T1 = (xb + xd) */</span> |
| <a name="l01842"></a>01842 T0 = __SSAT(T0 + U0, 16u); |
| <a name="l01843"></a>01843 T1 = __SSAT(T1 + U1, 16u); |
| <a name="l01844"></a>01844 |
| <a name="l01845"></a>01845 <span class="comment">/* writing the butterfly processed i0 sample */</span> |
| <a name="l01846"></a>01846 <span class="comment">/* xa' = xa + xb + xc + xd */</span> |
| <a name="l01847"></a>01847 <span class="comment">/* ya' = ya + yb + yc + yd */</span> |
| <a name="l01848"></a>01848 pSrc16[i0 * 2u] = (R0 >> 1u) + (T0 >> 1u); |
| <a name="l01849"></a>01849 pSrc16[(i0 * 2u) + 1u] = (R1 >> 1u) + (T1 >> 1u); |
| <a name="l01850"></a>01850 |
| <a name="l01851"></a>01851 <span class="comment">/* R0 = (ya + yc) - (yb + yd), R1 = (xa + xc) - (xb + xd) */</span> |
| <a name="l01852"></a>01852 R0 = (R0 >> 1u) - (T0 >> 1u); |
| <a name="l01853"></a>01853 R1 = (R1 >> 1u) - (T1 >> 1u); |
| <a name="l01854"></a>01854 |
| <a name="l01855"></a>01855 <span class="comment">/* Read yb (real), xb(imag) input */</span> |
| <a name="l01856"></a>01856 T0 = pSrc16[i1 * 2u]; |
| <a name="l01857"></a>01857 T1 = pSrc16[(i1 * 2u) + 1u]; |
| <a name="l01858"></a>01858 |
| <a name="l01859"></a>01859 <span class="comment">/* writing the butterfly processed i0 + fftLen/4 sample */</span> |
| <a name="l01860"></a>01860 <span class="comment">/* xc' = (xa-xb+xc-xd) */</span> |
| <a name="l01861"></a>01861 <span class="comment">/* yc' = (ya-yb+yc-yd) */</span> |
| <a name="l01862"></a>01862 pSrc16[i1 * 2u] = R0; |
| <a name="l01863"></a>01863 pSrc16[(i1 * 2u) + 1u] = R1; |
| <a name="l01864"></a>01864 |
| <a name="l01865"></a>01865 <span class="comment">/* Read yd (real), xd(imag) input */</span> |
| <a name="l01866"></a>01866 U0 = pSrc16[i3 * 2u]; |
| <a name="l01867"></a>01867 U1 = pSrc16[(i3 * 2u) + 1u]; |
| <a name="l01868"></a>01868 <span class="comment">/* T0 = (yb - yd), T1 = (xb - xd) */</span> |
| <a name="l01869"></a>01869 T0 = __SSAT(T0 - U0, 16u); |
| <a name="l01870"></a>01870 T1 = __SSAT(T1 - U1, 16u); |
| <a name="l01871"></a>01871 |
| <a name="l01872"></a>01872 <span class="comment">/* writing the butterfly processed i0 + fftLen/2 sample */</span> |
| <a name="l01873"></a>01873 <span class="comment">/* xb' = (xa-yb-xc+yd) */</span> |
| <a name="l01874"></a>01874 <span class="comment">/* yb' = (ya+xb-yc-xd) */</span> |
| <a name="l01875"></a>01875 pSrc16[i2 * 2u] = (S0 >> 1u) - (T1 >> 1u); |
| <a name="l01876"></a>01876 pSrc16[(i2 * 2u) + 1u] = (S1 >> 1u) + (T0 >> 1u); |
| <a name="l01877"></a>01877 |
| <a name="l01878"></a>01878 |
| <a name="l01879"></a>01879 <span class="comment">/* writing the butterfly processed i0 + 3fftLen/4 sample */</span> |
| <a name="l01880"></a>01880 <span class="comment">/* xd' = (xa+yb-xc-yd) */</span> |
| <a name="l01881"></a>01881 <span class="comment">/* yd' = (ya-xb-yc+xd) */</span> |
| <a name="l01882"></a>01882 pSrc16[i3 * 2u] = (S0 >> 1u) + (T1 >> 1u); |
| <a name="l01883"></a>01883 pSrc16[(i3 * 2u) + 1u] = (S1 >> 1u) - (T0 >> 1u); |
| <a name="l01884"></a>01884 } |
| <a name="l01885"></a>01885 <span class="comment">/* end of last stage process */</span> |
| <a name="l01886"></a>01886 |
| <a name="l01887"></a>01887 <span class="comment">/* output is in 11.5(q5) format for the 1024 point */</span> |
| <a name="l01888"></a>01888 <span class="comment">/* output is in 9.7(q7) format for the 256 point */</span> |
| <a name="l01889"></a>01889 <span class="comment">/* output is in 7.9(q9) format for the 64 point */</span> |
| <a name="l01890"></a>01890 <span class="comment">/* output is in 5.11(q11) format for the 16 point */</span> |
| <a name="l01891"></a>01891 |
| <a name="l01892"></a>01892 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_CM0 */</span> |
| <a name="l01893"></a>01893 |
| <a name="l01894"></a>01894 } |
| <a name="l01895"></a>01895 |
| <a name="l01896"></a>01896 |
| <a name="l01897"></a>01897 <span class="comment">/* </span> |
| <a name="l01898"></a>01898 <span class="comment"> * @brief In-place bit reversal function. </span> |
| <a name="l01899"></a>01899 <span class="comment"> * @param[in, out] *pSrc points to the in-place buffer of Q15 data type. </span> |
| <a name="l01900"></a>01900 <span class="comment"> * @param[in] fftLen length of the FFT. </span> |
| <a name="l01901"></a>01901 <span class="comment"> * @param[in] bitRevFactor bit reversal modifier that supports different size FFTs with the same bit reversal table </span> |
| <a name="l01902"></a>01902 <span class="comment"> * @param[in] *pBitRevTab points to bit reversal table. </span> |
| <a name="l01903"></a>01903 <span class="comment"> * @return none. </span> |
| <a name="l01904"></a>01904 <span class="comment"> */</span> |
| <a name="l01905"></a>01905 |
| <a name="l01906"></a><a class="code" href="arm__math_8h.html#a73f48eaea9297605705ae25d3405343e">01906</a> <span class="keywordtype">void</span> <a class="code" href="arm__cfft__radix4__q15_8c.html#a12a07b49948c354172ae07358309a4a5" title="In-place bit reversal function.">arm_bitreversal_q15</a>( |
| <a name="l01907"></a>01907 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> * pSrc16, |
| <a name="l01908"></a>01908 uint32_t fftLen, |
| <a name="l01909"></a>01909 uint16_t bitRevFactor, |
| <a name="l01910"></a>01910 uint16_t * pBitRevTab) |
| <a name="l01911"></a>01911 { |
| <a name="l01912"></a>01912 <a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *pSrc = (<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) pSrc16; |
| <a name="l01913"></a>01913 <a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> in; |
| <a name="l01914"></a>01914 uint32_t fftLenBy2, fftLenBy2p1; |
| <a name="l01915"></a>01915 uint32_t i, j; |
| <a name="l01916"></a>01916 |
| <a name="l01917"></a>01917 <span class="comment">/* Initializations */</span> |
| <a name="l01918"></a>01918 j = 0u; |
| <a name="l01919"></a>01919 fftLenBy2 = fftLen / 2u; |
| <a name="l01920"></a>01920 fftLenBy2p1 = (fftLen / 2u) + 1u; |
| <a name="l01921"></a>01921 |
| <a name="l01922"></a>01922 <span class="comment">/* Bit Reversal Implementation */</span> |
| <a name="l01923"></a>01923 <span class="keywordflow">for</span> (i = 0u; i <= (fftLenBy2 - 2u); i += 2u) |
| <a name="l01924"></a>01924 { |
| <a name="l01925"></a>01925 <span class="keywordflow">if</span>(i < j) |
| <a name="l01926"></a>01926 { |
| <a name="l01927"></a>01927 <span class="comment">/* pSrc[i] <-> pSrc[j]; */</span> |
| <a name="l01928"></a>01928 <span class="comment">/* pSrc[i+1u] <-> pSrc[j+1u] */</span> |
| <a name="l01929"></a>01929 in = pSrc[i]; |
| <a name="l01930"></a>01930 pSrc[i] = pSrc[j]; |
| <a name="l01931"></a>01931 pSrc[j] = in; |
| <a name="l01932"></a>01932 |
| <a name="l01933"></a>01933 <span class="comment">/* pSrc[i + fftLenBy2p1] <-> pSrc[j + fftLenBy2p1]; */</span> |
| <a name="l01934"></a>01934 <span class="comment">/* pSrc[i + fftLenBy2p1+1u] <-> pSrc[j + fftLenBy2p1+1u] */</span> |
| <a name="l01935"></a>01935 in = pSrc[i + fftLenBy2p1]; |
| <a name="l01936"></a>01936 pSrc[i + fftLenBy2p1] = pSrc[j + fftLenBy2p1]; |
| <a name="l01937"></a>01937 pSrc[j + fftLenBy2p1] = in; |
| <a name="l01938"></a>01938 } |
| <a name="l01939"></a>01939 |
| <a name="l01940"></a>01940 <span class="comment">/* pSrc[i+1u] <-> pSrc[j+fftLenBy2]; */</span> |
| <a name="l01941"></a>01941 <span class="comment">/* pSrc[i+2] <-> pSrc[j+fftLenBy2+1u] */</span> |
| <a name="l01942"></a>01942 in = pSrc[i + 1u]; |
| <a name="l01943"></a>01943 pSrc[i + 1u] = pSrc[j + fftLenBy2]; |
| <a name="l01944"></a>01944 pSrc[j + fftLenBy2] = in; |
| <a name="l01945"></a>01945 |
| <a name="l01946"></a>01946 <span class="comment">/* Reading the index for the bit reversal */</span> |
| <a name="l01947"></a>01947 j = *pBitRevTab; |
| <a name="l01948"></a>01948 |
| <a name="l01949"></a>01949 <span class="comment">/* Updating the bit reversal index depending on the fft length */</span> |
| <a name="l01950"></a>01950 pBitRevTab += bitRevFactor; |
| <a name="l01951"></a>01951 } |
| <a name="l01952"></a>01952 } |
| </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 13:16:15 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> |
