1 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
2 <html xmlns="http://www.w3.org/1999/xhtml">
4 <meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/>
5 <title>CMSIS DSP Software Library: arm_conv_q15.c Source File</title>
6 <link href="tabs.css" rel="stylesheet" type="text/css"/>
7 <link href="search/search.css" rel="stylesheet" type="text/css"/>
8 <script type="text/javaScript" src="search/search.js"></script>
9 <link href="doxygen.css" rel="stylesheet" type="text/css"/>
11 <body onload='searchBox.OnSelectItem(0);'>
12 <!-- Generated by Doxygen 1.7.2 -->
13 <script type="text/javascript"><!--
14 var searchBox = new SearchBox("searchBox", "search",false,'Search');
16 <div class="navigation" id="top">
19 <li><a href="index.html"><span>Main Page</span></a></li>
20 <li><a href="modules.html"><span>Modules</span></a></li>
21 <li><a href="annotated.html"><span>Data Structures</span></a></li>
22 <li class="current"><a href="files.html"><span>Files</span></a></li>
23 <li><a href="examples.html"><span>Examples</span></a></li>
25 <div id="MSearchBox" class="MSearchBoxInactive">
27 <img id="MSearchSelect" src="search/mag_sel.png"
28 onmouseover="return searchBox.OnSearchSelectShow()"
29 onmouseout="return searchBox.OnSearchSelectHide()"
31 <input type="text" id="MSearchField" value="Search" accesskey="S"
32 onfocus="searchBox.OnSearchFieldFocus(true)"
33 onblur="searchBox.OnSearchFieldFocus(false)"
34 onkeyup="searchBox.OnSearchFieldChange(event)"/>
35 </span><span class="right">
36 <a id="MSearchClose" href="javascript:searchBox.CloseResultsWindow()"><img id="MSearchCloseImg" border="0" src="search/close.png" alt=""/></a>
44 <li><a href="files.html"><span>File List</span></a></li>
45 <li><a href="globals.html"><span>Globals</span></a></li>
49 <div class="headertitle">
50 <h1>arm_conv_q15.c</h1> </div>
52 <div class="contents">
53 <a href="arm__conv__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>
54 <a name="l00002"></a>00002 <span class="comment">* Copyright (C) 2010 ARM Limited. All rights reserved. </span>
55 <a name="l00003"></a>00003 <span class="comment">* </span>
56 <a name="l00004"></a>00004 <span class="comment">* $Date: 15. July 2011 </span>
57 <a name="l00005"></a>00005 <span class="comment">* $Revision: V1.0.10 </span>
58 <a name="l00006"></a>00006 <span class="comment">* </span>
59 <a name="l00007"></a>00007 <span class="comment">* Project: CMSIS DSP Library </span>
60 <a name="l00008"></a>00008 <span class="comment">* Title: arm_conv_q15.c </span>
61 <a name="l00009"></a>00009 <span class="comment">* </span>
62 <a name="l00010"></a>00010 <span class="comment">* Description: Convolution of Q15 sequences. </span>
63 <a name="l00011"></a>00011 <span class="comment">* </span>
64 <a name="l00012"></a>00012 <span class="comment">* Target Processor: Cortex-M4/Cortex-M3/Cortex-M0</span>
65 <a name="l00013"></a>00013 <span class="comment">* </span>
66 <a name="l00014"></a>00014 <span class="comment">* Version 1.0.10 2011/7/15 </span>
67 <a name="l00015"></a>00015 <span class="comment">* Big Endian support added and Merged M0 and M3/M4 Source code. </span>
68 <a name="l00016"></a>00016 <span class="comment">* </span>
69 <a name="l00017"></a>00017 <span class="comment">* Version 1.0.3 2010/11/29 </span>
70 <a name="l00018"></a>00018 <span class="comment">* Re-organized the CMSIS folders and updated documentation. </span>
71 <a name="l00019"></a>00019 <span class="comment">* </span>
72 <a name="l00020"></a>00020 <span class="comment">* Version 1.0.2 2010/11/11 </span>
73 <a name="l00021"></a>00021 <span class="comment">* Documentation updated. </span>
74 <a name="l00022"></a>00022 <span class="comment">* </span>
75 <a name="l00023"></a>00023 <span class="comment">* Version 1.0.1 2010/10/05 </span>
76 <a name="l00024"></a>00024 <span class="comment">* Production release and review comments incorporated. </span>
77 <a name="l00025"></a>00025 <span class="comment">* </span>
78 <a name="l00026"></a>00026 <span class="comment">* Version 1.0.0 2010/09/20 </span>
79 <a name="l00027"></a>00027 <span class="comment">* Production release and review comments incorporated </span>
80 <a name="l00028"></a>00028 <span class="comment">* </span>
81 <a name="l00029"></a>00029 <span class="comment">* Version 0.0.7 2010/06/10 </span>
82 <a name="l00030"></a>00030 <span class="comment">* Misra-C changes done </span>
83 <a name="l00031"></a>00031 <span class="comment">* </span>
84 <a name="l00032"></a>00032 <span class="comment">* -------------------------------------------------------------------- */</span>
85 <a name="l00033"></a>00033
86 <a name="l00034"></a>00034 <span class="preprocessor">#include "<a class="code" href="arm__math_8h.html">arm_math.h</a>"</span>
87 <a name="l00035"></a>00035
88 <a name="l00068"></a><a class="code" href="group___conv.html#gaccd6a89b0ff7a94df64610598e6e6893">00068</a> <span class="keywordtype">void</span> <a class="code" href="group___conv.html#gaccd6a89b0ff7a94df64610598e6e6893" title="Convolution of Q15 sequences.">arm_conv_q15</a>(
89 <a name="l00069"></a>00069 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> * pSrcA,
90 <a name="l00070"></a>00070 uint32_t <a class="code" href="arm__convolution__example__f32_8c.html#ace48ed566e2cd6a680f0681192e6af28">srcALen</a>,
91 <a name="l00071"></a>00071 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> * pSrcB,
92 <a name="l00072"></a>00072 uint32_t <a class="code" href="arm__convolution__example__f32_8c.html#aea71286f498978c5ed3775609b974fc8">srcBLen</a>,
93 <a name="l00073"></a>00073 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> * pDst)
94 <a name="l00074"></a>00074 {
95 <a name="l00075"></a>00075
96 <a name="l00076"></a>00076 <span class="preprocessor">#ifndef ARM_MATH_CM0</span>
97 <a name="l00077"></a>00077 <span class="preprocessor"></span>
98 <a name="l00078"></a>00078 <span class="comment">/* Run the below code for Cortex-M4 and Cortex-M3 */</span>
99 <a name="l00079"></a>00079
100 <a name="l00080"></a>00080 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *pIn1; <span class="comment">/* inputA pointer */</span>
101 <a name="l00081"></a>00081 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *pIn2; <span class="comment">/* inputB pointer */</span>
102 <a name="l00082"></a>00082 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *pOut = pDst; <span class="comment">/* output pointer */</span>
103 <a name="l00083"></a>00083 <a class="code" href="arm__math_8h.html#a5aea1cb12fc02d9d44c8abf217eaa5c6" title="64-bit fractional data type in 1.63 format.">q63_t</a> sum, acc0, acc1, acc2, acc3; <span class="comment">/* Accumulator */</span>
104 <a name="l00084"></a>00084 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *px; <span class="comment">/* Intermediate inputA pointer */</span>
105 <a name="l00085"></a>00085 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *py; <span class="comment">/* Intermediate inputB pointer */</span>
106 <a name="l00086"></a>00086 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *pSrc1, *pSrc2; <span class="comment">/* Intermediate pointers */</span>
107 <a name="l00087"></a>00087 <a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> x0, x1, x2, x3, c0; <span class="comment">/* Temporary variables to hold state and coefficient values */</span>
108 <a name="l00088"></a>00088 uint32_t blockSize1, blockSize2, blockSize3, j, k, count, blkCnt; <span class="comment">/* loop counter */</span>
109 <a name="l00089"></a>00089 <a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *pb; <span class="comment">/* 32 bit pointer for inputB buffer */</span>
110 <a name="l00090"></a>00090
111 <a name="l00091"></a>00091
112 <a name="l00092"></a>00092 <span class="comment">/* The algorithm implementation is based on the lengths of the inputs. */</span>
113 <a name="l00093"></a>00093 <span class="comment">/* srcB is always made to slide across srcA. */</span>
114 <a name="l00094"></a>00094 <span class="comment">/* So srcBLen is always considered as shorter or equal to srcALen */</span>
115 <a name="l00095"></a>00095 <span class="keywordflow">if</span>(srcALen >= srcBLen)
116 <a name="l00096"></a>00096 {
117 <a name="l00097"></a>00097 <span class="comment">/* Initialization of inputA pointer */</span>
118 <a name="l00098"></a>00098 pIn1 = pSrcA;
119 <a name="l00099"></a>00099
120 <a name="l00100"></a>00100 <span class="comment">/* Initialization of inputB pointer */</span>
121 <a name="l00101"></a>00101 pIn2 = pSrcB;
122 <a name="l00102"></a>00102 }
123 <a name="l00103"></a>00103 <span class="keywordflow">else</span>
124 <a name="l00104"></a>00104 {
125 <a name="l00105"></a>00105 <span class="comment">/* Initialization of inputA pointer */</span>
126 <a name="l00106"></a>00106 pIn1 = pSrcB;
127 <a name="l00107"></a>00107
128 <a name="l00108"></a>00108 <span class="comment">/* Initialization of inputB pointer */</span>
129 <a name="l00109"></a>00109 pIn2 = pSrcA;
130 <a name="l00110"></a>00110
131 <a name="l00111"></a>00111 <span class="comment">/* srcBLen is always considered as shorter or equal to srcALen */</span>
132 <a name="l00112"></a>00112 j = <a class="code" href="arm__convolution__example__f32_8c.html#aea71286f498978c5ed3775609b974fc8">srcBLen</a>;
133 <a name="l00113"></a>00113 srcBLen = <a class="code" href="arm__convolution__example__f32_8c.html#ace48ed566e2cd6a680f0681192e6af28">srcALen</a>;
134 <a name="l00114"></a>00114 srcALen = j;
135 <a name="l00115"></a>00115 }
136 <a name="l00116"></a>00116
137 <a name="l00117"></a>00117 <span class="comment">/* conv(x,y) at n = x[n] * y[0] + x[n-1] * y[1] + x[n-2] * y[2] + ...+ x[n-N+1] * y[N -1] */</span>
138 <a name="l00118"></a>00118 <span class="comment">/* The function is internally </span>
139 <a name="l00119"></a>00119 <span class="comment"> * divided into three stages according to the number of multiplications that has to be </span>
140 <a name="l00120"></a>00120 <span class="comment"> * taken place between inputA samples and inputB samples. In the first stage of the </span>
141 <a name="l00121"></a>00121 <span class="comment"> * algorithm, the multiplications increase by one for every iteration. </span>
142 <a name="l00122"></a>00122 <span class="comment"> * In the second stage of the algorithm, srcBLen number of multiplications are done. </span>
143 <a name="l00123"></a>00123 <span class="comment"> * In the third stage of the algorithm, the multiplications decrease by one </span>
144 <a name="l00124"></a>00124 <span class="comment"> * for every iteration. */</span>
145 <a name="l00125"></a>00125
146 <a name="l00126"></a>00126 <span class="comment">/* The algorithm is implemented in three stages. </span>
147 <a name="l00127"></a>00127 <span class="comment"> The loop counters of each stage is initiated here. */</span>
148 <a name="l00128"></a>00128 blockSize1 = srcBLen - 1u;
149 <a name="l00129"></a>00129 blockSize2 = srcALen - (srcBLen - 1u);
150 <a name="l00130"></a>00130
151 <a name="l00131"></a>00131 <span class="comment">/* -------------------------- </span>
152 <a name="l00132"></a>00132 <span class="comment"> * Initializations of stage1 </span>
153 <a name="l00133"></a>00133 <span class="comment"> * -------------------------*/</span>
154 <a name="l00134"></a>00134
155 <a name="l00135"></a>00135 <span class="comment">/* sum = x[0] * y[0] </span>
156 <a name="l00136"></a>00136 <span class="comment"> * sum = x[0] * y[1] + x[1] * y[0] </span>
157 <a name="l00137"></a>00137 <span class="comment"> * .... </span>
158 <a name="l00138"></a>00138 <span class="comment"> * sum = x[0] * y[srcBlen - 1] + x[1] * y[srcBlen - 2] +...+ x[srcBLen - 1] * y[0] </span>
159 <a name="l00139"></a>00139 <span class="comment"> */</span>
160 <a name="l00140"></a>00140
161 <a name="l00141"></a>00141 <span class="comment">/* In this stage the MAC operations are increased by 1 for every iteration. </span>
162 <a name="l00142"></a>00142 <span class="comment"> The count variable holds the number of MAC operations performed */</span>
163 <a name="l00143"></a>00143 count = 1u;
164 <a name="l00144"></a>00144
165 <a name="l00145"></a>00145 <span class="comment">/* Working pointer of inputA */</span>
166 <a name="l00146"></a>00146 px = pIn1;
167 <a name="l00147"></a>00147
168 <a name="l00148"></a>00148 <span class="comment">/* Working pointer of inputB */</span>
169 <a name="l00149"></a>00149 py = pIn2;
170 <a name="l00150"></a>00150
171 <a name="l00151"></a>00151
172 <a name="l00152"></a>00152 <span class="comment">/* ------------------------ </span>
173 <a name="l00153"></a>00153 <span class="comment"> * Stage1 process </span>
174 <a name="l00154"></a>00154 <span class="comment"> * ----------------------*/</span>
175 <a name="l00155"></a>00155
176 <a name="l00156"></a>00156 <span class="comment">/* For loop unrolling by 4, this stage is divided into two. */</span>
177 <a name="l00157"></a>00157 <span class="comment">/* First part of this stage computes the MAC operations less than 4 */</span>
178 <a name="l00158"></a>00158 <span class="comment">/* Second part of this stage computes the MAC operations greater than or equal to 4 */</span>
179 <a name="l00159"></a>00159
180 <a name="l00160"></a>00160 <span class="comment">/* The first part of the stage starts here */</span>
181 <a name="l00161"></a>00161 <span class="keywordflow">while</span>((count < 4u) && (blockSize1 > 0u))
182 <a name="l00162"></a>00162 {
183 <a name="l00163"></a>00163 <span class="comment">/* Accumulator is made zero for every iteration */</span>
184 <a name="l00164"></a>00164 sum = 0;
185 <a name="l00165"></a>00165
186 <a name="l00166"></a>00166 <span class="comment">/* Loop over number of MAC operations between </span>
187 <a name="l00167"></a>00167 <span class="comment"> * inputA samples and inputB samples */</span>
188 <a name="l00168"></a>00168 k = count;
189 <a name="l00169"></a>00169
190 <a name="l00170"></a>00170 <span class="keywordflow">while</span>(k > 0u)
191 <a name="l00171"></a>00171 {
192 <a name="l00172"></a>00172 <span class="comment">/* Perform the multiply-accumulates */</span>
193 <a name="l00173"></a>00173 sum = __SMLALD(*px++, *py--, sum);
194 <a name="l00174"></a>00174
195 <a name="l00175"></a>00175 <span class="comment">/* Decrement the loop counter */</span>
196 <a name="l00176"></a>00176 k--;
197 <a name="l00177"></a>00177 }
198 <a name="l00178"></a>00178
199 <a name="l00179"></a>00179 <span class="comment">/* Store the result in the accumulator in the destination buffer. */</span>
200 <a name="l00180"></a>00180 *pOut++ = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) (__SSAT((sum >> 15), 16));
201 <a name="l00181"></a>00181
202 <a name="l00182"></a>00182 <span class="comment">/* Update the inputA and inputB pointers for next MAC calculation */</span>
203 <a name="l00183"></a>00183 py = pIn2 + count;
204 <a name="l00184"></a>00184 px = pIn1;
205 <a name="l00185"></a>00185
206 <a name="l00186"></a>00186 <span class="comment">/* Increment the MAC count */</span>
207 <a name="l00187"></a>00187 count++;
208 <a name="l00188"></a>00188
209 <a name="l00189"></a>00189 <span class="comment">/* Decrement the loop counter */</span>
210 <a name="l00190"></a>00190 blockSize1--;
211 <a name="l00191"></a>00191 }
212 <a name="l00192"></a>00192
213 <a name="l00193"></a>00193 <span class="comment">/* The second part of the stage starts here */</span>
214 <a name="l00194"></a>00194 <span class="comment">/* The internal loop, over count, is unrolled by 4 */</span>
215 <a name="l00195"></a>00195 <span class="comment">/* To, read the last two inputB samples using SIMD: </span>
216 <a name="l00196"></a>00196 <span class="comment"> * y[srcBLen] and y[srcBLen-1] coefficients, py is decremented by 1 */</span>
217 <a name="l00197"></a>00197 py = py - 1;
218 <a name="l00198"></a>00198
219 <a name="l00199"></a>00199 <span class="keywordflow">while</span>(blockSize1 > 0u)
220 <a name="l00200"></a>00200 {
221 <a name="l00201"></a>00201 <span class="comment">/* Accumulator is made zero for every iteration */</span>
222 <a name="l00202"></a>00202 sum = 0;
223 <a name="l00203"></a>00203
224 <a name="l00204"></a>00204 <span class="comment">/* Apply loop unrolling and compute 4 MACs simultaneously. */</span>
225 <a name="l00205"></a>00205 k = count >> 2u;
226 <a name="l00206"></a>00206
227 <a name="l00207"></a>00207 <span class="comment">/* First part of the processing with loop unrolling. Compute 4 MACs at a time. </span>
228 <a name="l00208"></a>00208 <span class="comment"> ** a second loop below computes MACs for the remaining 1 to 3 samples. */</span>
229 <a name="l00209"></a>00209 <span class="keywordflow">while</span>(k > 0u)
230 <a name="l00210"></a>00210 {
231 <a name="l00211"></a>00211 <span class="comment">/* Perform the multiply-accumulates */</span>
232 <a name="l00212"></a>00212 <span class="comment">/* x[0], x[1] are multiplied with y[srcBLen - 1], y[srcBLen - 2] respectively */</span>
233 <a name="l00213"></a>00213 sum = __SMLALDX(*<a class="code" href="arm__math_8h.html#a9de2e0a5785be82866bcb96012282248" title="definition to read/write two 16 bit values.">__SIMD32</a>(px)++, *<a class="code" href="arm__math_8h.html#a9de2e0a5785be82866bcb96012282248" title="definition to read/write two 16 bit values.">__SIMD32</a>(py)--, sum);
234 <a name="l00214"></a>00214 <span class="comment">/* x[2], x[3] are multiplied with y[srcBLen - 3], y[srcBLen - 4] respectively */</span>
235 <a name="l00215"></a>00215 sum = __SMLALDX(*<a class="code" href="arm__math_8h.html#a9de2e0a5785be82866bcb96012282248" title="definition to read/write two 16 bit values.">__SIMD32</a>(px)++, *<a class="code" href="arm__math_8h.html#a9de2e0a5785be82866bcb96012282248" title="definition to read/write two 16 bit values.">__SIMD32</a>(py)--, sum);
236 <a name="l00216"></a>00216
237 <a name="l00217"></a>00217 <span class="comment">/* Decrement the loop counter */</span>
238 <a name="l00218"></a>00218 k--;
239 <a name="l00219"></a>00219 }
240 <a name="l00220"></a>00220
241 <a name="l00221"></a>00221 <span class="comment">/* For the next MAC operations, the pointer py is used without SIMD </span>
242 <a name="l00222"></a>00222 <span class="comment"> * So, py is incremented by 1 */</span>
243 <a name="l00223"></a>00223 py = py + 1u;
244 <a name="l00224"></a>00224
245 <a name="l00225"></a>00225 <span class="comment">/* If the count is not a multiple of 4, compute any remaining MACs here. </span>
246 <a name="l00226"></a>00226 <span class="comment"> ** No loop unrolling is used. */</span>
247 <a name="l00227"></a>00227 k = count % 0x4u;
248 <a name="l00228"></a>00228
249 <a name="l00229"></a>00229 <span class="keywordflow">while</span>(k > 0u)
250 <a name="l00230"></a>00230 {
251 <a name="l00231"></a>00231 <span class="comment">/* Perform the multiply-accumulates */</span>
252 <a name="l00232"></a>00232 sum = __SMLALD(*px++, *py--, sum);
253 <a name="l00233"></a>00233
254 <a name="l00234"></a>00234 <span class="comment">/* Decrement the loop counter */</span>
255 <a name="l00235"></a>00235 k--;
256 <a name="l00236"></a>00236 }
257 <a name="l00237"></a>00237
258 <a name="l00238"></a>00238 <span class="comment">/* Store the result in the accumulator in the destination buffer. */</span>
259 <a name="l00239"></a>00239 *pOut++ = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) (__SSAT((sum >> 15), 16));
260 <a name="l00240"></a>00240
261 <a name="l00241"></a>00241 <span class="comment">/* Update the inputA and inputB pointers for next MAC calculation */</span>
262 <a name="l00242"></a>00242 py = pIn2 + (count - 1u);
263 <a name="l00243"></a>00243 px = pIn1;
264 <a name="l00244"></a>00244
265 <a name="l00245"></a>00245 <span class="comment">/* Increment the MAC count */</span>
266 <a name="l00246"></a>00246 count++;
267 <a name="l00247"></a>00247
268 <a name="l00248"></a>00248 <span class="comment">/* Decrement the loop counter */</span>
269 <a name="l00249"></a>00249 blockSize1--;
270 <a name="l00250"></a>00250 }
271 <a name="l00251"></a>00251
272 <a name="l00252"></a>00252 <span class="comment">/* -------------------------- </span>
273 <a name="l00253"></a>00253 <span class="comment"> * Initializations of stage2 </span>
274 <a name="l00254"></a>00254 <span class="comment"> * ------------------------*/</span>
275 <a name="l00255"></a>00255
276 <a name="l00256"></a>00256 <span class="comment">/* sum = x[0] * y[srcBLen-1] + x[1] * y[srcBLen-2] +...+ x[srcBLen-1] * y[0] </span>
277 <a name="l00257"></a>00257 <span class="comment"> * sum = x[1] * y[srcBLen-1] + x[2] * y[srcBLen-2] +...+ x[srcBLen] * y[0] </span>
278 <a name="l00258"></a>00258 <span class="comment"> * .... </span>
279 <a name="l00259"></a>00259 <span class="comment"> * sum = x[srcALen-srcBLen-2] * y[srcBLen-1] + x[srcALen] * y[srcBLen-2] +...+ x[srcALen-1] * y[0] </span>
280 <a name="l00260"></a>00260 <span class="comment"> */</span>
281 <a name="l00261"></a>00261
282 <a name="l00262"></a>00262 <span class="comment">/* Working pointer of inputA */</span>
283 <a name="l00263"></a>00263 px = pIn1;
284 <a name="l00264"></a>00264
285 <a name="l00265"></a>00265 <span class="comment">/* Working pointer of inputB */</span>
286 <a name="l00266"></a>00266 pSrc2 = pIn2 + (srcBLen - 1u);
287 <a name="l00267"></a>00267 py = pSrc2;
288 <a name="l00268"></a>00268
289 <a name="l00269"></a>00269 <span class="comment">/* Initialize inputB pointer of type q31 */</span>
290 <a name="l00270"></a>00270 pb = (<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) (py - 1u);
291 <a name="l00271"></a>00271
292 <a name="l00272"></a>00272 <span class="comment">/* count is the index by which the pointer pIn1 to be incremented */</span>
293 <a name="l00273"></a>00273 count = 1u;
294 <a name="l00274"></a>00274
295 <a name="l00275"></a>00275
296 <a name="l00276"></a>00276 <span class="comment">/* -------------------- </span>
297 <a name="l00277"></a>00277 <span class="comment"> * Stage2 process </span>
298 <a name="l00278"></a>00278 <span class="comment"> * -------------------*/</span>
299 <a name="l00279"></a>00279
300 <a name="l00280"></a>00280 <span class="comment">/* Stage2 depends on srcBLen as in this stage srcBLen number of MACS are performed. </span>
301 <a name="l00281"></a>00281 <span class="comment"> * So, to loop unroll over blockSize2, </span>
302 <a name="l00282"></a>00282 <span class="comment"> * srcBLen should be greater than or equal to 4 */</span>
303 <a name="l00283"></a>00283 <span class="keywordflow">if</span>(srcBLen >= 4u)
304 <a name="l00284"></a>00284 {
305 <a name="l00285"></a>00285 <span class="comment">/* Loop unroll over blockSize2, by 4 */</span>
306 <a name="l00286"></a>00286 blkCnt = blockSize2 >> 2u;
307 <a name="l00287"></a>00287
308 <a name="l00288"></a>00288 <span class="keywordflow">while</span>(blkCnt > 0u)
309 <a name="l00289"></a>00289 {
310 <a name="l00290"></a>00290 <span class="comment">/* Set all accumulators to zero */</span>
311 <a name="l00291"></a>00291 acc0 = 0;
312 <a name="l00292"></a>00292 acc1 = 0;
313 <a name="l00293"></a>00293 acc2 = 0;
314 <a name="l00294"></a>00294 acc3 = 0;
315 <a name="l00295"></a>00295
316 <a name="l00296"></a>00296
317 <a name="l00297"></a>00297 <span class="comment">/* read x[0], x[1] samples */</span>
318 <a name="l00298"></a>00298 x0 = *(<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) (px++);
319 <a name="l00299"></a>00299 <span class="comment">/* read x[1], x[2] samples */</span>
320 <a name="l00300"></a>00300 x1 = *(<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) (px++);
321 <a name="l00301"></a>00301
322 <a name="l00302"></a>00302
323 <a name="l00303"></a>00303 <span class="comment">/* Apply loop unrolling and compute 4 MACs simultaneously. */</span>
324 <a name="l00304"></a>00304 k = srcBLen >> 2u;
325 <a name="l00305"></a>00305
326 <a name="l00306"></a>00306 <span class="comment">/* First part of the processing with loop unrolling. Compute 4 MACs at a time. </span>
327 <a name="l00307"></a>00307 <span class="comment"> ** a second loop below computes MACs for the remaining 1 to 3 samples. */</span>
328 <a name="l00308"></a>00308 <span class="keywordflow">do</span>
329 <a name="l00309"></a>00309 {
330 <a name="l00310"></a>00310 <span class="comment">/* Read the last two inputB samples using SIMD: </span>
331 <a name="l00311"></a>00311 <span class="comment"> * y[srcBLen - 1] and y[srcBLen - 2] */</span>
332 <a name="l00312"></a>00312 c0 = *(pb--);
333 <a name="l00313"></a>00313
334 <a name="l00314"></a>00314 <span class="comment">/* acc0 += x[0] * y[srcBLen - 1] + x[1] * y[srcBLen - 2] */</span>
335 <a name="l00315"></a>00315 acc0 = __SMLALDX(x0, c0, acc0);
336 <a name="l00316"></a>00316
337 <a name="l00317"></a>00317 <span class="comment">/* acc1 += x[1] * y[srcBLen - 1] + x[2] * y[srcBLen - 2] */</span>
338 <a name="l00318"></a>00318 acc1 = __SMLALDX(x1, c0, acc1);
339 <a name="l00319"></a>00319
340 <a name="l00320"></a>00320 <span class="comment">/* Read x[2], x[3] */</span>
341 <a name="l00321"></a>00321 x2 = *(<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) (px++);
342 <a name="l00322"></a>00322
343 <a name="l00323"></a>00323 <span class="comment">/* Read x[3], x[4] */</span>
344 <a name="l00324"></a>00324 x3 = *(<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) (px++);
345 <a name="l00325"></a>00325
346 <a name="l00326"></a>00326 <span class="comment">/* acc2 += x[2] * y[srcBLen - 1] + x[3] * y[srcBLen - 2] */</span>
347 <a name="l00327"></a>00327 acc2 = __SMLALDX(x2, c0, acc2);
348 <a name="l00328"></a>00328
349 <a name="l00329"></a>00329 <span class="comment">/* acc3 += x[3] * y[srcBLen - 1] + x[4] * y[srcBLen - 2] */</span>
350 <a name="l00330"></a>00330 acc3 = __SMLALDX(x3, c0, acc3);
351 <a name="l00331"></a>00331
352 <a name="l00332"></a>00332 <span class="comment">/* Read y[srcBLen - 3] and y[srcBLen - 4] */</span>
353 <a name="l00333"></a>00333 c0 = *(pb--);
354 <a name="l00334"></a>00334
355 <a name="l00335"></a>00335 <span class="comment">/* acc0 += x[2] * y[srcBLen - 3] + x[3] * y[srcBLen - 4] */</span>
356 <a name="l00336"></a>00336 acc0 = __SMLALDX(x2, c0, acc0);
357 <a name="l00337"></a>00337
358 <a name="l00338"></a>00338 <span class="comment">/* acc1 += x[3] * y[srcBLen - 3] + x[4] * y[srcBLen - 4] */</span>
359 <a name="l00339"></a>00339 acc1 = __SMLALDX(x3, c0, acc1);
360 <a name="l00340"></a>00340
361 <a name="l00341"></a>00341 <span class="comment">/* Read x[4], x[5] */</span>
362 <a name="l00342"></a>00342 x0 = *(<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) (px++);
363 <a name="l00343"></a>00343
364 <a name="l00344"></a>00344 <span class="comment">/* Read x[5], x[6] */</span>
365 <a name="l00345"></a>00345 x1 = *(<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) (px++);
366 <a name="l00346"></a>00346
367 <a name="l00347"></a>00347 <span class="comment">/* acc2 += x[4] * y[srcBLen - 3] + x[5] * y[srcBLen - 4] */</span>
368 <a name="l00348"></a>00348 acc2 = __SMLALDX(x0, c0, acc2);
369 <a name="l00349"></a>00349
370 <a name="l00350"></a>00350 <span class="comment">/* acc3 += x[5] * y[srcBLen - 3] + x[6] * y[srcBLen - 4] */</span>
371 <a name="l00351"></a>00351 acc3 = __SMLALDX(x1, c0, acc3);
372 <a name="l00352"></a>00352
373 <a name="l00353"></a>00353 } <span class="keywordflow">while</span>(--k);
374 <a name="l00354"></a>00354
375 <a name="l00355"></a>00355 <span class="comment">/* For the next MAC operations, SIMD is not used </span>
376 <a name="l00356"></a>00356 <span class="comment"> * So, the 16 bit pointer if inputB, py is updated */</span>
377 <a name="l00357"></a>00357 py = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *) pb;
378 <a name="l00358"></a>00358 py = py + 1;
379 <a name="l00359"></a>00359
380 <a name="l00360"></a>00360 <span class="comment">/* If the srcBLen is not a multiple of 4, compute any remaining MACs here. </span>
381 <a name="l00361"></a>00361 <span class="comment"> ** No loop unrolling is used. */</span>
382 <a name="l00362"></a>00362 k = srcBLen % 0x4u;
383 <a name="l00363"></a>00363
384 <a name="l00364"></a>00364 <span class="keywordflow">if</span>(k == 1u)
385 <a name="l00365"></a>00365 {
386 <a name="l00366"></a>00366 <span class="comment">/* Read y[srcBLen - 5] */</span>
387 <a name="l00367"></a>00367 c0 = *(py);
388 <a name="l00368"></a>00368
389 <a name="l00369"></a>00369 <span class="preprocessor">#ifdef ARM_MATH_BIG_ENDIAN</span>
390 <a name="l00370"></a>00370 <span class="preprocessor"></span>
391 <a name="l00371"></a>00371 c0 = c0 << 16u;
392 <a name="l00372"></a>00372
393 <a name="l00373"></a>00373 <span class="preprocessor">#endif </span><span class="comment">/* #ifdef ARM_MATH_BIG_ENDIAN */</span>
394 <a name="l00374"></a>00374
395 <a name="l00375"></a>00375 <span class="comment">/* Read x[7] */</span>
396 <a name="l00376"></a>00376 x3 = *(<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) px++;
397 <a name="l00377"></a>00377
398 <a name="l00378"></a>00378 <span class="comment">/* Perform the multiply-accumulates */</span>
399 <a name="l00379"></a>00379 acc0 = __SMLALD(x0, c0, acc0);
400 <a name="l00380"></a>00380 acc1 = __SMLALD(x1, c0, acc1);
401 <a name="l00381"></a>00381 acc2 = __SMLALDX(x1, c0, acc2);
402 <a name="l00382"></a>00382 acc3 = __SMLALDX(x3, c0, acc3);
403 <a name="l00383"></a>00383 }
404 <a name="l00384"></a>00384
405 <a name="l00385"></a>00385 <span class="keywordflow">if</span>(k == 2u)
406 <a name="l00386"></a>00386 {
407 <a name="l00387"></a>00387 <span class="comment">/* Read y[srcBLen - 5], y[srcBLen - 6] */</span>
408 <a name="l00388"></a>00388 c0 = *(pb);
409 <a name="l00389"></a>00389
410 <a name="l00390"></a>00390 <span class="comment">/* Read x[7], x[8] */</span>
411 <a name="l00391"></a>00391 x3 = *(<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) px++;
412 <a name="l00392"></a>00392
413 <a name="l00393"></a>00393 <span class="comment">/* Read x[9] */</span>
414 <a name="l00394"></a>00394 x2 = *(<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) px++;
415 <a name="l00395"></a>00395
416 <a name="l00396"></a>00396 <span class="comment">/* Perform the multiply-accumulates */</span>
417 <a name="l00397"></a>00397 acc0 = __SMLALDX(x0, c0, acc0);
418 <a name="l00398"></a>00398 acc1 = __SMLALDX(x1, c0, acc1);
419 <a name="l00399"></a>00399 acc2 = __SMLALDX(x3, c0, acc2);
420 <a name="l00400"></a>00400 acc3 = __SMLALDX(x2, c0, acc3);
421 <a name="l00401"></a>00401 }
422 <a name="l00402"></a>00402
423 <a name="l00403"></a>00403 <span class="keywordflow">if</span>(k == 3u)
424 <a name="l00404"></a>00404 {
425 <a name="l00405"></a>00405 <span class="comment">/* Read y[srcBLen - 5], y[srcBLen - 6] */</span>
426 <a name="l00406"></a>00406 c0 = *pb--;
427 <a name="l00407"></a>00407
428 <a name="l00408"></a>00408 <span class="comment">/* Read x[7], x[8] */</span>
429 <a name="l00409"></a>00409 x3 = *(<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) px++;
430 <a name="l00410"></a>00410
431 <a name="l00411"></a>00411 <span class="comment">/* Read x[9] */</span>
432 <a name="l00412"></a>00412 x2 = *(<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) px++;
433 <a name="l00413"></a>00413
434 <a name="l00414"></a>00414 <span class="comment">/* Perform the multiply-accumulates */</span>
435 <a name="l00415"></a>00415 acc0 = __SMLALDX(x0, c0, acc0);
436 <a name="l00416"></a>00416 acc1 = __SMLALDX(x1, c0, acc1);
437 <a name="l00417"></a>00417 acc2 = __SMLALDX(x3, c0, acc2);
438 <a name="l00418"></a>00418 acc3 = __SMLALDX(x2, c0, acc3);
439 <a name="l00419"></a>00419
440 <a name="l00420"></a>00420 <span class="preprocessor">#ifdef ARM_MATH_BIG_ENDIAN</span>
441 <a name="l00421"></a>00421 <span class="preprocessor"></span>
442 <a name="l00422"></a>00422 <span class="comment">/* Read y[srcBLen - 7] */</span>
443 <a name="l00423"></a>00423 c0 = (*pb);
444 <a name="l00424"></a>00424
445 <a name="l00425"></a>00425 <span class="comment">//c0 = (c0 & 0x0000FFFF)<<16; </span>
446 <a name="l00426"></a>00426 c0 = (c0) << 16;
447 <a name="l00427"></a>00427
448 <a name="l00428"></a>00428 <span class="preprocessor">#else</span>
449 <a name="l00429"></a>00429 <span class="preprocessor"></span>
450 <a name="l00430"></a>00430 <span class="comment">/* Read y[srcBLen - 7] */</span>
451 <a name="l00431"></a>00431 c0 = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) (*pb >> 16);
452 <a name="l00432"></a>00432
453 <a name="l00433"></a>00433 <span class="preprocessor">#endif </span><span class="comment">/* #ifdef ARM_MATH_BIG_ENDIAN */</span>
454 <a name="l00434"></a>00434
455 <a name="l00435"></a>00435 <span class="comment">/* Read x[10] */</span>
456 <a name="l00436"></a>00436 x3 = *(<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) px++;
457 <a name="l00437"></a>00437
458 <a name="l00438"></a>00438 <span class="comment">/* Perform the multiply-accumulates */</span>
459 <a name="l00439"></a>00439 acc0 = __SMLALDX(x1, c0, acc0);
460 <a name="l00440"></a>00440 acc1 = __SMLALD(x2, c0, acc1);
461 <a name="l00441"></a>00441 acc2 = __SMLALDX(x2, c0, acc2);
462 <a name="l00442"></a>00442 acc3 = __SMLALDX(x3, c0, acc3);
463 <a name="l00443"></a>00443 }
464 <a name="l00444"></a>00444
465 <a name="l00445"></a>00445
466 <a name="l00446"></a>00446 <span class="comment">/* Store the results in the accumulators in the destination buffer. */</span>
467 <a name="l00447"></a>00447
468 <a name="l00448"></a>00448 <span class="preprocessor">#ifndef ARM_MATH_BIG_ENDIAN</span>
469 <a name="l00449"></a>00449 <span class="preprocessor"></span>
470 <a name="l00450"></a>00450 *<a class="code" href="arm__math_8h.html#a9de2e0a5785be82866bcb96012282248" title="definition to read/write two 16 bit values.">__SIMD32</a>(pOut)++ =
471 <a name="l00451"></a>00451 __PKHBT(__SSAT((acc0 >> 15), 16), __SSAT((acc1 >> 15), 16), 16);
472 <a name="l00452"></a>00452 *<a class="code" href="arm__math_8h.html#a9de2e0a5785be82866bcb96012282248" title="definition to read/write two 16 bit values.">__SIMD32</a>(pOut)++ =
473 <a name="l00453"></a>00453 __PKHBT(__SSAT((acc2 >> 15), 16), __SSAT((acc3 >> 15), 16), 16);
474 <a name="l00454"></a>00454
475 <a name="l00455"></a>00455 <span class="preprocessor">#else</span>
476 <a name="l00456"></a>00456 <span class="preprocessor"></span>
477 <a name="l00457"></a>00457 *<a class="code" href="arm__math_8h.html#a9de2e0a5785be82866bcb96012282248" title="definition to read/write two 16 bit values.">__SIMD32</a>(pOut)++ =
478 <a name="l00458"></a>00458 __PKHBT(__SSAT((acc1 >> 15), 16), __SSAT((acc0 >> 15), 16), 16);
479 <a name="l00459"></a>00459 *<a class="code" href="arm__math_8h.html#a9de2e0a5785be82866bcb96012282248" title="definition to read/write two 16 bit values.">__SIMD32</a>(pOut)++ =
480 <a name="l00460"></a>00460 __PKHBT(__SSAT((acc3 >> 15), 16), __SSAT((acc2 >> 15), 16), 16);
481 <a name="l00461"></a>00461
482 <a name="l00462"></a>00462 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_BIG_ENDIAN */</span>
483 <a name="l00463"></a>00463
484 <a name="l00464"></a>00464 <span class="comment">/* Update the inputA and inputB pointers for next MAC calculation */</span>
485 <a name="l00465"></a>00465 px = pIn1 + (count * 4u);
486 <a name="l00466"></a>00466 py = pSrc2;
487 <a name="l00467"></a>00467 pb = (<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a> *) (py - 1);
488 <a name="l00468"></a>00468
489 <a name="l00469"></a>00469 <span class="comment">/* Increment the pointer pIn1 index, count by 1 */</span>
490 <a name="l00470"></a>00470 count++;
491 <a name="l00471"></a>00471
492 <a name="l00472"></a>00472 <span class="comment">/* Decrement the loop counter */</span>
493 <a name="l00473"></a>00473 blkCnt--;
494 <a name="l00474"></a>00474 }
495 <a name="l00475"></a>00475
496 <a name="l00476"></a>00476 <span class="comment">/* If the blockSize2 is not a multiple of 4, compute any remaining output samples here. </span>
497 <a name="l00477"></a>00477 <span class="comment"> ** No loop unrolling is used. */</span>
498 <a name="l00478"></a>00478 blkCnt = blockSize2 % 0x4u;
499 <a name="l00479"></a>00479
500 <a name="l00480"></a>00480 <span class="keywordflow">while</span>(blkCnt > 0u)
501 <a name="l00481"></a>00481 {
502 <a name="l00482"></a>00482 <span class="comment">/* Accumulator is made zero for every iteration */</span>
503 <a name="l00483"></a>00483 sum = 0;
504 <a name="l00484"></a>00484
505 <a name="l00485"></a>00485 <span class="comment">/* Apply loop unrolling and compute 4 MACs simultaneously. */</span>
506 <a name="l00486"></a>00486 k = srcBLen >> 2u;
507 <a name="l00487"></a>00487
508 <a name="l00488"></a>00488 <span class="comment">/* First part of the processing with loop unrolling. Compute 4 MACs at a time. </span>
509 <a name="l00489"></a>00489 <span class="comment"> ** a second loop below computes MACs for the remaining 1 to 3 samples. */</span>
510 <a name="l00490"></a>00490 <span class="keywordflow">while</span>(k > 0u)
511 <a name="l00491"></a>00491 {
512 <a name="l00492"></a>00492 <span class="comment">/* Perform the multiply-accumulates */</span>
513 <a name="l00493"></a>00493 sum += (<a class="code" href="arm__math_8h.html#a5aea1cb12fc02d9d44c8abf217eaa5c6" title="64-bit fractional data type in 1.63 format.">q63_t</a>) ((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) * px++ * *py--);
514 <a name="l00494"></a>00494 sum += (<a class="code" href="arm__math_8h.html#a5aea1cb12fc02d9d44c8abf217eaa5c6" title="64-bit fractional data type in 1.63 format.">q63_t</a>) ((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) * px++ * *py--);
515 <a name="l00495"></a>00495 sum += (<a class="code" href="arm__math_8h.html#a5aea1cb12fc02d9d44c8abf217eaa5c6" title="64-bit fractional data type in 1.63 format.">q63_t</a>) ((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) * px++ * *py--);
516 <a name="l00496"></a>00496 sum += (<a class="code" href="arm__math_8h.html#a5aea1cb12fc02d9d44c8abf217eaa5c6" title="64-bit fractional data type in 1.63 format.">q63_t</a>) ((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) * px++ * *py--);
517 <a name="l00497"></a>00497
518 <a name="l00498"></a>00498 <span class="comment">/* Decrement the loop counter */</span>
519 <a name="l00499"></a>00499 k--;
520 <a name="l00500"></a>00500 }
521 <a name="l00501"></a>00501
522 <a name="l00502"></a>00502 <span class="comment">/* If the srcBLen is not a multiple of 4, compute any remaining MACs here. </span>
523 <a name="l00503"></a>00503 <span class="comment"> ** No loop unrolling is used. */</span>
524 <a name="l00504"></a>00504 k = srcBLen % 0x4u;
525 <a name="l00505"></a>00505
526 <a name="l00506"></a>00506 <span class="keywordflow">while</span>(k > 0u)
527 <a name="l00507"></a>00507 {
528 <a name="l00508"></a>00508 <span class="comment">/* Perform the multiply-accumulates */</span>
529 <a name="l00509"></a>00509 sum += (<a class="code" href="arm__math_8h.html#a5aea1cb12fc02d9d44c8abf217eaa5c6" title="64-bit fractional data type in 1.63 format.">q63_t</a>) ((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) * px++ * *py--);
530 <a name="l00510"></a>00510
531 <a name="l00511"></a>00511 <span class="comment">/* Decrement the loop counter */</span>
532 <a name="l00512"></a>00512 k--;
533 <a name="l00513"></a>00513 }
534 <a name="l00514"></a>00514
535 <a name="l00515"></a>00515 <span class="comment">/* Store the result in the accumulator in the destination buffer. */</span>
536 <a name="l00516"></a>00516 *pOut++ = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) (__SSAT(sum >> 15, 16));
537 <a name="l00517"></a>00517
538 <a name="l00518"></a>00518 <span class="comment">/* Update the inputA and inputB pointers for next MAC calculation */</span>
539 <a name="l00519"></a>00519 px = pIn1 + count;
540 <a name="l00520"></a>00520 py = pSrc2;
541 <a name="l00521"></a>00521
542 <a name="l00522"></a>00522 <span class="comment">/* Increment the pointer pIn1 index, count by 1 */</span>
543 <a name="l00523"></a>00523 count++;
544 <a name="l00524"></a>00524
545 <a name="l00525"></a>00525 <span class="comment">/* Decrement the loop counter */</span>
546 <a name="l00526"></a>00526 blkCnt--;
547 <a name="l00527"></a>00527 }
548 <a name="l00528"></a>00528 }
549 <a name="l00529"></a>00529 <span class="keywordflow">else</span>
550 <a name="l00530"></a>00530 {
551 <a name="l00531"></a>00531 <span class="comment">/* If the srcBLen is not a multiple of 4, </span>
552 <a name="l00532"></a>00532 <span class="comment"> * the blockSize2 loop cannot be unrolled by 4 */</span>
553 <a name="l00533"></a>00533 blkCnt = blockSize2;
554 <a name="l00534"></a>00534
555 <a name="l00535"></a>00535 <span class="keywordflow">while</span>(blkCnt > 0u)
556 <a name="l00536"></a>00536 {
557 <a name="l00537"></a>00537 <span class="comment">/* Accumulator is made zero for every iteration */</span>
558 <a name="l00538"></a>00538 sum = 0;
559 <a name="l00539"></a>00539
560 <a name="l00540"></a>00540 <span class="comment">/* srcBLen number of MACS should be performed */</span>
561 <a name="l00541"></a>00541 k = <a class="code" href="arm__convolution__example__f32_8c.html#aea71286f498978c5ed3775609b974fc8">srcBLen</a>;
562 <a name="l00542"></a>00542
563 <a name="l00543"></a>00543 <span class="keywordflow">while</span>(k > 0u)
564 <a name="l00544"></a>00544 {
565 <a name="l00545"></a>00545 <span class="comment">/* Perform the multiply-accumulate */</span>
566 <a name="l00546"></a>00546 sum += (<a class="code" href="arm__math_8h.html#a5aea1cb12fc02d9d44c8abf217eaa5c6" title="64-bit fractional data type in 1.63 format.">q63_t</a>) ((<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) * px++ * *py--);
567 <a name="l00547"></a>00547
568 <a name="l00548"></a>00548 <span class="comment">/* Decrement the loop counter */</span>
569 <a name="l00549"></a>00549 k--;
570 <a name="l00550"></a>00550 }
571 <a name="l00551"></a>00551
572 <a name="l00552"></a>00552 <span class="comment">/* Store the result in the accumulator in the destination buffer. */</span>
573 <a name="l00553"></a>00553 *pOut++ = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) (__SSAT(sum >> 15, 16));
574 <a name="l00554"></a>00554
575 <a name="l00555"></a>00555 <span class="comment">/* Update the inputA and inputB pointers for next MAC calculation */</span>
576 <a name="l00556"></a>00556 px = pIn1 + count;
577 <a name="l00557"></a>00557 py = pSrc2;
578 <a name="l00558"></a>00558
579 <a name="l00559"></a>00559 <span class="comment">/* Increment the MAC count */</span>
580 <a name="l00560"></a>00560 count++;
581 <a name="l00561"></a>00561
582 <a name="l00562"></a>00562 <span class="comment">/* Decrement the loop counter */</span>
583 <a name="l00563"></a>00563 blkCnt--;
584 <a name="l00564"></a>00564 }
585 <a name="l00565"></a>00565 }
586 <a name="l00566"></a>00566
587 <a name="l00567"></a>00567
588 <a name="l00568"></a>00568 <span class="comment">/* -------------------------- </span>
589 <a name="l00569"></a>00569 <span class="comment"> * Initializations of stage3 </span>
590 <a name="l00570"></a>00570 <span class="comment"> * -------------------------*/</span>
591 <a name="l00571"></a>00571
592 <a name="l00572"></a>00572 <span class="comment">/* sum += x[srcALen-srcBLen+1] * y[srcBLen-1] + x[srcALen-srcBLen+2] * y[srcBLen-2] +...+ x[srcALen-1] * y[1] </span>
593 <a name="l00573"></a>00573 <span class="comment"> * sum += x[srcALen-srcBLen+2] * y[srcBLen-1] + x[srcALen-srcBLen+3] * y[srcBLen-2] +...+ x[srcALen-1] * y[2] </span>
594 <a name="l00574"></a>00574 <span class="comment"> * .... </span>
595 <a name="l00575"></a>00575 <span class="comment"> * sum += x[srcALen-2] * y[srcBLen-1] + x[srcALen-1] * y[srcBLen-2] </span>
596 <a name="l00576"></a>00576 <span class="comment"> * sum += x[srcALen-1] * y[srcBLen-1] </span>
597 <a name="l00577"></a>00577 <span class="comment"> */</span>
598 <a name="l00578"></a>00578
599 <a name="l00579"></a>00579 <span class="comment">/* In this stage the MAC operations are decreased by 1 for every iteration. </span>
600 <a name="l00580"></a>00580 <span class="comment"> The blockSize3 variable holds the number of MAC operations performed */</span>
601 <a name="l00581"></a>00581
602 <a name="l00582"></a>00582 blockSize3 = srcBLen - 1u;
603 <a name="l00583"></a>00583
604 <a name="l00584"></a>00584 <span class="comment">/* Working pointer of inputA */</span>
605 <a name="l00585"></a>00585 pSrc1 = (pIn1 + <a class="code" href="arm__convolution__example__f32_8c.html#ace48ed566e2cd6a680f0681192e6af28">srcALen</a>) - (srcBLen - 1u);
606 <a name="l00586"></a>00586 px = pSrc1;
607 <a name="l00587"></a>00587
608 <a name="l00588"></a>00588 <span class="comment">/* Working pointer of inputB */</span>
609 <a name="l00589"></a>00589 pSrc2 = pIn2 + (srcBLen - 1u);
610 <a name="l00590"></a>00590 pIn2 = pSrc2 - 1u;
611 <a name="l00591"></a>00591 py = pIn2;
612 <a name="l00592"></a>00592
613 <a name="l00593"></a>00593 <span class="comment">/* ------------------- </span>
614 <a name="l00594"></a>00594 <span class="comment"> * Stage3 process </span>
615 <a name="l00595"></a>00595 <span class="comment"> * ------------------*/</span>
616 <a name="l00596"></a>00596
617 <a name="l00597"></a>00597 <span class="comment">/* For loop unrolling by 4, this stage is divided into two. */</span>
618 <a name="l00598"></a>00598 <span class="comment">/* First part of this stage computes the MAC operations greater than 4 */</span>
619 <a name="l00599"></a>00599 <span class="comment">/* Second part of this stage computes the MAC operations less than or equal to 4 */</span>
620 <a name="l00600"></a>00600
621 <a name="l00601"></a>00601 <span class="comment">/* The first part of the stage starts here */</span>
622 <a name="l00602"></a>00602 j = blockSize3 >> 2u;
623 <a name="l00603"></a>00603
624 <a name="l00604"></a>00604 <span class="keywordflow">while</span>((j > 0u) && (blockSize3 > 0u))
625 <a name="l00605"></a>00605 {
626 <a name="l00606"></a>00606 <span class="comment">/* Accumulator is made zero for every iteration */</span>
627 <a name="l00607"></a>00607 sum = 0;
628 <a name="l00608"></a>00608
629 <a name="l00609"></a>00609 <span class="comment">/* Apply loop unrolling and compute 4 MACs simultaneously. */</span>
630 <a name="l00610"></a>00610 k = blockSize3 >> 2u;
631 <a name="l00611"></a>00611
632 <a name="l00612"></a>00612 <span class="comment">/* First part of the processing with loop unrolling. Compute 4 MACs at a time. </span>
633 <a name="l00613"></a>00613 <span class="comment"> ** a second loop below computes MACs for the remaining 1 to 3 samples. */</span>
634 <a name="l00614"></a>00614 <span class="keywordflow">while</span>(k > 0u)
635 <a name="l00615"></a>00615 {
636 <a name="l00616"></a>00616 <span class="comment">/* x[srcALen - srcBLen + 1], x[srcALen - srcBLen + 2] are multiplied </span>
637 <a name="l00617"></a>00617 <span class="comment"> * with y[srcBLen - 1], y[srcBLen - 2] respectively */</span>
638 <a name="l00618"></a>00618 sum = __SMLALDX(*<a class="code" href="arm__math_8h.html#a9de2e0a5785be82866bcb96012282248" title="definition to read/write two 16 bit values.">__SIMD32</a>(px)++, *<a class="code" href="arm__math_8h.html#a9de2e0a5785be82866bcb96012282248" title="definition to read/write two 16 bit values.">__SIMD32</a>(py)--, sum);
639 <a name="l00619"></a>00619 <span class="comment">/* x[srcALen - srcBLen + 3], x[srcALen - srcBLen + 4] are multiplied </span>
640 <a name="l00620"></a>00620 <span class="comment"> * with y[srcBLen - 3], y[srcBLen - 4] respectively */</span>
641 <a name="l00621"></a>00621 sum = __SMLALDX(*<a class="code" href="arm__math_8h.html#a9de2e0a5785be82866bcb96012282248" title="definition to read/write two 16 bit values.">__SIMD32</a>(px)++, *<a class="code" href="arm__math_8h.html#a9de2e0a5785be82866bcb96012282248" title="definition to read/write two 16 bit values.">__SIMD32</a>(py)--, sum);
642 <a name="l00622"></a>00622
643 <a name="l00623"></a>00623 <span class="comment">/* Decrement the loop counter */</span>
644 <a name="l00624"></a>00624 k--;
645 <a name="l00625"></a>00625 }
646 <a name="l00626"></a>00626
647 <a name="l00627"></a>00627 <span class="comment">/* For the next MAC operations, the pointer py is used without SIMD </span>
648 <a name="l00628"></a>00628 <span class="comment"> * So, py is incremented by 1 */</span>
649 <a name="l00629"></a>00629 py = py + 1u;
650 <a name="l00630"></a>00630
651 <a name="l00631"></a>00631 <span class="comment">/* If the blockSize3 is not a multiple of 4, compute any remaining MACs here. </span>
652 <a name="l00632"></a>00632 <span class="comment"> ** No loop unrolling is used. */</span>
653 <a name="l00633"></a>00633 k = blockSize3 % 0x4u;
654 <a name="l00634"></a>00634
655 <a name="l00635"></a>00635 <span class="keywordflow">while</span>(k > 0u)
656 <a name="l00636"></a>00636 {
657 <a name="l00637"></a>00637 <span class="comment">/* sum += x[srcALen - srcBLen + 5] * y[srcBLen - 5] */</span>
658 <a name="l00638"></a>00638 sum = __SMLALD(*px++, *py--, sum);
659 <a name="l00639"></a>00639
660 <a name="l00640"></a>00640 <span class="comment">/* Decrement the loop counter */</span>
661 <a name="l00641"></a>00641 k--;
662 <a name="l00642"></a>00642 }
663 <a name="l00643"></a>00643
664 <a name="l00644"></a>00644 <span class="comment">/* Store the result in the accumulator in the destination buffer. */</span>
665 <a name="l00645"></a>00645 *pOut++ = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) (__SSAT((sum >> 15), 16));
666 <a name="l00646"></a>00646
667 <a name="l00647"></a>00647 <span class="comment">/* Update the inputA and inputB pointers for next MAC calculation */</span>
668 <a name="l00648"></a>00648 px = ++pSrc1;
669 <a name="l00649"></a>00649 py = pIn2;
670 <a name="l00650"></a>00650
671 <a name="l00651"></a>00651 <span class="comment">/* Decrement the loop counter */</span>
672 <a name="l00652"></a>00652 blockSize3--;
673 <a name="l00653"></a>00653
674 <a name="l00654"></a>00654 j--;
675 <a name="l00655"></a>00655 }
676 <a name="l00656"></a>00656
677 <a name="l00657"></a>00657 <span class="comment">/* The second part of the stage starts here */</span>
678 <a name="l00658"></a>00658 <span class="comment">/* SIMD is not used for the next MAC operations, </span>
679 <a name="l00659"></a>00659 <span class="comment"> * so pointer py is updated to read only one sample at a time */</span>
680 <a name="l00660"></a>00660 py = py + 1u;
681 <a name="l00661"></a>00661
682 <a name="l00662"></a>00662 <span class="keywordflow">while</span>(blockSize3 > 0u)
683 <a name="l00663"></a>00663 {
684 <a name="l00664"></a>00664 <span class="comment">/* Accumulator is made zero for every iteration */</span>
685 <a name="l00665"></a>00665 sum = 0;
686 <a name="l00666"></a>00666
687 <a name="l00667"></a>00667 <span class="comment">/* Apply loop unrolling and compute 4 MACs simultaneously. */</span>
688 <a name="l00668"></a>00668 k = blockSize3;
689 <a name="l00669"></a>00669
690 <a name="l00670"></a>00670 <span class="keywordflow">while</span>(k > 0u)
691 <a name="l00671"></a>00671 {
692 <a name="l00672"></a>00672 <span class="comment">/* Perform the multiply-accumulates */</span>
693 <a name="l00673"></a>00673 <span class="comment">/* sum += x[srcALen-1] * y[srcBLen-1] */</span>
694 <a name="l00674"></a>00674 sum = __SMLALD(*px++, *py--, sum);
695 <a name="l00675"></a>00675
696 <a name="l00676"></a>00676 <span class="comment">/* Decrement the loop counter */</span>
697 <a name="l00677"></a>00677 k--;
698 <a name="l00678"></a>00678 }
699 <a name="l00679"></a>00679
700 <a name="l00680"></a>00680 <span class="comment">/* Store the result in the accumulator in the destination buffer. */</span>
701 <a name="l00681"></a>00681 *pOut++ = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) (__SSAT((sum >> 15), 16));
702 <a name="l00682"></a>00682
703 <a name="l00683"></a>00683 <span class="comment">/* Update the inputA and inputB pointers for next MAC calculation */</span>
704 <a name="l00684"></a>00684 px = ++pSrc1;
705 <a name="l00685"></a>00685 py = pSrc2;
706 <a name="l00686"></a>00686
707 <a name="l00687"></a>00687 <span class="comment">/* Decrement the loop counter */</span>
708 <a name="l00688"></a>00688 blockSize3--;
709 <a name="l00689"></a>00689 }
710 <a name="l00690"></a>00690
711 <a name="l00691"></a>00691 <span class="preprocessor">#else</span>
712 <a name="l00692"></a>00692 <span class="preprocessor"></span>
713 <a name="l00693"></a>00693 <span class="comment">/* Run the below code for Cortex-M0 */</span>
714 <a name="l00694"></a>00694
715 <a name="l00695"></a>00695 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *pIn1 = pSrcA; <span class="comment">/* input pointer */</span>
716 <a name="l00696"></a>00696 <a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a> *pIn2 = pSrcB; <span class="comment">/* coefficient pointer */</span>
717 <a name="l00697"></a>00697 <a class="code" href="arm__math_8h.html#a5aea1cb12fc02d9d44c8abf217eaa5c6" title="64-bit fractional data type in 1.63 format.">q63_t</a> sum; <span class="comment">/* Accumulator */</span>
718 <a name="l00698"></a>00698 uint32_t i, j; <span class="comment">/* loop counter */</span>
719 <a name="l00699"></a>00699
720 <a name="l00700"></a>00700 <span class="comment">/* Loop to calculate output of convolution for output length number of times */</span>
721 <a name="l00701"></a>00701 <span class="keywordflow">for</span> (i = 0; i < (srcALen + srcBLen - 1); i++)
722 <a name="l00702"></a>00702 {
723 <a name="l00703"></a>00703 <span class="comment">/* Initialize sum with zero to carry on MAC operations */</span>
724 <a name="l00704"></a>00704 sum = 0;
725 <a name="l00705"></a>00705
726 <a name="l00706"></a>00706 <span class="comment">/* Loop to perform MAC operations according to convolution equation */</span>
727 <a name="l00707"></a>00707 <span class="keywordflow">for</span> (j = 0; j <= i; j++)
728 <a name="l00708"></a>00708 {
729 <a name="l00709"></a>00709 <span class="comment">/* Check the array limitations */</span>
730 <a name="l00710"></a>00710 <span class="keywordflow">if</span>(((i - j) < <a class="code" href="arm__convolution__example__f32_8c.html#aea71286f498978c5ed3775609b974fc8">srcBLen</a>) && (j < srcALen))
731 <a name="l00711"></a>00711 {
732 <a name="l00712"></a>00712 <span class="comment">/* z[i] += x[i-j] * y[j] */</span>
733 <a name="l00713"></a>00713 sum += (<a class="code" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0" title="32-bit fractional data type in 1.31 format.">q31_t</a>) pIn1[j] * (pIn2[i - j]);
734 <a name="l00714"></a>00714 }
735 <a name="l00715"></a>00715 }
736 <a name="l00716"></a>00716
737 <a name="l00717"></a>00717 <span class="comment">/* Store the output in the destination buffer */</span>
738 <a name="l00718"></a>00718 pDst[i] = (<a class="code" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea" title="16-bit fractional data type in 1.15 format.">q15_t</a>) __SSAT((sum >> 15u), 16u);
739 <a name="l00719"></a>00719 }
740 <a name="l00720"></a>00720
741 <a name="l00721"></a>00721 <span class="preprocessor">#endif </span><span class="comment">/* #ifndef ARM_MATH_CM0 */</span>
742 <a name="l00722"></a>00722
743 <a name="l00723"></a>00723 }
744 <a name="l00724"></a>00724
747 <!--- window showing the filter options -->
748 <div id="MSearchSelectWindow"
749 onmouseover="return searchBox.OnSearchSelectShow()"
750 onmouseout="return searchBox.OnSearchSelectHide()"
751 onkeydown="return searchBox.OnSearchSelectKey(event)">
752 <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>
754 <!-- iframe showing the search results (closed by default) -->
755 <div id="MSearchResultsWindow">
756 <iframe src="" frameborder="0"
757 name="MSearchResults" id="MSearchResults">
761 <hr class="footer"/><address class="footer"><small>Generated on Fri Jul 15 2011 13:16:16 for CMSIS DSP Software Library by 
762 <a href="http://www.doxygen.org/index.html">
763 <img class="footer" src="doxygen.png" alt="doxygen"/></a> 1.7.2 </small></address>
projects / fw / stlink / blob