• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /* ----------------------------------------------------------------------
2  * Project:      CMSIS DSP Library
3  * Title:        arm_offset_q15.c
4  * Description:  Q15 vector offset
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 BasicOffset
37   @{
38  */
39 
40 /**
41   @brief         Adds a constant offset to a Q15 vector.
42   @param[in]     pSrc       points to the input vector
43   @param[in]     offset     is the offset to be added
44   @param[out]    pDst       points to the output vector
45   @param[in]     blockSize  number of samples in each vector
46   @return        none
47 
48   @par           Scaling and Overflow Behavior
49                    The function uses saturating arithmetic.
50                    Results outside of the allowable Q15 range [0x8000 0x7FFF] are saturated.
51  */
52 #if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
53 
54 #include "arm_helium_utils.h"
55 
arm_offset_q15(const q15_t * pSrc,q15_t offset,q15_t * pDst,uint32_t blockSize)56 void arm_offset_q15(
57     const q15_t * pSrc,
58     q15_t   offset,
59     q15_t * pDst,
60     uint32_t blockSize)
61 {
62     uint32_t  blkCnt;           /* loop counters */
63     q15x8_t vecSrc;
64 
65     /* Compute 8 outputs at a time */
66     blkCnt = blockSize >> 3;
67     while (blkCnt > 0U)
68     {
69         /*
70          * C = A + offset
71          * Add offset and then store the result in the destination buffer.
72          */
73         vecSrc = vld1q(pSrc);
74         vst1q(pDst, vqaddq(vecSrc, offset));
75         /*
76          * Decrement the blockSize loop counter
77          */
78         blkCnt--;
79         /*
80          * advance vector source and destination pointers
81          */
82         pSrc += 8;
83         pDst += 8;
84     }
85     /*
86      * tail
87      */
88     blkCnt = blockSize & 7;
89     if (blkCnt > 0U)
90     {
91         mve_pred16_t p0 = vctp16q(blkCnt);
92         vecSrc = vld1q(pSrc);
93         vstrhq_p(pDst, vqaddq(vecSrc, offset), p0);
94     }
95 }
96 
97 
98 #else
arm_offset_q15(const q15_t * pSrc,q15_t offset,q15_t * pDst,uint32_t blockSize)99 void arm_offset_q15(
100   const q15_t * pSrc,
101         q15_t offset,
102         q15_t * pDst,
103         uint32_t blockSize)
104 {
105         uint32_t blkCnt;                               /* Loop counter */
106 
107 #if defined (ARM_MATH_LOOPUNROLL)
108 
109 #if defined (ARM_MATH_DSP)
110   q31_t offset_packed;                           /* Offset packed to 32 bit */
111 
112   /* Offset is packed to 32 bit in order to use SIMD32 for addition */
113   offset_packed = __PKHBT(offset, offset, 16);
114 #endif
115 
116   /* Loop unrolling: Compute 4 outputs at a time */
117   blkCnt = blockSize >> 2U;
118 
119   while (blkCnt > 0U)
120   {
121     /* C = A + offset */
122 
123 #if defined (ARM_MATH_DSP)
124     /* Add offset and store result in destination buffer (2 samples at a time). */
125     write_q15x2_ia (&pDst, __QADD16(read_q15x2_ia ((q15_t **) &pSrc), offset_packed));
126     write_q15x2_ia (&pDst, __QADD16(read_q15x2_ia ((q15_t **) &pSrc), offset_packed));
127 #else
128     *pDst++ = (q15_t) __SSAT(((q31_t) *pSrc++ + offset), 16);
129     *pDst++ = (q15_t) __SSAT(((q31_t) *pSrc++ + offset), 16);
130     *pDst++ = (q15_t) __SSAT(((q31_t) *pSrc++ + offset), 16);
131     *pDst++ = (q15_t) __SSAT(((q31_t) *pSrc++ + offset), 16);
132 #endif
133 
134     /* Decrement loop counter */
135     blkCnt--;
136   }
137 
138   /* Loop unrolling: Compute remaining outputs */
139   blkCnt = blockSize % 0x4U;
140 
141 #else
142 
143   /* Initialize blkCnt with number of samples */
144   blkCnt = blockSize;
145 
146 #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
147 
148   while (blkCnt > 0U)
149   {
150     /* C = A + offset */
151 
152     /* Add offset and store result in destination buffer. */
153 #if defined (ARM_MATH_DSP)
154     *pDst++ = (q15_t) __QADD16(*pSrc++, offset);
155 #else
156     *pDst++ = (q15_t) __SSAT(((q31_t) *pSrc++ + offset), 16);
157 #endif
158 
159     /* Decrement loop counter */
160     blkCnt--;
161   }
162 
163 }
164 #endif /* defined(ARM_MATH_MVEI) */
165 
166 /**
167   @} end of BasicOffset group
168  */
169