1 /* ----------------------------------------------------------------------
2 * Project: CMSIS DSP Library
3 * Title: arm_cmplx_mult_real_q31.c
4 * Description: Q31 complex by real multiplication
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 CmplxByRealMult
37 @{
38 */
39
40 /**
41 @brief Q31 complex-by-real multiplication.
42 @param[in] pSrcCmplx points to complex input vector
43 @param[in] pSrcReal points to real input vector
44 @param[out] pCmplxDst points to complex output vector
45 @param[in] numSamples number of samples in each vector
46
47 @par Scaling and Overflow Behavior
48 The function uses saturating arithmetic.
49 Results outside of the allowable Q31 range[0x80000000 0x7FFFFFFF] are saturated.
50 */
51
52 #if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
arm_cmplx_mult_real_q31(const q31_t * pSrcCmplx,const q31_t * pSrcReal,q31_t * pCmplxDst,uint32_t numSamples)53 ARM_DSP_ATTRIBUTE void arm_cmplx_mult_real_q31(
54 const q31_t * pSrcCmplx,
55 const q31_t * pSrcReal,
56 q31_t * pCmplxDst,
57 uint32_t numSamples)
58 {
59
60 static const uint32_t stride_cmplx_x_real_32[4] = {
61 0, 0, 1, 1
62 };
63 q31x4_t rVec;
64 q31x4_t cmplxVec;
65 q31x4_t dstVec;
66 uint32x4_t strideVec;
67 uint32_t blockSizeC = numSamples * CMPLX_DIM; /* loop counters */
68 uint32_t blkCnt;
69 q31_t in;
70
71 /*
72 * stride vector for pairs of real generation
73 */
74 strideVec = vld1q(stride_cmplx_x_real_32);
75
76 /* Compute 4 complex outputs at a time */
77 blkCnt = blockSizeC >> 2;
78 while (blkCnt > 0U)
79 {
80 cmplxVec = vld1q(pSrcCmplx);
81 rVec = vldrwq_gather_shifted_offset_s32(pSrcReal, strideVec);
82 dstVec = vqdmulhq(cmplxVec, rVec);
83 vst1q(pCmplxDst, dstVec);
84
85 pSrcReal += 2;
86 pSrcCmplx += 4;
87 pCmplxDst += 4;
88 blkCnt --;
89 }
90
91 blkCnt = (blockSizeC & 3) >> 1;
92 while (blkCnt > 0U)
93 {
94 /* C[2 * i ] = A[2 * i ] * B[i]. */
95 /* C[2 * i + 1] = A[2 * i + 1] * B[i]. */
96
97 in = *pSrcReal++;
98 /* store saturated result in 1.31 format to destination buffer */
99 *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
100 *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
101
102 /* Decrement loop counter */
103 blkCnt--;
104 }
105 }
106 #else
arm_cmplx_mult_real_q31(const q31_t * pSrcCmplx,const q31_t * pSrcReal,q31_t * pCmplxDst,uint32_t numSamples)107 ARM_DSP_ATTRIBUTE void arm_cmplx_mult_real_q31(
108 const q31_t * pSrcCmplx,
109 const q31_t * pSrcReal,
110 q31_t * pCmplxDst,
111 uint32_t numSamples)
112 {
113 uint32_t blkCnt; /* Loop counter */
114 q31_t in; /* Temporary variable */
115
116 #if defined (ARM_MATH_LOOPUNROLL)
117
118 /* Loop unrolling: Compute 4 outputs at a time */
119 blkCnt = numSamples >> 2U;
120
121 while (blkCnt > 0U)
122 {
123 /* C[2 * i ] = A[2 * i ] * B[i]. */
124 /* C[2 * i + 1] = A[2 * i + 1] * B[i]. */
125
126 in = *pSrcReal++;
127 #if defined (ARM_MATH_DSP)
128 /* store saturated result in 1.31 format to destination buffer */
129 *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
130 *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
131 #else
132 /* store result in destination buffer. */
133 *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
134 *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
135 #endif
136
137 in = *pSrcReal++;
138 #if defined (ARM_MATH_DSP)
139 *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
140 *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
141 #else
142 *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
143 *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
144 #endif
145
146 in = *pSrcReal++;
147 #if defined (ARM_MATH_DSP)
148 *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
149 *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
150 #else
151 *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
152 *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
153 #endif
154
155 in = *pSrcReal++;
156 #if defined (ARM_MATH_DSP)
157 *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
158 *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
159 #else
160 *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
161 *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
162 #endif
163
164 /* Decrement loop counter */
165 blkCnt--;
166 }
167
168 /* Loop unrolling: Compute remaining outputs */
169 blkCnt = numSamples % 0x4U;
170
171 #else
172
173 /* Initialize blkCnt with number of samples */
174 blkCnt = numSamples;
175
176 #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
177
178 while (blkCnt > 0U)
179 {
180 /* C[2 * i ] = A[2 * i ] * B[i]. */
181 /* C[2 * i + 1] = A[2 * i + 1] * B[i]. */
182
183 in = *pSrcReal++;
184 #if defined (ARM_MATH_DSP)
185 /* store saturated result in 1.31 format to destination buffer */
186 *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
187 *pCmplxDst++ = (__SSAT((q31_t) (((q63_t) *pSrcCmplx++ * in) >> 32), 31) << 1);
188 #else
189 /* store result in destination buffer. */
190 *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
191 *pCmplxDst++ = (q31_t) clip_q63_to_q31(((q63_t) *pSrcCmplx++ * in) >> 31);
192 #endif
193
194 /* Decrement loop counter */
195 blkCnt--;
196 }
197
198 }
199 #endif /* defined(ARM_MATH_MVEI) */
200
201 /**
202 @} end of CmplxByRealMult group
203 */
204
