diff options
Diffstat (limited to 'Drivers/CMSIS/DSP/Source/FilteringFunctions/arm_biquad_cascade_df1_q31.c')
| -rw-r--r-- | Drivers/CMSIS/DSP/Source/FilteringFunctions/arm_biquad_cascade_df1_q31.c | 339 |
1 files changed, 97 insertions, 242 deletions
diff --git a/Drivers/CMSIS/DSP/Source/FilteringFunctions/arm_biquad_cascade_df1_q31.c b/Drivers/CMSIS/DSP/Source/FilteringFunctions/arm_biquad_cascade_df1_q31.c index 4ca3f85af..011e21dfe 100644 --- a/Drivers/CMSIS/DSP/Source/FilteringFunctions/arm_biquad_cascade_df1_q31.c +++ b/Drivers/CMSIS/DSP/Source/FilteringFunctions/arm_biquad_cascade_df1_q31.c @@ -3,13 +3,13 @@ * Title: arm_biquad_cascade_df1_q31.c * Description: Processing function for the Q31 Biquad cascade filter * - * $Date: 27. January 2017 - * $Revision: V.1.5.1 + * $Date: 18. March 2019 + * $Revision: V1.6.0 * * Target Processor: Cortex-M cores * -------------------------------------------------------------------- */ /* - * Copyright (C) 2010-2017 ARM Limited or its affiliates. All rights reserved. + * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved. * * SPDX-License-Identifier: Apache-2.0 * @@ -29,59 +29,54 @@ #include "arm_math.h" /** - * @ingroup groupFilters + @ingroup groupFilters */ /** - * @addtogroup BiquadCascadeDF1 - * @{ + @addtogroup BiquadCascadeDF1 + @{ */ /** - * @brief Processing function for the Q31 Biquad cascade filter. - * @param[in] *S points to an instance of the Q31 Biquad cascade 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. - * - * <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 2 bits and lie in the range [-0.25 +0.25). - * After all 5 multiply-accumulates are performed, the 2.62 accumulator is shifted by <code>postShift</code> bits and the result truncated to - * 1.31 format by discarding the low 32 bits. - * - * \par - * Refer to the function <code>arm_biquad_cascade_df1_fast_q31()</code> for a faster but less precise implementation of this filter for Cortex-M3 and Cortex-M4. + @brief Processing function for the Q31 Biquad cascade filter. + @param[in] S points to an instance of the Q31 Biquad cascade 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 + @return none + + @par Scaling and Overflow Behavior + 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 2 bits and lie in the range [-0.25 +0.25). + After all 5 multiply-accumulates are performed, the 2.62 accumulator is shifted by <code>postShift</code> bits and the result truncated to + 1.31 format by discarding the low 32 bits. + @remark + Refer to \ref arm_biquad_cascade_df1_fast_q31() for a faster but less precise implementation of this filter. */ void arm_biquad_cascade_df1_q31( const arm_biquad_casd_df1_inst_q31 * S, - q31_t * pSrc, - q31_t * pDst, - uint32_t blockSize) + const q31_t * pSrc, + q31_t * pDst, + uint32_t blockSize) { - q63_t acc; /* accumulator */ - uint32_t uShift = ((uint32_t) S->postShift + 1U); - uint32_t lShift = 32U - uShift; /* Shift to be applied to the output */ - q31_t *pIn = pSrc; /* input pointer initialization */ - q31_t *pOut = pDst; /* output pointer initialization */ - q31_t *pState = S->pState; /* pState pointer initialization */ - q31_t *pCoeffs = S->pCoeffs; /* coeff pointer initialization */ - q31_t Xn1, Xn2, Yn1, Yn2; /* Filter state variables */ - q31_t b0, b1, b2, a1, a2; /* Filter coefficients */ - q31_t Xn; /* temporary input */ - uint32_t sample, stage = S->numStages; /* loop counters */ - - -#if defined (ARM_MATH_DSP) - - q31_t acc_l, acc_h; /* temporary output variables */ - - /* Run the below code for Cortex-M4 and Cortex-M3 */ + const q31_t *pIn = pSrc; /* Source pointer */ + q31_t *pOut = pDst; /* Destination pointer */ + q31_t *pState = S->pState; /* pState pointer */ + const q31_t *pCoeffs = S->pCoeffs; /* Coefficient pointer */ + q63_t acc; /* Accumulator */ + q31_t b0, b1, b2, a1, a2; /* Filter coefficients */ + q31_t Xn1, Xn2, Yn1, Yn2; /* Filter pState variables */ + q31_t Xn; /* Temporary input */ + uint32_t uShift = ((uint32_t) S->postShift + 1U); + uint32_t lShift = 32U - uShift; /* Shift to be applied to the output */ + uint32_t sample, stage = S->numStages; /* Loop counters */ + +#if defined (ARM_MATH_LOOPUNROLL) + q31_t acc_l, acc_h; /* temporary output variables */ +#endif do { @@ -92,301 +87,161 @@ void arm_biquad_cascade_df1_q31( a1 = *pCoeffs++; a2 = *pCoeffs++; - /* Reading the state values */ + /* Reading the pState values */ Xn1 = pState[0]; Xn2 = pState[1]; Yn1 = pState[2]; Yn2 = pState[3]; +#if defined (ARM_MATH_LOOPUNROLL) + /* Apply loop unrolling and compute 4 output values simultaneously. */ - /* The variable acc hold output values that are being computed: + /* Variable acc hold output values that are being computed: * - * acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] + * acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */ + /* Loop unrolling: Compute 4 outputs at a time */ sample = 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 (sample > 0U) { - /* Read the input */ + /* Read the first input */ Xn = *pIn++; /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */ - - /* acc = b0 * x[n] */ - acc = (q63_t) b0 *Xn; - /* acc += b1 * x[n-1] */ - acc += (q63_t) b1 *Xn1; - /* acc += b[2] * x[n-2] */ - acc += (q63_t) b2 *Xn2; - /* acc += a1 * y[n-1] */ - acc += (q63_t) a1 *Yn1; - /* acc += a2 * y[n-2] */ - acc += (q63_t) a2 *Yn2; + acc = ((q63_t) b0 * Xn) + ((q63_t) b1 * Xn1) + ((q63_t) b2 * Xn2) + ((q63_t) a1 * Yn1) + ((q63_t) a2 * Yn2); /* The result is converted to 1.31 , Yn2 variable is reused */ - - /* Calc lower part of acc */ - acc_l = acc & 0xffffffff; - - /* Calc upper part of acc */ - acc_h = (acc >> 32) & 0xffffffff; + acc_l = (acc ) & 0xffffffff; /* Calc lower part of acc */ + acc_h = (acc >> 32) & 0xffffffff; /* Calc upper part of acc */ /* Apply shift for lower part of acc and upper part of acc */ Yn2 = (uint32_t) acc_l >> lShift | acc_h << uShift; - /* Store the output in the destination buffer. */ + /* Store output in destination buffer. */ *pOut++ = Yn2; /* Read the second input */ Xn2 = *pIn++; /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */ - - /* acc = b0 * x[n] */ - acc = (q63_t) b0 *Xn2; - /* acc += b1 * x[n-1] */ - acc += (q63_t) b1 *Xn; - /* acc += b[2] * x[n-2] */ - acc += (q63_t) b2 *Xn1; - /* acc += a1 * y[n-1] */ - acc += (q63_t) a1 *Yn2; - /* acc += a2 * y[n-2] */ - acc += (q63_t) a2 *Yn1; - + acc = ((q63_t) b0 * Xn2) + ((q63_t) b1 * Xn) + ((q63_t) b2 * Xn1) + ((q63_t) a1 * Yn2) + ((q63_t) a2 * Yn1); /* The result is converted to 1.31, Yn1 variable is reused */ - - /* Calc lower part of acc */ - acc_l = acc & 0xffffffff; - - /* Calc upper part of acc */ - acc_h = (acc >> 32) & 0xffffffff; - + acc_l = (acc ) & 0xffffffff; /* Calc lower part of acc */ + acc_h = (acc >> 32) & 0xffffffff; /* Calc upper part of acc */ /* Apply shift for lower part of acc and upper part of acc */ Yn1 = (uint32_t) acc_l >> lShift | acc_h << uShift; - /* Store the output in the destination buffer. */ + /* Store output in destination buffer. */ *pOut++ = Yn1; - /* Read the third input */ + /* Read the third input */ Xn1 = *pIn++; /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */ - - /* acc = b0 * x[n] */ - acc = (q63_t) b0 *Xn1; - /* acc += b1 * x[n-1] */ - acc += (q63_t) b1 *Xn2; - /* acc += b[2] * x[n-2] */ - acc += (q63_t) b2 *Xn; - /* acc += a1 * y[n-1] */ - acc += (q63_t) a1 *Yn1; - /* acc += a2 * y[n-2] */ - acc += (q63_t) a2 *Yn2; + acc = ((q63_t) b0 * Xn1) + ((q63_t) b1 * Xn2) + ((q63_t) b2 * Xn) + ((q63_t) a1 * Yn1) + ((q63_t) a2 * Yn2); /* The result is converted to 1.31, Yn2 variable is reused */ - /* Calc lower part of acc */ - acc_l = acc & 0xffffffff; - - /* Calc upper part of acc */ - acc_h = (acc >> 32) & 0xffffffff; - + acc_l = (acc ) & 0xffffffff; /* Calc lower part of acc */ + acc_h = (acc >> 32) & 0xffffffff; /* Calc upper part of acc */ /* Apply shift for lower part of acc and upper part of acc */ Yn2 = (uint32_t) acc_l >> lShift | acc_h << uShift; - /* Store the output in the destination buffer. */ + /* Store output in destination buffer. */ *pOut++ = Yn2; /* Read the forth input */ Xn = *pIn++; /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */ - - /* acc = b0 * x[n] */ - acc = (q63_t) b0 *Xn; - /* acc += b1 * x[n-1] */ - acc += (q63_t) b1 *Xn1; - /* acc += b[2] * x[n-2] */ - acc += (q63_t) b2 *Xn2; - /* acc += a1 * y[n-1] */ - acc += (q63_t) a1 *Yn2; - /* acc += a2 * y[n-2] */ - acc += (q63_t) a2 *Yn1; + acc = ((q63_t) b0 * Xn) + ((q63_t) b1 * Xn1) + ((q63_t) b2 * Xn2) + ((q63_t) a1 * Yn2) + ((q63_t) a2 * Yn1); /* The result is converted to 1.31, Yn1 variable is reused */ - /* Calc lower part of acc */ - acc_l = acc & 0xffffffff; - - /* Calc upper part of acc */ - acc_h = (acc >> 32) & 0xffffffff; + acc_l = (acc ) & 0xffffffff; /* Calc lower part of acc */ + acc_h = (acc >> 32) & 0xffffffff; /* Calc upper part of acc */ /* Apply shift for lower part of acc and upper part of acc */ Yn1 = (uint32_t) acc_l >> lShift | acc_h << uShift; - /* Every time after the output is computed state should be updated. */ - /* The states should be updated as: */ - /* Xn2 = Xn1 */ - /* Xn1 = Xn */ - /* Yn2 = Yn1 */ - /* Yn1 = acc */ - Xn2 = Xn1; - Xn1 = Xn; - - /* Store the output in the destination buffer. */ + /* Store output in destination buffer. */ *pOut++ = Yn1; - /* decrement the loop counter */ - sample--; - } - - /* If the blockSize is not a multiple of 4, compute any remaining output samples here. - ** No loop unrolling is used. */ - sample = (blockSize & 0x3U); - - while (sample > 0U) - { - /* Read the input */ - Xn = *pIn++; - - /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */ - - /* acc = b0 * x[n] */ - acc = (q63_t) b0 *Xn; - /* acc += b1 * x[n-1] */ - acc += (q63_t) b1 *Xn1; - /* acc += b[2] * x[n-2] */ - acc += (q63_t) b2 *Xn2; - /* acc += a1 * y[n-1] */ - acc += (q63_t) a1 *Yn1; - /* acc += a2 * y[n-2] */ - acc += (q63_t) a2 *Yn2; - - /* The result is converted to 1.31 */ - acc = acc >> lShift; - /* Every time after the output is computed state should be updated. */ - /* The states should be updated as: */ - /* Xn2 = Xn1 */ - /* Xn1 = Xn */ - /* Yn2 = Yn1 */ - /* Yn1 = acc */ + /* The states should be updated as: */ + /* Xn2 = Xn1 */ + /* Xn1 = Xn */ + /* Yn2 = Yn1 */ + /* Yn1 = acc */ Xn2 = Xn1; Xn1 = Xn; - Yn2 = Yn1; - Yn1 = (q31_t) acc; - /* Store the output in the destination buffer. */ - *pOut++ = (q31_t) acc; - - /* decrement the loop counter */ + /* decrement loop counter */ sample--; } - /* The first stage goes from the input buffer to the output buffer. */ - /* Subsequent stages occur in-place in the output buffer */ - pIn = pDst; - - /* Reset to destination pointer */ - pOut = pDst; - - /* Store the updated state variables back into the pState array */ - *pState++ = Xn1; - *pState++ = Xn2; - *pState++ = Yn1; - *pState++ = Yn2; - - } while (--stage); + /* Loop unrolling: Compute remaining outputs */ + sample = blockSize & 0x3U; #else - /* Run the below code for Cortex-M0 */ - - do - { - /* Reading the coefficients */ - b0 = *pCoeffs++; - b1 = *pCoeffs++; - b2 = *pCoeffs++; - a1 = *pCoeffs++; - a2 = *pCoeffs++; - - /* Reading the state values */ - Xn1 = pState[0]; - Xn2 = pState[1]; - Yn1 = pState[2]; - Yn2 = pState[3]; - - /* The variables acc holds the output value that is computed: - * acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] - */ - + /* Initialize blkCnt with number of samples */ sample = blockSize; +#endif /* #if defined (ARM_MATH_LOOPUNROLL) */ + while (sample > 0U) { /* Read the input */ Xn = *pIn++; /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */ - /* acc = b0 * x[n] */ - acc = (q63_t) b0 *Xn; - - /* acc += b1 * x[n-1] */ - acc += (q63_t) b1 *Xn1; - /* acc += b[2] * x[n-2] */ - acc += (q63_t) b2 *Xn2; - /* acc += a1 * y[n-1] */ - acc += (q63_t) a1 *Yn1; - /* acc += a2 * y[n-2] */ - acc += (q63_t) a2 *Yn2; + acc = ((q63_t) b0 * Xn) + ((q63_t) b1 * Xn1) + ((q63_t) b2 * Xn2) + ((q63_t) a1 * Yn1) + ((q63_t) a2 * Yn2); /* The result is converted to 1.31 */ acc = acc >> lShift; + /* Store output in destination buffer. */ + *pOut++ = (q31_t) acc; + /* Every time after the output is computed state should be updated. */ - /* The states should be updated as: */ - /* Xn2 = Xn1 */ - /* Xn1 = Xn */ - /* Yn2 = Yn1 */ - /* Yn1 = acc */ + /* The states should be updated as: */ + /* Xn2 = Xn1 */ + /* Xn1 = Xn */ + /* Yn2 = Yn1 */ + /* Yn1 = acc */ Xn2 = Xn1; Xn1 = Xn; Yn2 = Yn1; Yn1 = (q31_t) acc; - /* Store the output in the destination buffer. */ - *pOut++ = (q31_t) acc; - - /* decrement the loop counter */ + /* decrement loop counter */ sample--; } - /* The first stage goes from the input buffer to the output buffer. */ - /* Subsequent stages occur in-place in the output buffer */ - pIn = pDst; - - /* Reset to destination pointer */ - pOut = pDst; - - /* Store the updated state variables back into the pState array */ + /* Store the updated state variables back into the pState array */ *pState++ = Xn1; *pState++ = Xn2; *pState++ = Yn1; *pState++ = Yn2; - } while (--stage); + /* The first stage goes from the input buffer to the output buffer. */ + /* Subsequent numStages occur in-place in the output buffer */ + pIn = pDst; -#endif /* #if defined (ARM_MATH_DSP) */ -} + /* Reset output pointer */ + pOut = pDst; + /* decrement loop counter */ + stage--; + } while (stage > 0U); +} /** - * @} end of BiquadCascadeDF1 group - */ + @} end of BiquadCascadeDF1 group + */ |