Robert Lopez / CMSIS5

Aded CMSIS5 DSP and NN folder. Needs some work

DSP/source/StatisticsFunctions/arm_power_f32.c@0:eedb7d567a5d, 2018-04-12 (annotated)

Committer:
robert_lp
Date:
Thu Apr 12 01:31:58 2018 +0000
Revision:
0:eedb7d567a5d
CMSIS5 Library

Who changed what in which revision?

UserRevisionLine numberNew contents of line
robert_lp 0:eedb7d567a5d 1 /* ----------------------------------------------------------------------
robert_lp 0:eedb7d567a5d 2 * Project: CMSIS DSP Library
robert_lp 0:eedb7d567a5d 3 * Title: arm_power_f32.c
robert_lp 0:eedb7d567a5d 4 * Description: Sum of the squares of the elements of a floating-point vector
robert_lp 0:eedb7d567a5d 5 *
robert_lp 0:eedb7d567a5d 6 * $Date: 27. January 2017
robert_lp 0:eedb7d567a5d 7 * $Revision: V.1.5.1
robert_lp 0:eedb7d567a5d 8 *
robert_lp 0:eedb7d567a5d 9 * Target Processor: Cortex-M cores
robert_lp 0:eedb7d567a5d 10 * -------------------------------------------------------------------- */
robert_lp 0:eedb7d567a5d 11 /*
robert_lp 0:eedb7d567a5d 12 * Copyright (C) 2010-2017 ARM Limited or its affiliates. All rights reserved.
robert_lp 0:eedb7d567a5d 13 *
robert_lp 0:eedb7d567a5d 14 * SPDX-License-Identifier: Apache-2.0
robert_lp 0:eedb7d567a5d 15 *
robert_lp 0:eedb7d567a5d 16 * Licensed under the Apache License, Version 2.0 (the License); you may
robert_lp 0:eedb7d567a5d 17 * not use this file except in compliance with the License.
robert_lp 0:eedb7d567a5d 18 * You may obtain a copy of the License at
robert_lp 0:eedb7d567a5d 19 *
robert_lp 0:eedb7d567a5d 20 * www.apache.org/licenses/LICENSE-2.0
robert_lp 0:eedb7d567a5d 21 *
robert_lp 0:eedb7d567a5d 22 * Unless required by applicable law or agreed to in writing, software
robert_lp 0:eedb7d567a5d 23 * distributed under the License is distributed on an AS IS BASIS, WITHOUT
robert_lp 0:eedb7d567a5d 24 * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
robert_lp 0:eedb7d567a5d 25 * See the License for the specific language governing permissions and
robert_lp 0:eedb7d567a5d 26 * limitations under the License.
robert_lp 0:eedb7d567a5d 27 */
robert_lp 0:eedb7d567a5d 28
robert_lp 0:eedb7d567a5d 29 #include "arm_math.h"
robert_lp 0:eedb7d567a5d 30
robert_lp 0:eedb7d567a5d 31 /**
robert_lp 0:eedb7d567a5d 32 * @ingroup groupStats
robert_lp 0:eedb7d567a5d 33 */
robert_lp 0:eedb7d567a5d 34
robert_lp 0:eedb7d567a5d 35 /**
robert_lp 0:eedb7d567a5d 36 * @defgroup power Power
robert_lp 0:eedb7d567a5d 37 *
robert_lp 0:eedb7d567a5d 38 * Calculates the sum of the squares of the elements in the input vector.
robert_lp 0:eedb7d567a5d 39 * The underlying algorithm is used:
robert_lp 0:eedb7d567a5d 40 *
robert_lp 0:eedb7d567a5d 41 * <pre>
robert_lp 0:eedb7d567a5d 42 * Result = pSrc[0] * pSrc[0] + pSrc[1] * pSrc[1] + pSrc[2] * pSrc[2] + ... + pSrc[blockSize-1] * pSrc[blockSize-1];
robert_lp 0:eedb7d567a5d 43 * </pre>
robert_lp 0:eedb7d567a5d 44 *
robert_lp 0:eedb7d567a5d 45 * There are separate functions for floating point, Q31, Q15, and Q7 data types.
robert_lp 0:eedb7d567a5d 46 */
robert_lp 0:eedb7d567a5d 47
robert_lp 0:eedb7d567a5d 48 /**
robert_lp 0:eedb7d567a5d 49 * @addtogroup power
robert_lp 0:eedb7d567a5d 50 * @{
robert_lp 0:eedb7d567a5d 51 */
robert_lp 0:eedb7d567a5d 52
robert_lp 0:eedb7d567a5d 53
robert_lp 0:eedb7d567a5d 54 /**
robert_lp 0:eedb7d567a5d 55 * @brief Sum of the squares of the elements of a floating-point vector.
robert_lp 0:eedb7d567a5d 56 * @param[in] *pSrc points to the input vector
robert_lp 0:eedb7d567a5d 57 * @param[in] blockSize length of the input vector
robert_lp 0:eedb7d567a5d 58 * @param[out] *pResult sum of the squares value returned here
robert_lp 0:eedb7d567a5d 59 * @return none.
robert_lp 0:eedb7d567a5d 60 *
robert_lp 0:eedb7d567a5d 61 */
robert_lp 0:eedb7d567a5d 62
robert_lp 0:eedb7d567a5d 63
robert_lp 0:eedb7d567a5d 64 void arm_power_f32(
robert_lp 0:eedb7d567a5d 65 float32_t * pSrc,
robert_lp 0:eedb7d567a5d 66 uint32_t blockSize,
robert_lp 0:eedb7d567a5d 67 float32_t * pResult)
robert_lp 0:eedb7d567a5d 68 {
robert_lp 0:eedb7d567a5d 69 float32_t sum = 0.0f; /* accumulator */
robert_lp 0:eedb7d567a5d 70 float32_t in; /* Temporary variable to store input value */
robert_lp 0:eedb7d567a5d 71 uint32_t blkCnt; /* loop counter */
robert_lp 0:eedb7d567a5d 72
robert_lp 0:eedb7d567a5d 73 #if defined (ARM_MATH_DSP)
robert_lp 0:eedb7d567a5d 74 /* Run the below code for Cortex-M4 and Cortex-M3 */
robert_lp 0:eedb7d567a5d 75
robert_lp 0:eedb7d567a5d 76 /*loop Unrolling */
robert_lp 0:eedb7d567a5d 77 blkCnt = blockSize >> 2U;
robert_lp 0:eedb7d567a5d 78
robert_lp 0:eedb7d567a5d 79 /* First part of the processing with loop unrolling. Compute 4 outputs at a time.
robert_lp 0:eedb7d567a5d 80 ** a second loop below computes the remaining 1 to 3 samples. */
robert_lp 0:eedb7d567a5d 81 while (blkCnt > 0U)
robert_lp 0:eedb7d567a5d 82 {
robert_lp 0:eedb7d567a5d 83 /* C = A[0] * A[0] + A[1] * A[1] + A[2] * A[2] + ... + A[blockSize-1] * A[blockSize-1] */
robert_lp 0:eedb7d567a5d 84 /* Compute Power and then store the result in a temporary variable, sum. */
robert_lp 0:eedb7d567a5d 85 in = *pSrc++;
robert_lp 0:eedb7d567a5d 86 sum += in * in;
robert_lp 0:eedb7d567a5d 87 in = *pSrc++;
robert_lp 0:eedb7d567a5d 88 sum += in * in;
robert_lp 0:eedb7d567a5d 89 in = *pSrc++;
robert_lp 0:eedb7d567a5d 90 sum += in * in;
robert_lp 0:eedb7d567a5d 91 in = *pSrc++;
robert_lp 0:eedb7d567a5d 92 sum += in * in;
robert_lp 0:eedb7d567a5d 93
robert_lp 0:eedb7d567a5d 94 /* Decrement the loop counter */
robert_lp 0:eedb7d567a5d 95 blkCnt--;
robert_lp 0:eedb7d567a5d 96 }
robert_lp 0:eedb7d567a5d 97
robert_lp 0:eedb7d567a5d 98 /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
robert_lp 0:eedb7d567a5d 99 ** No loop unrolling is used. */
robert_lp 0:eedb7d567a5d 100 blkCnt = blockSize % 0x4U;
robert_lp 0:eedb7d567a5d 101
robert_lp 0:eedb7d567a5d 102
robert_lp 0:eedb7d567a5d 103 #else
robert_lp 0:eedb7d567a5d 104 /* Run the below code for Cortex-M0 */
robert_lp 0:eedb7d567a5d 105
robert_lp 0:eedb7d567a5d 106 /* Loop over blockSize number of values */
robert_lp 0:eedb7d567a5d 107 blkCnt = blockSize;
robert_lp 0:eedb7d567a5d 108
robert_lp 0:eedb7d567a5d 109 #endif /* #if defined (ARM_MATH_DSP) */
robert_lp 0:eedb7d567a5d 110
robert_lp 0:eedb7d567a5d 111
robert_lp 0:eedb7d567a5d 112 while (blkCnt > 0U)
robert_lp 0:eedb7d567a5d 113 {
robert_lp 0:eedb7d567a5d 114 /* C = A[0] * A[0] + A[1] * A[1] + A[2] * A[2] + ... + A[blockSize-1] * A[blockSize-1] */
robert_lp 0:eedb7d567a5d 115 /* compute power and then store the result in a temporary variable, sum. */
robert_lp 0:eedb7d567a5d 116 in = *pSrc++;
robert_lp 0:eedb7d567a5d 117 sum += in * in;
robert_lp 0:eedb7d567a5d 118
robert_lp 0:eedb7d567a5d 119 /* Decrement the loop counter */
robert_lp 0:eedb7d567a5d 120 blkCnt--;
robert_lp 0:eedb7d567a5d 121 }
robert_lp 0:eedb7d567a5d 122
robert_lp 0:eedb7d567a5d 123 /* Store the result to the destination */
robert_lp 0:eedb7d567a5d 124 *pResult = sum;
robert_lp 0:eedb7d567a5d 125 }
robert_lp 0:eedb7d567a5d 126
robert_lp 0:eedb7d567a5d 127 /**
robert_lp 0:eedb7d567a5d 128 * @} end of power group
robert_lp 0:eedb7d567a5d 129 */
robert_lp 0:eedb7d567a5d 130