|
|
|
|
@ -1,8 +1,8 @@
|
|
|
|
|
/* ----------------------------------------------------------------------
|
|
|
|
|
* Copyright (C) 2010-2014 ARM Limited. All rights reserved.
|
|
|
|
|
*
|
|
|
|
|
* $Date: 19. March 2015
|
|
|
|
|
* $Revision: V.1.4.5
|
|
|
|
|
* $Date: 22. December 2016
|
|
|
|
|
* $Revision: V.1.4.5 a
|
|
|
|
|
*
|
|
|
|
|
* Project: CMSIS DSP Library
|
|
|
|
|
* Title: arm_var_f32.c
|
|
|
|
|
@ -48,14 +48,13 @@
|
|
|
|
|
* @defgroup variance Variance
|
|
|
|
|
*
|
|
|
|
|
* Calculates the variance of the elements in the input vector.
|
|
|
|
|
* The underlying algorithm is used:
|
|
|
|
|
* The underlying algorithm used is the direct method sometimes referred to as the two-pass method:
|
|
|
|
|
*
|
|
|
|
|
* <pre>
|
|
|
|
|
* Result = (sumOfSquares - sum<sup>2</sup> / blockSize) / (blockSize - 1)
|
|
|
|
|
* Result = sum(element - meanOfElements)^2) / numElement - 1
|
|
|
|
|
*
|
|
|
|
|
* where, sumOfSquares = pSrc[0] * pSrc[0] + pSrc[1] * pSrc[1] + ... + pSrc[blockSize-1] * pSrc[blockSize-1]
|
|
|
|
|
* where, meanOfElements = ( pSrc[0] * pSrc[0] + pSrc[1] * pSrc[1] + ... + pSrc[blockSize-1] ) / blockSize
|
|
|
|
|
*
|
|
|
|
|
* sum = pSrc[0] + pSrc[1] + pSrc[2] + ... + pSrc[blockSize-1]
|
|
|
|
|
* </pre>
|
|
|
|
|
*
|
|
|
|
|
* There are separate functions for floating point, Q31, and Q15 data types.
|
|
|
|
|
@ -76,116 +75,117 @@
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
void arm_var_f32(
|
|
|
|
|
float32_t * pSrc,
|
|
|
|
|
uint32_t blockSize,
|
|
|
|
|
float32_t * pResult)
|
|
|
|
|
float32_t * pSrc,
|
|
|
|
|
uint32_t blockSize,
|
|
|
|
|
float32_t * pResult)
|
|
|
|
|
{
|
|
|
|
|
float32_t sum = 0.0f; /* Temporary result storage */
|
|
|
|
|
float32_t sumOfSquares = 0.0f; /* Sum of squares */
|
|
|
|
|
float32_t in; /* input value */
|
|
|
|
|
uint32_t blkCnt; /* loop counter */
|
|
|
|
|
#ifndef ARM_MATH_CM0_FAMILY
|
|
|
|
|
float32_t meanOfSquares, mean, squareOfMean; /* Temporary variables */
|
|
|
|
|
#else
|
|
|
|
|
float32_t squareOfSum; /* Square of Sum */
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
if(blockSize == 1u)
|
|
|
|
|
{
|
|
|
|
|
*pResult = 0;
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#ifndef ARM_MATH_CM0_FAMILY
|
|
|
|
|
|
|
|
|
|
/*loop Unrolling */
|
|
|
|
|
blkCnt = blockSize >> 2u;
|
|
|
|
|
|
|
|
|
|
/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
|
|
|
|
|
** a second loop below computes the remaining 1 to 3 samples. */
|
|
|
|
|
while(blkCnt > 0u)
|
|
|
|
|
{
|
|
|
|
|
/* C = (A[0] * A[0] + A[1] * A[1] + ... + A[blockSize-1] * A[blockSize-1]) */
|
|
|
|
|
/* Compute Sum of squares of the input samples
|
|
|
|
|
* and then store the result in a temporary variable, sum. */
|
|
|
|
|
in = *pSrc++;
|
|
|
|
|
sum += in;
|
|
|
|
|
sumOfSquares += in * in;
|
|
|
|
|
in = *pSrc++;
|
|
|
|
|
sum += in;
|
|
|
|
|
sumOfSquares += in * in;
|
|
|
|
|
in = *pSrc++;
|
|
|
|
|
sum += in;
|
|
|
|
|
sumOfSquares += in * in;
|
|
|
|
|
in = *pSrc++;
|
|
|
|
|
sum += in;
|
|
|
|
|
sumOfSquares += in * in;
|
|
|
|
|
|
|
|
|
|
/* Decrement the loop counter */
|
|
|
|
|
blkCnt--;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* If the blockSize is not a multiple of 4, compute any remaining output samples here.
|
|
|
|
|
** No loop unrolling is used. */
|
|
|
|
|
blkCnt = blockSize % 0x4u;
|
|
|
|
|
|
|
|
|
|
while(blkCnt > 0u)
|
|
|
|
|
{
|
|
|
|
|
/* C = (A[0] * A[0] + A[1] * A[1] + ... + A[blockSize-1] * A[blockSize-1]) */
|
|
|
|
|
/* Compute Sum of squares of the input samples
|
|
|
|
|
* and then store the result in a temporary variable, sum. */
|
|
|
|
|
in = *pSrc++;
|
|
|
|
|
sum += in;
|
|
|
|
|
sumOfSquares += in * in;
|
|
|
|
|
|
|
|
|
|
/* Decrement the loop counter */
|
|
|
|
|
blkCnt--;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Compute Mean of squares of the input samples
|
|
|
|
|
* and then store the result in a temporary variable, meanOfSquares. */
|
|
|
|
|
meanOfSquares = sumOfSquares / ((float32_t) blockSize - 1.0f);
|
|
|
|
|
|
|
|
|
|
/* Compute mean of all input values */
|
|
|
|
|
mean = sum / (float32_t) blockSize;
|
|
|
|
|
|
|
|
|
|
/* Compute square of mean */
|
|
|
|
|
squareOfMean = (mean * mean) * (((float32_t) blockSize) /
|
|
|
|
|
((float32_t) blockSize - 1.0f));
|
|
|
|
|
|
|
|
|
|
/* Compute variance and then store the result to the destination */
|
|
|
|
|
*pResult = meanOfSquares - squareOfMean;
|
|
|
|
|
|
|
|
|
|
#else
|
|
|
|
|
/* Run the below code for Cortex-M0 */
|
|
|
|
|
|
|
|
|
|
/* Loop over blockSize number of values */
|
|
|
|
|
blkCnt = blockSize;
|
|
|
|
|
|
|
|
|
|
while(blkCnt > 0u)
|
|
|
|
|
{
|
|
|
|
|
/* C = (A[0] * A[0] + A[1] * A[1] + ... + A[blockSize-1] * A[blockSize-1]) */
|
|
|
|
|
/* Compute Sum of squares of the input samples
|
|
|
|
|
* and then store the result in a temporary variable, sumOfSquares. */
|
|
|
|
|
in = *pSrc++;
|
|
|
|
|
sumOfSquares += in * in;
|
|
|
|
|
|
|
|
|
|
/* C = (A[0] + A[1] + ... + A[blockSize-1]) */
|
|
|
|
|
/* Compute Sum of the input samples
|
|
|
|
|
* and then store the result in a temporary variable, sum. */
|
|
|
|
|
sum += in;
|
|
|
|
|
|
|
|
|
|
/* Decrement the loop counter */
|
|
|
|
|
blkCnt--;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Compute the square of sum */
|
|
|
|
|
squareOfSum = ((sum * sum) / (float32_t) blockSize);
|
|
|
|
|
|
|
|
|
|
/* Compute the variance */
|
|
|
|
|
*pResult = ((sumOfSquares - squareOfSum) / (float32_t) (blockSize - 1.0f));
|
|
|
|
|
|
|
|
|
|
#endif /* #ifndef ARM_MATH_CM0_FAMILY */
|
|
|
|
|
float32_t fMean, fValue;
|
|
|
|
|
uint32_t blkCnt; /* loop counter */
|
|
|
|
|
float32_t * pInput = pSrc;
|
|
|
|
|
float32_t sum = 0.0f;
|
|
|
|
|
float32_t fSum = 0.0f;
|
|
|
|
|
#if !defined(ARM_MATH_CM0_FAMILY) && !defined(ARM_MATH_CM3_FAMILY)
|
|
|
|
|
float32_t in1, in2, in3, in4;
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
if (blockSize <= 1u)
|
|
|
|
|
{
|
|
|
|
|
*pResult = 0;
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#if !defined(ARM_MATH_CM0_FAMILY) && !defined(ARM_MATH_CM3_FAMILY)
|
|
|
|
|
/* Run the below code for Cortex-M4 and Cortex-M7 */
|
|
|
|
|
|
|
|
|
|
/*loop Unrolling */
|
|
|
|
|
blkCnt = blockSize >> 2u;
|
|
|
|
|
|
|
|
|
|
/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
|
|
|
|
|
** a second loop below computes the remaining 1 to 3 samples. */
|
|
|
|
|
while (blkCnt > 0u)
|
|
|
|
|
{
|
|
|
|
|
/* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) */
|
|
|
|
|
in1 = *pInput++;
|
|
|
|
|
in2 = *pInput++;
|
|
|
|
|
in3 = *pInput++;
|
|
|
|
|
in4 = *pInput++;
|
|
|
|
|
|
|
|
|
|
sum += in1;
|
|
|
|
|
sum += in2;
|
|
|
|
|
sum += in3;
|
|
|
|
|
sum += in4;
|
|
|
|
|
|
|
|
|
|
/* Decrement the loop counter */
|
|
|
|
|
blkCnt--;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* If the blockSize is not a multiple of 4, compute any remaining output samples here.
|
|
|
|
|
** No loop unrolling is used. */
|
|
|
|
|
blkCnt = blockSize % 0x4u;
|
|
|
|
|
|
|
|
|
|
#else
|
|
|
|
|
/* Run the below code for Cortex-M0 or Cortex-M3 */
|
|
|
|
|
|
|
|
|
|
/* Loop over blockSize number of values */
|
|
|
|
|
blkCnt = blockSize;
|
|
|
|
|
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
while (blkCnt > 0u)
|
|
|
|
|
{
|
|
|
|
|
/* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) */
|
|
|
|
|
sum += *pInput++;
|
|
|
|
|
|
|
|
|
|
/* Decrement the loop counter */
|
|
|
|
|
blkCnt--;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) / blockSize */
|
|
|
|
|
fMean = sum / (float32_t) blockSize;
|
|
|
|
|
|
|
|
|
|
pInput = pSrc;
|
|
|
|
|
|
|
|
|
|
#if !defined(ARM_MATH_CM0_FAMILY) && !defined(ARM_MATH_CM3_FAMILY)
|
|
|
|
|
|
|
|
|
|
/*loop Unrolling */
|
|
|
|
|
blkCnt = blockSize >> 2u;
|
|
|
|
|
|
|
|
|
|
/* First part of the processing with loop unrolling. Compute 4 outputs at a time.
|
|
|
|
|
** a second loop below computes the remaining 1 to 3 samples. */
|
|
|
|
|
while (blkCnt > 0u)
|
|
|
|
|
{
|
|
|
|
|
fValue = *pInput++ - fMean;
|
|
|
|
|
fSum += fValue * fValue;
|
|
|
|
|
fValue = *pInput++ - fMean;
|
|
|
|
|
fSum += fValue * fValue;
|
|
|
|
|
fValue = *pInput++ - fMean;
|
|
|
|
|
fSum += fValue * fValue;
|
|
|
|
|
fValue = *pInput++ - fMean;
|
|
|
|
|
fSum += fValue * fValue;
|
|
|
|
|
|
|
|
|
|
/* Decrement the loop counter */
|
|
|
|
|
blkCnt--;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
blkCnt = blockSize % 0x4u;
|
|
|
|
|
#else
|
|
|
|
|
/* Run the below code for Cortex-M0 or Cortex-M3 */
|
|
|
|
|
|
|
|
|
|
/* Loop over blockSize number of values */
|
|
|
|
|
blkCnt = blockSize;
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
while (blkCnt > 0u)
|
|
|
|
|
{
|
|
|
|
|
fValue = *pInput++ - fMean;
|
|
|
|
|
fSum += fValue * fValue;
|
|
|
|
|
|
|
|
|
|
/* Decrement the loop counter */
|
|
|
|
|
blkCnt--;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Variance */
|
|
|
|
|
*pResult = fSum / (float32_t)(blockSize - 1.0f);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
|