1 /* ----------------------------------------------------------------------
2 * Project: CMSIS DSP Library
3 * Title: arm_cmplx_mag_squared_q15.c
4 * Description: Q15 complex magnitude squared
5 *
6 * $Date: 23 April 2021
7 * $Revision: V1.9.0
8 *
9 * Target Processor: Cortex-M and Cortex-A cores
10 * -------------------------------------------------------------------- */
11 /*
12 * Copyright (C) 2010-2021 ARM Limited or its affiliates. All rights reserved.
13 *
14 * SPDX-License-Identifier: Apache-2.0
15 *
16 * Licensed under the Apache License, Version 2.0 (the License); you may
17 * not use this file except in compliance with the License.
18 * You may obtain a copy of the License at
19 *
20 * www.apache.org/licenses/LICENSE-2.0
21 *
22 * Unless required by applicable law or agreed to in writing, software
23 * distributed under the License is distributed on an AS IS BASIS, WITHOUT
24 * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
25 * See the License for the specific language governing permissions and
26 * limitations under the License.
27 */
28
29 #include "dsp/complex_math_functions.h"
30
31 /**
32 @ingroup groupCmplxMath
33 */
34
35 /**
36 @addtogroup cmplx_mag_squared
37 @{
38 */
39
40 /**
41 @brief Q15 complex magnitude squared.
42 @param[in] pSrc points to input vector
43 @param[out] pDst points to output vector
44 @param[in] numSamples number of samples in each vector
45
46 @par Scaling and Overflow Behavior
47 The function implements 1.15 by 1.15 multiplications and finally output is converted into 3.13 format.
48 */
49
50 #if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
51
arm_cmplx_mag_squared_q15(const q15_t * pSrc,q15_t * pDst,uint32_t numSamples)52 void arm_cmplx_mag_squared_q15(
53 const q15_t * pSrc,
54 q15_t * pDst,
55 uint32_t numSamples)
56 {
57 int32_t blockSize = numSamples; /* loop counters */
58 uint32_t blkCnt; /* loop counters */
59 q31_t in;
60 q31_t acc0; /* Accumulators */
61 q15x8x2_t vecSrc;
62 q15x8_t vReal, vImag;
63 q15x8_t vMagSq;
64
65
66 blkCnt = blockSize >> 3;
67 while (blkCnt > 0U)
68 {
69 vecSrc = vld2q(pSrc);
70 vReal = vmulhq(vecSrc.val[0], vecSrc.val[0]);
71 vImag = vmulhq(vecSrc.val[1], vecSrc.val[1]);
72 vMagSq = vqaddq(vReal, vImag);
73 vMagSq = vshrq(vMagSq, 1);
74
75 vst1q(pDst, vMagSq);
76
77 pSrc += 16;
78 pDst += 8;
79 /*
80 * Decrement the blkCnt loop counter
81 * Advance vector source and destination pointers
82 */
83 blkCnt --;
84 }
85
86 /*
87 * tail
88 */
89 blkCnt = blockSize & 7;
90 while (blkCnt > 0U)
91 {
92 /* C[0] = (A[0] * A[0] + A[1] * A[1]) */
93
94 in = read_q15x2_ia ((q15_t **) &pSrc);
95 acc0 = __SMUAD(in, in);
96
97 /* store result in 3.13 format in destination buffer. */
98 *pDst++ = (q15_t) (acc0 >> 17);
99
100
101 /* Decrement loop counter */
102 blkCnt--;
103 }
104
105 }
106
107 #else
arm_cmplx_mag_squared_q15(const q15_t * pSrc,q15_t * pDst,uint32_t numSamples)108 void arm_cmplx_mag_squared_q15(
109 const q15_t * pSrc,
110 q15_t * pDst,
111 uint32_t numSamples)
112 {
113 uint32_t blkCnt; /* Loop counter */
114
115 #if defined (ARM_MATH_DSP)
116 q31_t in;
117 q31_t acc0; /* Accumulators */
118 #else
119 q15_t real, imag; /* Temporary input variables */
120 q31_t acc0, acc1; /* Accumulators */
121 #endif
122
123 #if defined (ARM_MATH_LOOPUNROLL)
124
125 /* Loop unrolling: Compute 4 outputs at a time */
126 blkCnt = numSamples >> 2U;
127
128 while (blkCnt > 0U)
129 {
130 /* C[0] = (A[0] * A[0] + A[1] * A[1]) */
131
132 #if defined (ARM_MATH_DSP)
133 in = read_q15x2_ia (&pSrc);
134 acc0 = __SMUAD(in, in);
135 /* store result in 3.13 format in destination buffer. */
136 *pDst++ = (q15_t) (acc0 >> 17);
137
138 in = read_q15x2_ia (&pSrc);
139 acc0 = __SMUAD(in, in);
140 *pDst++ = (q15_t) (acc0 >> 17);
141
142 in = read_q15x2_ia (&pSrc);
143 acc0 = __SMUAD(in, in);
144 *pDst++ = (q15_t) (acc0 >> 17);
145
146 in = read_q15x2_ia (&pSrc);
147 acc0 = __SMUAD(in, in);
148 *pDst++ = (q15_t) (acc0 >> 17);
149 #else
150 real = *pSrc++;
151 imag = *pSrc++;
152 acc0 = ((q31_t) real * real);
153 acc1 = ((q31_t) imag * imag);
154 /* store result in 3.13 format in destination buffer. */
155 *pDst++ = (q15_t) (((q63_t) acc0 + acc1) >> 17);
156
157 real = *pSrc++;
158 imag = *pSrc++;
159 acc0 = ((q31_t) real * real);
160 acc1 = ((q31_t) imag * imag);
161 *pDst++ = (q15_t) (((q63_t) acc0 + acc1) >> 17);
162
163 real = *pSrc++;
164 imag = *pSrc++;
165 acc0 = ((q31_t) real * real);
166 acc1 = ((q31_t) imag * imag);
167 *pDst++ = (q15_t) (((q63_t) acc0 + acc1) >> 17);
168
169 real = *pSrc++;
170 imag = *pSrc++;
171 acc0 = ((q31_t) real * real);
172 acc1 = ((q31_t) imag * imag);
173 *pDst++ = (q15_t) (((q63_t) acc0 + acc1) >> 17);
174 #endif /* #if defined (ARM_MATH_DSP) */
175
176 /* Decrement loop counter */
177 blkCnt--;
178 }
179
180 /* Loop unrolling: Compute remaining outputs */
181 blkCnt = numSamples % 0x4U;
182
183 #else
184
185 /* Initialize blkCnt with number of samples */
186 blkCnt = numSamples;
187
188 #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
189
190 while (blkCnt > 0U)
191 {
192 /* C[0] = (A[0] * A[0] + A[1] * A[1]) */
193
194 #if defined (ARM_MATH_DSP)
195 in = read_q15x2_ia (&pSrc);
196 acc0 = __SMUAD(in, in);
197
198 /* store result in 3.13 format in destination buffer. */
199 *pDst++ = (q15_t) (acc0 >> 17);
200 #else
201 real = *pSrc++;
202 imag = *pSrc++;
203 acc0 = ((q31_t) real * real);
204 acc1 = ((q31_t) imag * imag);
205
206 /* store result in 3.13 format in destination buffer. */
207 *pDst++ = (q15_t) (((q63_t) acc0 + acc1) >> 17);
208 #endif
209
210 /* Decrement loop counter */
211 blkCnt--;
212 }
213
214 }
215
216 #endif /* defined(ARM_MATH_MVEI) */
217
218 /**
219 @} end of cmplx_mag_squared group
220 */
221
