Renesas / mbed-dsp-neon

Fork of mbed-dsp. CMSIS-DSP library of supporting NEON

Dependents:   mbed-os-example-cmsis_dsp_neon

Fork of mbed-dsp by mbed official

Information

Japanese version is available in lower part of this page.
このページの後半に日本語版が用意されています．

CMSIS-DSP of supporting NEON

What is this ?

A library for CMSIS-DSP of supporting NEON.
We supported the NEON to CMSIS-DSP Ver1.4.3(CMSIS V4.1) that ARM supplied, has achieved the processing speed improvement.
If you use the mbed-dsp library, you can use to replace this library.
CMSIS-DSP of supporting NEON is provied as a library.

Library Creation environment

CMSIS-DSP library of supporting NEON was created by the following environment.

  • Compiler
    ARMCC Version 5.03
  • Compile option switch[C Compiler]
   -DARM_MATH_MATRIX_CHECK -DARM_MATH_ROUNDING -O3 -Otime --cpu=Cortex-A9 --littleend --arm 
   --apcs=/interwork --no_unaligned_access --fpu=vfpv3_fp16 --fpmode=fast --apcs=/hardfp 
   --vectorize --asm
  • Compile option switch[Assembler]
   --cpreproc --cpu=Cortex-A9 --littleend --arm --apcs=/interwork --no_unaligned_access 
   --fpu=vfpv3_fp16 --fpmode=fast --apcs=/hardfp


Effects of NEON support

In the data which passes to each function, large size will be expected more effective than small size.
Also if the data is a multiple of 16, effect will be expected in every function in the CMSIS-DSP.


NEON対応CMSIS-DSP

概要

NEON対応したCMSIS-DSPのライブラリです。
ARM社提供のCMSIS-DSP Ver1.4.3(CMSIS V4.1)をターゲットにNEON対応を行ない、処理速度向上を実現しております。
mbed-dspライブラリを使用している場合は、本ライブラリに置き換えて使用することができます。
NEON対応したCMSIS-DSPはライブラリで提供します。

ライブラリ作成環境

NEON対応CMSIS-DSPライブラリは、以下の環境で作成しています。

  • コンパイラ
    ARMCC Version 5.03
  • コンパイルオプションスイッチ[C Compiler]
   -DARM_MATH_MATRIX_CHECK -DARM_MATH_ROUNDING -O3 -Otime --cpu=Cortex-A9 --littleend --arm 
   --apcs=/interwork --no_unaligned_access --fpu=vfpv3_fp16 --fpmode=fast --apcs=/hardfp 
   --vectorize --asm
  • コンパイルオプションスイッチ[Assembler]
   --cpreproc --cpu=Cortex-A9 --littleend --arm --apcs=/interwork --no_unaligned_access 
   --fpu=vfpv3_fp16 --fpmode=fast --apcs=/hardfp


NEON対応による効果について

CMSIS-DSP内の各関数へ渡すデータは、小さいサイズよりも大きいサイズの方が効果が見込めます。
また、16の倍数のデータであれば、CMSIS-DSP内のどの関数でも効果が見込めます。


cmsis_dsp/StatisticsFunctions/arm_var_f32.c@1:fdd22bb7aa52, 2012-11-28 (annotated)

Committer:
emilmont
Date:
Wed Nov 28 12:30:09 2012 +0000
Revision:
1:fdd22bb7aa52
Child:
2:da51fb522205
DSP library code

Who changed what in which revision?

UserRevisionLine numberNew contents of line
emilmont 1:fdd22bb7aa52 1 /* ----------------------------------------------------------------------
emilmont 1:fdd22bb7aa52 2 * Copyright (C) 2010 ARM Limited. All rights reserved.
emilmont 1:fdd22bb7aa52 3 *
emilmont 1:fdd22bb7aa52 4 * $Date: 15. February 2012
emilmont 1:fdd22bb7aa52 5 * $Revision: V1.1.0
emilmont 1:fdd22bb7aa52 6 *
emilmont 1:fdd22bb7aa52 7 * Project: CMSIS DSP Library
emilmont 1:fdd22bb7aa52 8 * Title: arm_var_f32.c
emilmont 1:fdd22bb7aa52 9 *
emilmont 1:fdd22bb7aa52 10 * Description: Variance of the elements of a floating-point vector.
emilmont 1:fdd22bb7aa52 11 *
emilmont 1:fdd22bb7aa52 12 * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
emilmont 1:fdd22bb7aa52 13 *
emilmont 1:fdd22bb7aa52 14 * Version 1.1.0 2012/02/15
emilmont 1:fdd22bb7aa52 15 * Updated with more optimizations, bug fixes and minor API changes.
emilmont 1:fdd22bb7aa52 16 *
emilmont 1:fdd22bb7aa52 17 * Version 1.0.10 2011/7/15
emilmont 1:fdd22bb7aa52 18 * Big Endian support added and Merged M0 and M3/M4 Source code.
emilmont 1:fdd22bb7aa52 19 *
emilmont 1:fdd22bb7aa52 20 * Version 1.0.3 2010/11/29
emilmont 1:fdd22bb7aa52 21 * Re-organized the CMSIS folders and updated documentation.
emilmont 1:fdd22bb7aa52 22 *
emilmont 1:fdd22bb7aa52 23 * Version 1.0.2 2010/11/11
emilmont 1:fdd22bb7aa52 24 * Documentation updated.
emilmont 1:fdd22bb7aa52 25 *
emilmont 1:fdd22bb7aa52 26 * Version 1.0.1 2010/10/05
emilmont 1:fdd22bb7aa52 27 * Production release and review comments incorporated.
emilmont 1:fdd22bb7aa52 28 *
emilmont 1:fdd22bb7aa52 29 * Version 1.0.0 2010/09/20
emilmont 1:fdd22bb7aa52 30 * Production release and review comments incorporated.
emilmont 1:fdd22bb7aa52 31 * ---------------------------------------------------------------------------- */
emilmont 1:fdd22bb7aa52 32
emilmont 1:fdd22bb7aa52 33 #include "arm_math.h"
emilmont 1:fdd22bb7aa52 34
emilmont 1:fdd22bb7aa52 35 /**
emilmont 1:fdd22bb7aa52 36 * @ingroup groupStats
emilmont 1:fdd22bb7aa52 37 */
emilmont 1:fdd22bb7aa52 38
emilmont 1:fdd22bb7aa52 39 /**
emilmont 1:fdd22bb7aa52 40 * @defgroup variance Variance
emilmont 1:fdd22bb7aa52 41 *
emilmont 1:fdd22bb7aa52 42 * Calculates the variance of the elements in the input vector.
emilmont 1:fdd22bb7aa52 43 * The underlying algorithm is used:
emilmont 1:fdd22bb7aa52 44 *
emilmont 1:fdd22bb7aa52 45 * <pre>
emilmont 1:fdd22bb7aa52 46 * Result = (sumOfSquares - sum<sup>2</sup> / blockSize) / (blockSize - 1)
emilmont 1:fdd22bb7aa52 47 *
emilmont 1:fdd22bb7aa52 48 * where, sumOfSquares = pSrc[0] * pSrc[0] + pSrc[1] * pSrc[1] + ... + pSrc[blockSize-1] * pSrc[blockSize-1]
emilmont 1:fdd22bb7aa52 49 *
emilmont 1:fdd22bb7aa52 50 * sum = pSrc[0] + pSrc[1] + pSrc[2] + ... + pSrc[blockSize-1]
emilmont 1:fdd22bb7aa52 51 * </pre>
emilmont 1:fdd22bb7aa52 52 *
emilmont 1:fdd22bb7aa52 53 * There are separate functions for floating point, Q31, and Q15 data types.
emilmont 1:fdd22bb7aa52 54 */
emilmont 1:fdd22bb7aa52 55
emilmont 1:fdd22bb7aa52 56 /**
emilmont 1:fdd22bb7aa52 57 * @addtogroup variance
emilmont 1:fdd22bb7aa52 58 * @{
emilmont 1:fdd22bb7aa52 59 */
emilmont 1:fdd22bb7aa52 60
emilmont 1:fdd22bb7aa52 61
emilmont 1:fdd22bb7aa52 62 /**
emilmont 1:fdd22bb7aa52 63 * @brief Variance of the elements of a floating-point vector.
emilmont 1:fdd22bb7aa52 64 * @param[in] *pSrc points to the input vector
emilmont 1:fdd22bb7aa52 65 * @param[in] blockSize length of the input vector
emilmont 1:fdd22bb7aa52 66 * @param[out] *pResult variance value returned here
emilmont 1:fdd22bb7aa52 67 * @return none.
emilmont 1:fdd22bb7aa52 68 *
emilmont 1:fdd22bb7aa52 69 */
emilmont 1:fdd22bb7aa52 70
emilmont 1:fdd22bb7aa52 71
emilmont 1:fdd22bb7aa52 72 void arm_var_f32(
emilmont 1:fdd22bb7aa52 73 float32_t * pSrc,
emilmont 1:fdd22bb7aa52 74 uint32_t blockSize,
emilmont 1:fdd22bb7aa52 75 float32_t * pResult)
emilmont 1:fdd22bb7aa52 76 {
emilmont 1:fdd22bb7aa52 77
emilmont 1:fdd22bb7aa52 78 float32_t sum = 0.0f; /* Temporary result storage */
emilmont 1:fdd22bb7aa52 79 float32_t sumOfSquares = 0.0f; /* Sum of squares */
emilmont 1:fdd22bb7aa52 80 float32_t in; /* input value */
emilmont 1:fdd22bb7aa52 81 uint32_t blkCnt; /* loop counter */
emilmont 1:fdd22bb7aa52 82
emilmont 1:fdd22bb7aa52 83 #ifndef ARM_MATH_CM0
emilmont 1:fdd22bb7aa52 84
emilmont 1:fdd22bb7aa52 85 /* Run the below code for Cortex-M4 and Cortex-M3 */
emilmont 1:fdd22bb7aa52 86
emilmont 1:fdd22bb7aa52 87 float32_t meanOfSquares, mean, squareOfMean; /* Temporary variables */
emilmont 1:fdd22bb7aa52 88
emilmont 1:fdd22bb7aa52 89 /*loop Unrolling */
emilmont 1:fdd22bb7aa52 90 blkCnt = blockSize >> 2u;
emilmont 1:fdd22bb7aa52 91
emilmont 1:fdd22bb7aa52 92 /* First part of the processing with loop unrolling. Compute 4 outputs at a time.
emilmont 1:fdd22bb7aa52 93 ** a second loop below computes the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 94 while(blkCnt > 0u)
emilmont 1:fdd22bb7aa52 95 {
emilmont 1:fdd22bb7aa52 96 /* C = (A[0] * A[0] + A[1] * A[1] + ... + A[blockSize-1] * A[blockSize-1]) */
emilmont 1:fdd22bb7aa52 97 /* Compute Sum of squares of the input samples
emilmont 1:fdd22bb7aa52 98 * and then store the result in a temporary variable, sum. */
emilmont 1:fdd22bb7aa52 99 in = *pSrc++;
emilmont 1:fdd22bb7aa52 100 sum += in;
emilmont 1:fdd22bb7aa52 101 sumOfSquares += in * in;
emilmont 1:fdd22bb7aa52 102 in = *pSrc++;
emilmont 1:fdd22bb7aa52 103 sum += in;
emilmont 1:fdd22bb7aa52 104 sumOfSquares += in * in;
emilmont 1:fdd22bb7aa52 105 in = *pSrc++;
emilmont 1:fdd22bb7aa52 106 sum += in;
emilmont 1:fdd22bb7aa52 107 sumOfSquares += in * in;
emilmont 1:fdd22bb7aa52 108 in = *pSrc++;
emilmont 1:fdd22bb7aa52 109 sum += in;
emilmont 1:fdd22bb7aa52 110 sumOfSquares += in * in;
emilmont 1:fdd22bb7aa52 111
emilmont 1:fdd22bb7aa52 112 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 113 blkCnt--;
emilmont 1:fdd22bb7aa52 114 }
emilmont 1:fdd22bb7aa52 115
emilmont 1:fdd22bb7aa52 116 /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
emilmont 1:fdd22bb7aa52 117 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 118 blkCnt = blockSize % 0x4u;
emilmont 1:fdd22bb7aa52 119
emilmont 1:fdd22bb7aa52 120 while(blkCnt > 0u)
emilmont 1:fdd22bb7aa52 121 {
emilmont 1:fdd22bb7aa52 122 /* C = (A[0] * A[0] + A[1] * A[1] + ... + A[blockSize-1] * A[blockSize-1]) */
emilmont 1:fdd22bb7aa52 123 /* Compute Sum of squares of the input samples
emilmont 1:fdd22bb7aa52 124 * and then store the result in a temporary variable, sum. */
emilmont 1:fdd22bb7aa52 125 in = *pSrc++;
emilmont 1:fdd22bb7aa52 126 sum += in;
emilmont 1:fdd22bb7aa52 127 sumOfSquares += in * in;
emilmont 1:fdd22bb7aa52 128
emilmont 1:fdd22bb7aa52 129 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 130 blkCnt--;
emilmont 1:fdd22bb7aa52 131 }
emilmont 1:fdd22bb7aa52 132
emilmont 1:fdd22bb7aa52 133 /* Compute Mean of squares of the input samples
emilmont 1:fdd22bb7aa52 134 * and then store the result in a temporary variable, meanOfSquares. */
emilmont 1:fdd22bb7aa52 135 meanOfSquares = sumOfSquares / ((float32_t) blockSize - 1.0f);
emilmont 1:fdd22bb7aa52 136
emilmont 1:fdd22bb7aa52 137 /* Compute mean of all input values */
emilmont 1:fdd22bb7aa52 138 mean = sum / (float32_t) blockSize;
emilmont 1:fdd22bb7aa52 139
emilmont 1:fdd22bb7aa52 140 /* Compute square of mean */
emilmont 1:fdd22bb7aa52 141 squareOfMean = (mean * mean) * (((float32_t) blockSize) /
emilmont 1:fdd22bb7aa52 142 ((float32_t) blockSize - 1.0f));
emilmont 1:fdd22bb7aa52 143
emilmont 1:fdd22bb7aa52 144 /* Compute variance and then store the result to the destination */
emilmont 1:fdd22bb7aa52 145 *pResult = meanOfSquares - squareOfMean;
emilmont 1:fdd22bb7aa52 146
emilmont 1:fdd22bb7aa52 147 #else
emilmont 1:fdd22bb7aa52 148
emilmont 1:fdd22bb7aa52 149 /* Run the below code for Cortex-M0 */
emilmont 1:fdd22bb7aa52 150 float32_t squareOfSum; /* Square of Sum */
emilmont 1:fdd22bb7aa52 151
emilmont 1:fdd22bb7aa52 152 /* Loop over blockSize number of values */
emilmont 1:fdd22bb7aa52 153 blkCnt = blockSize;
emilmont 1:fdd22bb7aa52 154
emilmont 1:fdd22bb7aa52 155 while(blkCnt > 0u)
emilmont 1:fdd22bb7aa52 156 {
emilmont 1:fdd22bb7aa52 157 /* C = (A[0] * A[0] + A[1] * A[1] + ... + A[blockSize-1] * A[blockSize-1]) */
emilmont 1:fdd22bb7aa52 158 /* Compute Sum of squares of the input samples
emilmont 1:fdd22bb7aa52 159 * and then store the result in a temporary variable, sumOfSquares. */
emilmont 1:fdd22bb7aa52 160 in = *pSrc++;
emilmont 1:fdd22bb7aa52 161 sumOfSquares += in * in;
emilmont 1:fdd22bb7aa52 162
emilmont 1:fdd22bb7aa52 163 /* C = (A[0] + A[1] + ... + A[blockSize-1]) */
emilmont 1:fdd22bb7aa52 164 /* Compute Sum of the input samples
emilmont 1:fdd22bb7aa52 165 * and then store the result in a temporary variable, sum. */
emilmont 1:fdd22bb7aa52 166 sum += in;
emilmont 1:fdd22bb7aa52 167
emilmont 1:fdd22bb7aa52 168 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 169 blkCnt--;
emilmont 1:fdd22bb7aa52 170 }
emilmont 1:fdd22bb7aa52 171
emilmont 1:fdd22bb7aa52 172 /* Compute the square of sum */
emilmont 1:fdd22bb7aa52 173 squareOfSum = ((sum * sum) / (float32_t) blockSize);
emilmont 1:fdd22bb7aa52 174
emilmont 1:fdd22bb7aa52 175 /* Compute the variance */
emilmont 1:fdd22bb7aa52 176 *pResult = ((sumOfSquares - squareOfSum) / (float32_t) (blockSize - 1.0f));
emilmont 1:fdd22bb7aa52 177
emilmont 1:fdd22bb7aa52 178 #endif /* #ifndef ARM_MATH_CM0 */
emilmont 1:fdd22bb7aa52 179
emilmont 1:fdd22bb7aa52 180 }
emilmont 1:fdd22bb7aa52 181
emilmont 1:fdd22bb7aa52 182 /**
emilmont 1:fdd22bb7aa52 183 * @} end of variance group
emilmont 1:fdd22bb7aa52 184 */