123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411 |
- /* ----------------------------------------------------------------------
- * Copyright (C) 2010-2013 ARM Limited. All rights reserved.
- *
- * $Date: 17. January 2013
- * $Revision: V1.4.1
- *
- * Project: CMSIS DSP Library
- * Title: arm_biquad_cascade_df1_q15.c
- *
- * Description: Processing function for the
- * Q15 Biquad cascade DirectFormI(DF1) filter.
- *
- * 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 BiquadCascadeDF1
- * @{
- */
-
- /**
- * @brief Processing function for the Q15 Biquad cascade filter.
- * @param[in] *S points to an instance of the Q15 Biquad cascade structure.
- * @param[in] *pSrc points to the block of input data.
- * @param[out] *pDst points to the location where the output result is written.
- * @param[in] blockSize number of samples to process per call.
- * @return none.
- *
- *
- * <b>Scaling and Overflow Behavior:</b>
- * \par
- * The function is implemented using a 64-bit internal accumulator.
- * Both coefficients and state variables are represented in 1.15 format and multiplications yield a 2.30 result.
- * The 2.30 intermediate results are accumulated in a 64-bit accumulator in 34.30 format.
- * There is no risk of internal overflow with this approach and the full precision of intermediate multiplications is preserved.
- * The accumulator is then shifted by <code>postShift</code> bits to truncate the result to 1.15 format by discarding the low 16 bits.
- * Finally, the result is saturated to 1.15 format.
- *
- * \par
- * Refer to the function <code>arm_biquad_cascade_df1_fast_q15()</code> for a faster but less precise implementation of this filter for Cortex-M3 and Cortex-M4.
- */
-
- void arm_biquad_cascade_df1_q15(
- const arm_biquad_casd_df1_inst_q15 * S,
- q15_t * pSrc,
- q15_t * pDst,
- uint32_t blockSize)
- {
-
-
- #ifndef ARM_MATH_CM0_FAMILY
-
- /* Run the below code for Cortex-M4 and Cortex-M3 */
-
- q15_t *pIn = pSrc; /* Source pointer */
- q15_t *pOut = pDst; /* Destination pointer */
- q31_t in; /* Temporary variable to hold input value */
- q31_t out; /* Temporary variable to hold output value */
- q31_t b0; /* Temporary variable to hold bo value */
- q31_t b1, a1; /* Filter coefficients */
- q31_t state_in, state_out; /* Filter state variables */
- q31_t acc_l, acc_h;
- q63_t acc; /* Accumulator */
- int32_t lShift = (15 - (int32_t) S->postShift); /* Post shift */
- q15_t *pState = S->pState; /* State pointer */
- q15_t *pCoeffs = S->pCoeffs; /* Coefficient pointer */
- uint32_t sample, stage = (uint32_t) S->numStages; /* Stage loop counter */
- int32_t uShift = (32 - lShift);
-
- do
- {
- /* Read the b0 and 0 coefficients using SIMD */
- b0 = *__SIMD32(pCoeffs)++;
-
- /* Read the b1 and b2 coefficients using SIMD */
- b1 = *__SIMD32(pCoeffs)++;
-
- /* Read the a1 and a2 coefficients using SIMD */
- a1 = *__SIMD32(pCoeffs)++;
-
- /* Read the input state values from the state buffer: x[n-1], x[n-2] */
- state_in = *__SIMD32(pState)++;
-
- /* Read the output state values from the state buffer: y[n-1], y[n-2] */
- state_out = *__SIMD32(pState)--;
-
- /* Apply loop unrolling and compute 2 output values simultaneously. */
- /* The 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]
- */
- sample = blockSize >> 1u;
-
- /* First part of the processing with loop unrolling. Compute 2 outputs at a time.
- ** a second loop below computes the remaining 1 sample. */
- while(sample > 0u)
- {
-
- /* Read the input */
- in = *__SIMD32(pIn)++;
-
- /* out = b0 * x[n] + 0 * 0 */
- out = __SMUAD(b0, in);
-
- /* acc += b1 * x[n-1] + b2 * x[n-2] + out */
- acc = __SMLALD(b1, state_in, out);
- /* acc += a1 * y[n-1] + a2 * y[n-2] */
- acc = __SMLALD(a1, state_out, acc);
-
- /* The result is converted from 3.29 to 1.31 if postShift = 1, and then saturation is applied */
- /* Calc lower part of acc */
- acc_l = acc & 0xffffffff;
-
- /* Calc upper part of acc */
- acc_h = (acc >> 32) & 0xffffffff;
-
- /* Apply shift for lower part of acc and upper part of acc */
- out = (uint32_t) acc_l >> lShift | acc_h << uShift;
-
- out = __SSAT(out, 16);
-
- /* 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 */
- /* x[n-N], x[n-N-1] are packed together to make state_in of type q31 */
- /* y[n-N], y[n-N-1] are packed together to make state_out of type q31 */
-
- #ifndef ARM_MATH_BIG_ENDIAN
-
- state_in = __PKHBT(in, state_in, 16);
- state_out = __PKHBT(out, state_out, 16);
-
- #else
-
- state_in = __PKHBT(state_in >> 16, (in >> 16), 16);
- state_out = __PKHBT(state_out >> 16, (out), 16);
-
- #endif /* #ifndef ARM_MATH_BIG_ENDIAN */
-
- /* out = b0 * x[n] + 0 * 0 */
- out = __SMUADX(b0, in);
- /* acc += b1 * x[n-1] + b2 * x[n-2] + out */
- acc = __SMLALD(b1, state_in, out);
- /* acc += a1 * y[n-1] + a2 * y[n-2] */
- acc = __SMLALD(a1, state_out, acc);
-
- /* The result is converted from 3.29 to 1.31 if postShift = 1, and then saturation is applied */
- /* Calc lower part of acc */
- acc_l = acc & 0xffffffff;
-
- /* Calc upper part of acc */
- acc_h = (acc >> 32) & 0xffffffff;
-
- /* Apply shift for lower part of acc and upper part of acc */
- out = (uint32_t) acc_l >> lShift | acc_h << uShift;
-
- out = __SSAT(out, 16);
-
- /* Store the output in the destination buffer. */
-
- #ifndef ARM_MATH_BIG_ENDIAN
-
- *__SIMD32(pOut)++ = __PKHBT(state_out, out, 16);
-
- #else
-
- *__SIMD32(pOut)++ = __PKHBT(out, state_out >> 16, 16);
-
- #endif /* #ifndef ARM_MATH_BIG_ENDIAN */
-
- /* 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 */
- /* x[n-N], x[n-N-1] are packed together to make state_in of type q31 */
- /* y[n-N], y[n-N-1] are packed together to make state_out of type q31 */
- #ifndef ARM_MATH_BIG_ENDIAN
-
- state_in = __PKHBT(in >> 16, state_in, 16);
- state_out = __PKHBT(out, state_out, 16);
-
- #else
-
- state_in = __PKHBT(state_in >> 16, in, 16);
- state_out = __PKHBT(state_out >> 16, out, 16);
-
- #endif /* #ifndef ARM_MATH_BIG_ENDIAN */
-
-
- /* Decrement the loop counter */
- sample--;
-
- }
-
- /* If the blockSize is not a multiple of 2, compute any remaining output samples here.
- ** No loop unrolling is used. */
-
- if((blockSize & 0x1u) != 0u)
- {
- /* Read the input */
- in = *pIn++;
-
- /* out = b0 * x[n] + 0 * 0 */
-
- #ifndef ARM_MATH_BIG_ENDIAN
-
- out = __SMUAD(b0, in);
-
- #else
-
- out = __SMUADX(b0, in);
-
- #endif /* #ifndef ARM_MATH_BIG_ENDIAN */
-
- /* acc = b1 * x[n-1] + b2 * x[n-2] + out */
- acc = __SMLALD(b1, state_in, out);
- /* acc += a1 * y[n-1] + a2 * y[n-2] */
- acc = __SMLALD(a1, state_out, acc);
-
- /* The result is converted from 3.29 to 1.31 if postShift = 1, and then saturation is applied */
- /* Calc lower part of acc */
- acc_l = acc & 0xffffffff;
-
- /* Calc upper part of acc */
- acc_h = (acc >> 32) & 0xffffffff;
-
- /* Apply shift for lower part of acc and upper part of acc */
- out = (uint32_t) acc_l >> lShift | acc_h << uShift;
-
- out = __SSAT(out, 16);
-
- /* Store the output in the destination buffer. */
- *pOut++ = (q15_t) out;
-
- /* 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 */
- /* x[n-N], x[n-N-1] are packed together to make state_in of type q31 */
- /* y[n-N], y[n-N-1] are packed together to make state_out of type q31 */
-
- #ifndef ARM_MATH_BIG_ENDIAN
-
- state_in = __PKHBT(in, state_in, 16);
- state_out = __PKHBT(out, state_out, 16);
-
- #else
-
- state_in = __PKHBT(state_in >> 16, in, 16);
- state_out = __PKHBT(state_out >> 16, out, 16);
-
- #endif /* #ifndef ARM_MATH_BIG_ENDIAN */
-
- }
-
- /* The first stage goes from the input wire to the output wire. */
- /* Subsequent numStages occur in-place in the output wire */
- pIn = pDst;
-
- /* Reset the output pointer */
- pOut = pDst;
-
- /* Store the updated state variables back into the state array */
- *__SIMD32(pState)++ = state_in;
- *__SIMD32(pState)++ = state_out;
-
-
- /* Decrement the loop counter */
- stage--;
-
- } while(stage > 0u);
-
- #else
-
- /* Run the below code for Cortex-M0 */
-
- q15_t *pIn = pSrc; /* Source pointer */
- q15_t *pOut = pDst; /* Destination pointer */
- q15_t b0, b1, b2, a1, a2; /* Filter coefficients */
- q15_t Xn1, Xn2, Yn1, Yn2; /* Filter state variables */
- q15_t Xn; /* temporary input */
- q63_t acc; /* Accumulator */
- int32_t shift = (15 - (int32_t) S->postShift); /* Post shift */
- q15_t *pState = S->pState; /* State pointer */
- q15_t *pCoeffs = S->pCoeffs; /* Coefficient pointer */
- uint32_t sample, stage = (uint32_t) S->numStages; /* Stage loop counter */
-
- do
- {
- /* Reading the coefficients */
- b0 = *pCoeffs++;
- pCoeffs++; // skip the 0 coefficient
- 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]
- */
-
- sample = blockSize;
-
- 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 = (q31_t) b0 *Xn;
-
- /* acc += b1 * x[n-1] */
- acc += (q31_t) b1 *Xn1;
- /* acc += b[2] * x[n-2] */
- acc += (q31_t) b2 *Xn2;
- /* acc += a1 * y[n-1] */
- acc += (q31_t) a1 *Yn1;
- /* acc += a2 * y[n-2] */
- acc += (q31_t) a2 *Yn2;
-
- /* The result is converted to 1.31 */
- acc = __SSAT((acc >> shift), 16);
-
- /* 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;
- Yn2 = Yn1;
- Yn1 = (q15_t) acc;
-
- /* Store the output in the destination buffer. */
- *pOut++ = (q15_t) acc;
-
- /* decrement the 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);
-
- #endif /* #ifndef ARM_MATH_CM0_FAMILY */
-
- }
-
-
- /**
- * @} end of BiquadCascadeDF1 group
- */
|