1 /* ----------------------------------------------------------------------
2 * Project: CMSIS DSP Library
3 * Title: arm_mult_q7.c
4 * Description: Q7 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 @addtogroup BasicMult
37 @{
38 */
39
40 /**
41 @brief Q7 vector multiplication
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 Q7 range [0x80 0x7F] are saturated.
50 */
51 #if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
52
53 #include "arm_helium_utils.h"
54
arm_mult_q7(const q7_t * pSrcA,const q7_t * pSrcB,q7_t * pDst,uint32_t blockSize)55 ARM_DSP_ATTRIBUTE void arm_mult_q7(
56 const q7_t * pSrcA,
57 const q7_t * pSrcB,
58 q7_t * pDst,
59 uint32_t blockSize)
60 {
61 uint32_t blkCnt; /* loop counters */
62 q7x16_t vecA, vecB;
63
64 /* Compute 16 outputs at a time */
65 blkCnt = blockSize >> 4;
66 while (blkCnt > 0U)
67 {
68 /*
69 * C = A * B
70 * Multiply the inputs and then store the results in the destination buffer.
71 */
72 vecA = vld1q(pSrcA);
73 vecB = vld1q(pSrcB);
74 vst1q(pDst, vqdmulhq(vecA, vecB));
75 /*
76 * Decrement the blockSize loop counter
77 */
78 blkCnt--;
79 /*
80 * advance vector source and destination pointers
81 */
82 pSrcA += 16;
83 pSrcB += 16;
84 pDst += 16;
85 }
86 /*
87 * tail
88 */
89 blkCnt = blockSize & 0xF;
90 if (blkCnt > 0U)
91 {
92 mve_pred16_t p0 = vctp8q(blkCnt);
93 vecA = vld1q(pSrcA);
94 vecB = vld1q(pSrcB);
95 vstrbq_p(pDst, vqdmulhq(vecA, vecB), p0);
96 }
97 }
98
99 #else
arm_mult_q7(const q7_t * pSrcA,const q7_t * pSrcB,q7_t * pDst,uint32_t blockSize)100 ARM_DSP_ATTRIBUTE void arm_mult_q7(
101 const q7_t * pSrcA,
102 const q7_t * pSrcB,
103 q7_t * pDst,
104 uint32_t blockSize)
105 {
106 uint32_t blkCnt; /* Loop counter */
107
108 #if defined (ARM_MATH_LOOPUNROLL)
109
110 #if defined (ARM_MATH_DSP)
111 q7_t out1, out2, out3, out4; /* Temporary output variables */
112 #endif
113
114 /* Loop unrolling: Compute 4 outputs at a time */
115 blkCnt = blockSize >> 2U;
116
117 while (blkCnt > 0U)
118 {
119 /* C = A * B */
120
121 #if defined (ARM_MATH_DSP)
122 /* Multiply inputs and store results in temporary variables */
123 out1 = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
124 out2 = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
125 out3 = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
126 out4 = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
127
128 /* Pack and store result in destination buffer (in single write) */
129 write_q7x4_ia (&pDst, __PACKq7(out1, out2, out3, out4));
130 #else
131 *pDst++ = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
132 *pDst++ = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
133 *pDst++ = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
134 *pDst++ = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
135 #endif
136
137 /* Decrement loop counter */
138 blkCnt--;
139 }
140
141 /* Loop unrolling: Compute remaining outputs */
142 blkCnt = blockSize % 0x4U;
143
144 #else
145
146 /* Initialize blkCnt with number of samples */
147 blkCnt = blockSize;
148
149 #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
150
151 while (blkCnt > 0U)
152 {
153 /* C = A * B */
154
155 /* Multiply input and store result in destination buffer. */
156 *pDst++ = (q7_t) __SSAT((((q15_t) (*pSrcA++) * (*pSrcB++)) >> 7), 8);
157
158 /* Decrement loop counter */
159 blkCnt--;
160 }
161
162 }
163 #endif /* defined(ARM_MATH_MVEI) */
164
165 /**
166 @} end of BasicMult group
167 */
168
