1 /* ----------------------------------------------------------------------
2 * Project: CMSIS DSP Library
3 * Title: arm_add_q31.c
4 * Description: Q31 vector addition
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 @addtogroup BasicAdd
37 @{
38 */
39
40 /**
41 @brief Q31 vector addition.
42 @param[in] pSrcA points to the first input vector
43 @param[in] pSrcB points to the second input vector
44 @param[out] pDst points to the output vector
45 @param[in] blockSize 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)
53
54 #include "arm_helium_utils.h"
55
arm_add_q31(const q31_t * pSrcA,const q31_t * pSrcB,q31_t * pDst,uint32_t blockSize)56 void arm_add_q31(
57 const q31_t * pSrcA,
58 const q31_t * pSrcB,
59 q31_t * pDst,
60 uint32_t blockSize)
61 {
62 uint32_t blkCnt;
63 q31x4_t vecA;
64 q31x4_t vecB;
65
66 /* Compute 4 outputs at a time */
67 blkCnt = blockSize >> 2;
68 while (blkCnt > 0U)
69 {
70 /*
71 * C = A + B
72 * Add and then store the results in the destination buffer.
73 */
74 vecA = vld1q(pSrcA);
75 vecB = vld1q(pSrcB);
76 vst1q(pDst, vqaddq(vecA, vecB));
77 /*
78 * Decrement the blockSize loop counter
79 */
80 blkCnt--;
81 /*
82 * advance vector source and destination pointers
83 */
84 pSrcA += 4;
85 pSrcB += 4;
86 pDst += 4;
87 }
88 /*
89 * tail
90 */
91 blkCnt = blockSize & 3;
92 if (blkCnt > 0U)
93 {
94 mve_pred16_t p0 = vctp32q(blkCnt);
95 vecA = vld1q(pSrcA);
96 vecB = vld1q(pSrcB);
97 vstrwq_p(pDst, vqaddq(vecA, vecB), p0);
98 }
99 }
100
101 #else
arm_add_q31(const q31_t * pSrcA,const q31_t * pSrcB,q31_t * pDst,uint32_t blockSize)102 void arm_add_q31(
103 const q31_t * pSrcA,
104 const q31_t * pSrcB,
105 q31_t * pDst,
106 uint32_t blockSize)
107 {
108 uint32_t blkCnt; /* Loop counter */
109
110 #if defined (ARM_MATH_LOOPUNROLL)
111
112 /* Loop unrolling: Compute 4 outputs at a time */
113 blkCnt = blockSize >> 2U;
114
115 while (blkCnt > 0U)
116 {
117 /* C = A + B */
118
119 /* Add and store result in destination buffer. */
120 *pDst++ = __QADD(*pSrcA++, *pSrcB++);
121
122 *pDst++ = __QADD(*pSrcA++, *pSrcB++);
123
124 *pDst++ = __QADD(*pSrcA++, *pSrcB++);
125
126 *pDst++ = __QADD(*pSrcA++, *pSrcB++);
127
128 /* Decrement loop counter */
129 blkCnt--;
130 }
131
132 /* Loop unrolling: Compute remaining outputs */
133 blkCnt = blockSize % 0x4U;
134
135 #else
136
137 /* Initialize blkCnt with number of samples */
138 blkCnt = blockSize;
139
140 #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
141
142 while (blkCnt > 0U)
143 {
144 /* C = A + B */
145
146 /* Add and store result in destination buffer. */
147 *pDst++ = __QADD(*pSrcA++, *pSrcB++);
148
149 /* Decrement loop counter */
150 blkCnt--;
151 }
152
153 }
154
155 #endif /* defined(ARM_MATH_MVEI) */
156 /**
157 @} end of BasicAdd group
158 */
159
