1 /* ----------------------------------------------------------------------
2 * Project: CMSIS DSP Library
3 * Title: arm_mse_f64.c
4 * Description: Double floating point mean square error
5 *
6 * $Date: 10 August 2022
7 * $Revision: V1.10.1
8 *
9 * Target Processor: Cortex-M and Cortex-A cores
10 * -------------------------------------------------------------------- */
11 /*
12 * Copyright (C) 2010-2022 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/statistics_functions.h"
30
31 /**
32 @ingroup groupStats
33 */
34
35 /**
36 @addtogroup MSE
37 @{
38 */
39
40 /**
41 @brief Mean square error between two double floating point vectors.
42 @param[in] pSrcA points to the first input vector
43 @param[in] pSrcB points to the second input vector
44 @param[in] blockSize number of samples in input vector
45 @param[out] pResult mean square error
46 */
47
arm_mse_f64(const float64_t * pSrcA,const float64_t * pSrcB,uint32_t blockSize,float64_t * pResult)48 ARM_DSP_ATTRIBUTE void arm_mse_f64(
49 const float64_t * pSrcA,
50 const float64_t * pSrcB,
51 uint32_t blockSize,
52 float64_t * pResult)
53
54 {
55
56 uint32_t blkCnt; /* Loop counter */
57 float64_t inA, inB;
58 float64_t sum = 0.0;
59 #if defined(ARM_MATH_NEON) && defined(__aarch64__)
60
61 float64x2_t inAV , inBV , subV, sumV;
62 sumV = vdupq_n_f64(0.0);
63
64 blkCnt = blockSize >> 1U ;
65
66 while (blkCnt > 0U)
67 {
68 inAV = vld1q_f64(pSrcA);
69 pSrcA+=2;
70 inBV = vld1q_f64(pSrcB);
71 pSrcB+=2;
72 subV = vsubq_f64(inAV, inBV);
73 sumV = vmlaq_f64(sumV, subV, subV);
74
75 blkCnt--;
76
77 }
78 sum = vaddvq_f64(sumV);
79 blkCnt = (blockSize) & 1;
80
81 #else
82 /* Temporary return variable */
83 #if defined (ARM_MATH_LOOPUNROLL)
84 blkCnt = (blockSize) >> 1;
85
86 while (blkCnt > 0U)
87 {
88
89
90 inA = *pSrcA++;
91 inB = *pSrcB++;
92 inA = inA - inB;
93 sum += inA * inA;
94
95 inA = *pSrcA++;
96 inB = *pSrcB++;
97 inA = inA - inB;
98 sum += inA * inA;
99
100 /* Decrement loop counter */
101 blkCnt--;
102 }
103
104
105 /* Loop unrolling: Compute remaining outputs */
106 blkCnt = (blockSize) & 1;
107 #else
108 /* Initialize blkCnt with number of samples */
109 blkCnt = blockSize;
110 #endif
111 #endif
112
113 #if defined(__clang__) && defined(ARM_MATH_NEON) && defined(__aarch64__)
114 #pragma clang loop vectorize(enable) unroll(disable)
115 #endif
116 while (blkCnt > 0U)
117 {
118 inA = *pSrcA++;
119 inB = *pSrcB++;
120 inA = inA - inB;
121 sum += inA * inA;
122
123 /* Decrement loop counter */
124 blkCnt--;
125 }
126
127 /* Store result in destination buffer */
128 *pResult = sum / blockSize;
129 }
130
131
132
133 /**
134 @} end of MSE group
135 */
136
