1 /* ----------------------------------------------------------------------
2 * Project: CMSIS DSP Library
3 * Title: arm_mult_f32.c
4 * Description: Floating-point vector 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/basic_math_functions.h"
30
31 /**
32 @ingroup groupMath
33 */
34
35 /**
36 @defgroup BasicMult Vector Multiplication
37
38 Element-by-element multiplication of two vectors.
39
40 <pre>
41 pDst[n] = pSrcA[n] * pSrcB[n], 0 <= n < blockSize.
42 </pre>
43
44 There are separate functions for floating-point, Q7, Q15, and Q31 data types.
45 */
46
47 /**
48 @addtogroup BasicMult
49 @{
50 */
51
52 /**
53 @brief Floating-point vector multiplication.
54 @param[in] pSrcA points to the first input vector.
55 @param[in] pSrcB points to the second input vector.
56 @param[out] pDst points to the output vector.
57 @param[in] blockSize number of samples in each vector.
58 */
59
60 #if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
61
62 #include "arm_helium_utils.h"
63
arm_mult_f32(const float32_t * pSrcA,const float32_t * pSrcB,float32_t * pDst,uint32_t blockSize)64 void arm_mult_f32(
65 const float32_t * pSrcA,
66 const float32_t * pSrcB,
67 float32_t * pDst,
68 uint32_t blockSize)
69 {
70 uint32_t blkCnt; /* Loop counter */
71
72 f32x4_t vec1;
73 f32x4_t vec2;
74 f32x4_t res;
75
76 /* Compute 4 outputs at a time */
77 blkCnt = blockSize >> 2U;
78 while (blkCnt > 0U)
79 {
80 /* C = A + B */
81
82 /* Add and then store the results in the destination buffer. */
83 vec1 = vld1q(pSrcA);
84 vec2 = vld1q(pSrcB);
85 res = vmulq(vec1, vec2);
86 vst1q(pDst, res);
87
88 /* Increment pointers */
89 pSrcA += 4;
90 pSrcB += 4;
91 pDst += 4;
92
93 /* Decrement the loop counter */
94 blkCnt--;
95 }
96
97 /* Tail */
98 blkCnt = blockSize & 0x3;
99 if (blkCnt > 0U)
100 {
101 /* C = A + B */
102 mve_pred16_t p0 = vctp32q(blkCnt);
103 vec1 = vld1q(pSrcA);
104 vec2 = vld1q(pSrcB);
105 vstrwq_p(pDst, vmulq(vec1,vec2), p0);
106 }
107
108 }
109
110 #else
arm_mult_f32(const float32_t * pSrcA,const float32_t * pSrcB,float32_t * pDst,uint32_t blockSize)111 void arm_mult_f32(
112 const float32_t * pSrcA,
113 const float32_t * pSrcB,
114 float32_t * pDst,
115 uint32_t blockSize)
116 {
117 uint32_t blkCnt; /* Loop counter */
118
119 #if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
120 f32x4_t vec1;
121 f32x4_t vec2;
122 f32x4_t res;
123
124 /* Compute 4 outputs at a time */
125 blkCnt = blockSize >> 2U;
126
127 while (blkCnt > 0U)
128 {
129 /* C = A * B */
130
131 /* Multiply the inputs and then store the results in the destination buffer. */
132 vec1 = vld1q_f32(pSrcA);
133 vec2 = vld1q_f32(pSrcB);
134 res = vmulq_f32(vec1, vec2);
135 vst1q_f32(pDst, res);
136
137 /* Increment pointers */
138 pSrcA += 4;
139 pSrcB += 4;
140 pDst += 4;
141
142 /* Decrement the loop counter */
143 blkCnt--;
144 }
145
146 /* Tail */
147 blkCnt = blockSize & 0x3;
148
149 #else
150 #if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
151
152 /* Loop unrolling: Compute 4 outputs at a time */
153 blkCnt = blockSize >> 2U;
154
155 while (blkCnt > 0U)
156 {
157 /* C = A * B */
158
159 /* Multiply inputs and store result in destination buffer. */
160 *pDst++ = (*pSrcA++) * (*pSrcB++);
161
162 *pDst++ = (*pSrcA++) * (*pSrcB++);
163
164 *pDst++ = (*pSrcA++) * (*pSrcB++);
165
166 *pDst++ = (*pSrcA++) * (*pSrcB++);
167
168 /* Decrement loop counter */
169 blkCnt--;
170 }
171
172 /* Loop unrolling: Compute remaining outputs */
173 blkCnt = blockSize % 0x4U;
174
175 #else
176
177 /* Initialize blkCnt with number of samples */
178 blkCnt = blockSize;
179
180 #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
181 #endif /* #if defined(ARM_MATH_NEON) */
182
183 while (blkCnt > 0U)
184 {
185 /* C = A * B */
186
187 /* Multiply input and store result in destination buffer. */
188 *pDst++ = (*pSrcA++) * (*pSrcB++);
189
190 /* Decrement loop counter */
191 blkCnt--;
192 }
193
194 }
195 #endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
196
197 /**
198 @} end of BasicMult group
199 */
200
