blob: 35360595b8f0f4afa268be5197209e6874f9bbab [file] [log] [blame]
/* ----------------------------------------------------------------------
* Copyright (C) 2010-2014 ARM Limited. All rights reserved.
*
* $Date: 19. March 2015
* $Revision: V.1.4.5
*
* Project: CMSIS DSP Library
* Title: arm_fir_q31.c
*
* Description: Q31 FIR filter processing function.
*
* Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* - Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* - Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in
* the documentation and/or other materials provided with the
* distribution.
* - Neither the name of ARM LIMITED nor the names of its contributors
* may be used to endorse or promote products derived from this
* software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
* FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
* COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
* BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
* LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
* CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
* ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
* -------------------------------------------------------------------- */
#include "arm_math.h"
/**
* @ingroup groupFilters
*/
/**
* @addtogroup FIR
* @{
*/
/**
* @param[in] *S points to an instance of the Q31 FIR filter structure.
* @param[in] *pSrc points to the block of input data.
* @param[out] *pDst points to the block of output data.
* @param[in] blockSize number of samples to process per call.
* @return none.
*
* @details
* <b>Scaling and Overflow Behavior:</b>
* \par
* The function is implemented using an internal 64-bit accumulator.
* The accumulator has a 2.62 format and maintains full precision of the intermediate multiplication results but provides only a single guard bit.
* Thus, if the accumulator result overflows it wraps around rather than clip.
* In order to avoid overflows completely the input signal must be scaled down by log2(numTaps) bits.
* After all multiply-accumulates are performed, the 2.62 accumulator is right shifted by 31 bits and saturated to 1.31 format to yield the final result.
*
* \par
* Refer to the function <code>arm_fir_fast_q31()</code> for a faster but less precise implementation of this filter for Cortex-M3 and Cortex-M4.
*/
void arm_fir_q31(
const arm_fir_instance_q31 * S,
q31_t * pSrc,
q31_t * pDst,
uint32_t blockSize)
{
q31_t *pState = S->pState; /* State pointer */
q31_t *pCoeffs = S->pCoeffs; /* Coefficient pointer */
q31_t *pStateCurnt; /* Points to the current sample of the state */
#ifndef ARM_MATH_CM0_FAMILY
/* Run the below code for Cortex-M4 and Cortex-M3 */
q31_t x0, x1, x2; /* Temporary variables to hold state */
q31_t c0; /* Temporary variable to hold coefficient value */
q31_t *px; /* Temporary pointer for state */
q31_t *pb; /* Temporary pointer for coefficient buffer */
q63_t acc0, acc1, acc2; /* Accumulators */
uint32_t numTaps = S->numTaps; /* Number of filter coefficients in the filter */
uint32_t i, tapCnt, blkCnt, tapCntN3; /* Loop counters */
/* S->pState points to state array which contains previous frame (numTaps - 1) samples */
/* pStateCurnt points to the location where the new input data should be written */
pStateCurnt = &(S->pState[(numTaps - 1u)]);
/* Apply loop unrolling and compute 4 output values simultaneously.
* The variables acc0 ... acc3 hold output values that are being computed:
*
* acc0 = b[numTaps-1] * x[n-numTaps-1] + b[numTaps-2] * x[n-numTaps-2] + b[numTaps-3] * x[n-numTaps-3] +...+ b[0] * x[0]
* acc1 = b[numTaps-1] * x[n-numTaps] + b[numTaps-2] * x[n-numTaps-1] + b[numTaps-3] * x[n-numTaps-2] +...+ b[0] * x[1]
* acc2 = b[numTaps-1] * x[n-numTaps+1] + b[numTaps-2] * x[n-numTaps] + b[numTaps-3] * x[n-numTaps-1] +...+ b[0] * x[2]
* acc3 = b[numTaps-1] * x[n-numTaps+2] + b[numTaps-2] * x[n-numTaps+1] + b[numTaps-3] * x[n-numTaps] +...+ b[0] * x[3]
*/
blkCnt = blockSize / 3;
blockSize = blockSize - (3 * blkCnt);
tapCnt = numTaps / 3;
tapCntN3 = numTaps - (3 * tapCnt);
/* 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)
{
/* Copy three new input samples into the state buffer */
*pStateCurnt++ = *pSrc++;
*pStateCurnt++ = *pSrc++;
*pStateCurnt++ = *pSrc++;
/* Set all accumulators to zero */
acc0 = 0;
acc1 = 0;
acc2 = 0;
/* Initialize state pointer */
px = pState;
/* Initialize coefficient pointer */
pb = pCoeffs;
/* Read the first two samples from the state buffer:
* x[n-numTaps], x[n-numTaps-1] */
x0 = *(px++);
x1 = *(px++);
/* Loop unrolling. Process 3 taps at a time. */
i = tapCnt;
while(i > 0u)
{
/* Read the b[numTaps] coefficient */
c0 = *pb;
/* Read x[n-numTaps-2] sample */
x2 = *(px++);
/* Perform the multiply-accumulates */
acc0 += ((q63_t) x0 * c0);
acc1 += ((q63_t) x1 * c0);
acc2 += ((q63_t) x2 * c0);
/* Read the coefficient and state */
c0 = *(pb + 1u);
x0 = *(px++);
/* Perform the multiply-accumulates */
acc0 += ((q63_t) x1 * c0);
acc1 += ((q63_t) x2 * c0);
acc2 += ((q63_t) x0 * c0);
/* Read the coefficient and state */
c0 = *(pb + 2u);
x1 = *(px++);
/* update coefficient pointer */
pb += 3u;
/* Perform the multiply-accumulates */
acc0 += ((q63_t) x2 * c0);
acc1 += ((q63_t) x0 * c0);
acc2 += ((q63_t) x1 * c0);
/* Decrement the loop counter */
i--;
}
/* If the filter length is not a multiple of 3, compute the remaining filter taps */
i = tapCntN3;
while(i > 0u)
{
/* Read coefficients */
c0 = *(pb++);
/* Fetch 1 state variable */
x2 = *(px++);
/* Perform the multiply-accumulates */
acc0 += ((q63_t) x0 * c0);
acc1 += ((q63_t) x1 * c0);
acc2 += ((q63_t) x2 * c0);
/* Reuse the present sample states for next sample */
x0 = x1;
x1 = x2;
/* Decrement the loop counter */
i--;
}
/* Advance the state pointer by 3 to process the next group of 3 samples */
pState = pState + 3;
/* The results in the 3 accumulators are in 2.30 format. Convert to 1.31
** Then store the 3 outputs in the destination buffer. */
*pDst++ = (q31_t) (acc0 >> 31u);
*pDst++ = (q31_t) (acc1 >> 31u);
*pDst++ = (q31_t) (acc2 >> 31u);
/* Decrement the samples loop counter */
blkCnt--;
}
/* If the blockSize is not a multiple of 3, compute any remaining output samples here.
** No loop unrolling is used. */
while(blockSize > 0u)
{
/* Copy one sample at a time into state buffer */
*pStateCurnt++ = *pSrc++;
/* Set the accumulator to zero */
acc0 = 0;
/* Initialize state pointer */
px = pState;
/* Initialize Coefficient pointer */
pb = (pCoeffs);
i = numTaps;
/* Perform the multiply-accumulates */
do
{
acc0 += (q63_t) * (px++) * (*(pb++));
i--;
} while(i > 0u);
/* The result is in 2.62 format. Convert to 1.31
** Then store the output in the destination buffer. */
*pDst++ = (q31_t) (acc0 >> 31u);
/* Advance state pointer by 1 for the next sample */
pState = pState + 1;
/* Decrement the samples loop counter */
blockSize--;
}
/* Processing is complete.
** Now copy the last numTaps - 1 samples to the satrt of the state buffer.
** This prepares the state buffer for the next function call. */
/* Points to the start of the state buffer */
pStateCurnt = S->pState;
tapCnt = (numTaps - 1u) >> 2u;
/* copy data */
while(tapCnt > 0u)
{
*pStateCurnt++ = *pState++;
*pStateCurnt++ = *pState++;
*pStateCurnt++ = *pState++;
*pStateCurnt++ = *pState++;
/* Decrement the loop counter */
tapCnt--;
}
/* Calculate remaining number of copies */
tapCnt = (numTaps - 1u) % 0x4u;
/* Copy the remaining q31_t data */
while(tapCnt > 0u)
{
*pStateCurnt++ = *pState++;
/* Decrement the loop counter */
tapCnt--;
}
#else
/* Run the below code for Cortex-M0 */
q31_t *px; /* Temporary pointer for state */
q31_t *pb; /* Temporary pointer for coefficient buffer */
q63_t acc; /* Accumulator */
uint32_t numTaps = S->numTaps; /* Length of the filter */
uint32_t i, tapCnt, blkCnt; /* Loop counters */
/* S->pState buffer contains previous frame (numTaps - 1) samples */
/* pStateCurnt points to the location where the new input data should be written */
pStateCurnt = &(S->pState[(numTaps - 1u)]);
/* Initialize blkCnt with blockSize */
blkCnt = blockSize;
while(blkCnt > 0u)
{
/* Copy one sample at a time into state buffer */
*pStateCurnt++ = *pSrc++;
/* Set the accumulator to zero */
acc = 0;
/* Initialize state pointer */
px = pState;
/* Initialize Coefficient pointer */
pb = pCoeffs;
i = numTaps;
/* Perform the multiply-accumulates */
do
{
/* acc = b[numTaps-1] * x[n-numTaps-1] + b[numTaps-2] * x[n-numTaps-2] + b[numTaps-3] * x[n-numTaps-3] +...+ b[0] * x[0] */
acc += (q63_t) * px++ * *pb++;
i--;
} while(i > 0u);
/* The result is in 2.62 format. Convert to 1.31
** Then store the output in the destination buffer. */
*pDst++ = (q31_t) (acc >> 31u);
/* Advance state pointer by 1 for the next sample */
pState = pState + 1;
/* Decrement the samples loop counter */
blkCnt--;
}
/* Processing is complete.
** Now copy the last numTaps - 1 samples to the starting of the state buffer.
** This prepares the state buffer for the next function call. */
/* Points to the start of the state buffer */
pStateCurnt = S->pState;
/* Copy numTaps number of values */
tapCnt = numTaps - 1u;
/* Copy the data */
while(tapCnt > 0u)
{
*pStateCurnt++ = *pState++;
/* Decrement the loop counter */
tapCnt--;
}
#endif /* #ifndef ARM_MATH_CM0_FAMILY */
}
/**
* @} end of FIR group
*/