CMSIS DSP Library from CMSIS 2.0. See http://www.onarm.com/cmsis/ for full details
Dependents: K22F_DSP_Matrix_least_square BNO055-ELEC3810 1BNO055 ECE4180Project--Slave2 ... more
src/Cortex-M4-M3/FilteringFunctions/arm_fir_decimate_fast_q15.c@0:1014af42efd9, 2011-03-10 (annotated)
- Committer:
- simon
- Date:
- Thu Mar 10 15:07:50 2011 +0000
- Revision:
- 0:1014af42efd9
Who changed what in which revision?
User | Revision | Line number | New contents of line |
---|---|---|---|
simon | 0:1014af42efd9 | 1 | /* ---------------------------------------------------------------------- |
simon | 0:1014af42efd9 | 2 | * Copyright (C) 2010 ARM Limited. All rights reserved. |
simon | 0:1014af42efd9 | 3 | * |
simon | 0:1014af42efd9 | 4 | * $Date: 29. November 2010 |
simon | 0:1014af42efd9 | 5 | * $Revision: V1.0.3 |
simon | 0:1014af42efd9 | 6 | * |
simon | 0:1014af42efd9 | 7 | * Project: CMSIS DSP Library |
simon | 0:1014af42efd9 | 8 | * Title: arm_fir_decimate_fast_q15.c |
simon | 0:1014af42efd9 | 9 | * |
simon | 0:1014af42efd9 | 10 | * Description: Fast Q15 FIR Decimator. |
simon | 0:1014af42efd9 | 11 | * |
simon | 0:1014af42efd9 | 12 | * Target Processor: Cortex-M4/Cortex-M3 |
simon | 0:1014af42efd9 | 13 | * |
simon | 0:1014af42efd9 | 14 | * Version 1.0.3 2010/11/29 |
simon | 0:1014af42efd9 | 15 | * Re-organized the CMSIS folders and updated documentation. |
simon | 0:1014af42efd9 | 16 | * |
simon | 0:1014af42efd9 | 17 | * Version 1.0.2 2010/11/11 |
simon | 0:1014af42efd9 | 18 | * Documentation updated. |
simon | 0:1014af42efd9 | 19 | * |
simon | 0:1014af42efd9 | 20 | * Version 1.0.1 2010/10/05 |
simon | 0:1014af42efd9 | 21 | * Production release and review comments incorporated. |
simon | 0:1014af42efd9 | 22 | * |
simon | 0:1014af42efd9 | 23 | * Version 1.0.0 2010/09/20 |
simon | 0:1014af42efd9 | 24 | * Production release and review comments incorporated. |
simon | 0:1014af42efd9 | 25 | * -------------------------------------------------------------------- */ |
simon | 0:1014af42efd9 | 26 | |
simon | 0:1014af42efd9 | 27 | #include "arm_math.h" |
simon | 0:1014af42efd9 | 28 | |
simon | 0:1014af42efd9 | 29 | /** |
simon | 0:1014af42efd9 | 30 | * @ingroup groupFilters |
simon | 0:1014af42efd9 | 31 | */ |
simon | 0:1014af42efd9 | 32 | |
simon | 0:1014af42efd9 | 33 | /** |
simon | 0:1014af42efd9 | 34 | * @addtogroup FIR_decimate |
simon | 0:1014af42efd9 | 35 | * @{ |
simon | 0:1014af42efd9 | 36 | */ |
simon | 0:1014af42efd9 | 37 | |
simon | 0:1014af42efd9 | 38 | /** |
simon | 0:1014af42efd9 | 39 | * @brief Processing function for the Q15 FIR decimator (fast variant). |
simon | 0:1014af42efd9 | 40 | * @param[in] *S points to an instance of the Q15 FIR decimator structure. |
simon | 0:1014af42efd9 | 41 | * @param[in] *pSrc points to the block of input data. |
simon | 0:1014af42efd9 | 42 | * @param[out] *pDst points to the block of output data |
simon | 0:1014af42efd9 | 43 | * @param[in] blockSize number of input samples to process per call. |
simon | 0:1014af42efd9 | 44 | * @return none |
simon | 0:1014af42efd9 | 45 | * |
simon | 0:1014af42efd9 | 46 | * <b>Scaling and Overflow Behavior:</b> |
simon | 0:1014af42efd9 | 47 | * \par |
simon | 0:1014af42efd9 | 48 | * This fast version uses a 32-bit accumulator with 2.30 format. |
simon | 0:1014af42efd9 | 49 | * The accumulator maintains full precision of the intermediate multiplication results but provides only a single guard bit. |
simon | 0:1014af42efd9 | 50 | * Thus, if the accumulator result overflows it wraps around and distorts the result. |
simon | 0:1014af42efd9 | 51 | * In order to avoid overflows completely the input signal must be scaled down by log2(numTaps) bits (log2 is read as log to the base 2). |
simon | 0:1014af42efd9 | 52 | * The 2.30 accumulator is then truncated to 2.15 format and saturated to yield the 1.15 result. |
simon | 0:1014af42efd9 | 53 | * |
simon | 0:1014af42efd9 | 54 | * \par |
simon | 0:1014af42efd9 | 55 | * Refer to the function <code>arm_fir_decimate_q15()</code> for a slower implementation of this function which uses 64-bit accumulation to avoid wrap around distortion. |
simon | 0:1014af42efd9 | 56 | * Both the slow and the fast versions use the same instance structure. |
simon | 0:1014af42efd9 | 57 | * Use the function <code>arm_fir_decimate_init_q15()</code> to initialize the filter structure. |
simon | 0:1014af42efd9 | 58 | */ |
simon | 0:1014af42efd9 | 59 | |
simon | 0:1014af42efd9 | 60 | void arm_fir_decimate_fast_q15( |
simon | 0:1014af42efd9 | 61 | const arm_fir_decimate_instance_q15 * S, |
simon | 0:1014af42efd9 | 62 | q15_t * pSrc, |
simon | 0:1014af42efd9 | 63 | q15_t * pDst, |
simon | 0:1014af42efd9 | 64 | uint32_t blockSize) |
simon | 0:1014af42efd9 | 65 | { |
simon | 0:1014af42efd9 | 66 | q15_t *pState = S->pState; /* State pointer */ |
simon | 0:1014af42efd9 | 67 | q15_t *pCoeffs = S->pCoeffs; /* Coefficient pointer */ |
simon | 0:1014af42efd9 | 68 | q15_t *pStateCurnt; /* Points to the current sample of the state */ |
simon | 0:1014af42efd9 | 69 | q15_t *px; /* Temporary pointer for state buffer */ |
simon | 0:1014af42efd9 | 70 | q15_t *pb; /* Temporary pointer coefficient buffer */ |
simon | 0:1014af42efd9 | 71 | q31_t x0, c0; /* Temporary variables to hold state and coefficient values */ |
simon | 0:1014af42efd9 | 72 | q31_t sum0; /* Accumulators */ |
simon | 0:1014af42efd9 | 73 | uint32_t numTaps = S->numTaps; /* Number of taps */ |
simon | 0:1014af42efd9 | 74 | uint32_t i, blkCnt, tapCnt, outBlockSize = blockSize / S->M; /* Loop counters */ |
simon | 0:1014af42efd9 | 75 | |
simon | 0:1014af42efd9 | 76 | |
simon | 0:1014af42efd9 | 77 | /* S->pState buffer contains previous frame (numTaps - 1) samples */ |
simon | 0:1014af42efd9 | 78 | /* pStateCurnt points to the location where the new input data should be written */ |
simon | 0:1014af42efd9 | 79 | pStateCurnt = S->pState + (numTaps - 1u); |
simon | 0:1014af42efd9 | 80 | |
simon | 0:1014af42efd9 | 81 | /* Total number of output samples to be computed */ |
simon | 0:1014af42efd9 | 82 | blkCnt = outBlockSize; |
simon | 0:1014af42efd9 | 83 | |
simon | 0:1014af42efd9 | 84 | while(blkCnt > 0u) |
simon | 0:1014af42efd9 | 85 | { |
simon | 0:1014af42efd9 | 86 | /* Copy decimation factor number of new input samples into the state buffer */ |
simon | 0:1014af42efd9 | 87 | i = S->M; |
simon | 0:1014af42efd9 | 88 | |
simon | 0:1014af42efd9 | 89 | do |
simon | 0:1014af42efd9 | 90 | { |
simon | 0:1014af42efd9 | 91 | *pStateCurnt++ = *pSrc++; |
simon | 0:1014af42efd9 | 92 | |
simon | 0:1014af42efd9 | 93 | } while(--i); |
simon | 0:1014af42efd9 | 94 | |
simon | 0:1014af42efd9 | 95 | /*Set sum to zero */ |
simon | 0:1014af42efd9 | 96 | sum0 = 0; |
simon | 0:1014af42efd9 | 97 | |
simon | 0:1014af42efd9 | 98 | /* Initialize state pointer */ |
simon | 0:1014af42efd9 | 99 | px = pState; |
simon | 0:1014af42efd9 | 100 | |
simon | 0:1014af42efd9 | 101 | /* Initialize coeff pointer */ |
simon | 0:1014af42efd9 | 102 | pb = pCoeffs; |
simon | 0:1014af42efd9 | 103 | |
simon | 0:1014af42efd9 | 104 | /* Loop unrolling. Process 4 taps at a time. */ |
simon | 0:1014af42efd9 | 105 | tapCnt = numTaps >> 2; |
simon | 0:1014af42efd9 | 106 | |
simon | 0:1014af42efd9 | 107 | /* Loop over the number of taps. Unroll by a factor of 4. |
simon | 0:1014af42efd9 | 108 | ** Repeat until we've computed numTaps-4 coefficients. */ |
simon | 0:1014af42efd9 | 109 | while(tapCnt > 0u) |
simon | 0:1014af42efd9 | 110 | { |
simon | 0:1014af42efd9 | 111 | /* Read the Read b[numTaps-1] and b[numTaps-2] coefficients */ |
simon | 0:1014af42efd9 | 112 | c0 = *__SIMD32(pb)++; |
simon | 0:1014af42efd9 | 113 | |
simon | 0:1014af42efd9 | 114 | /* Read x[n-numTaps-1] and x[n-numTaps-2]sample */ |
simon | 0:1014af42efd9 | 115 | x0 = *__SIMD32(px)++; |
simon | 0:1014af42efd9 | 116 | |
simon | 0:1014af42efd9 | 117 | /* Perform the multiply-accumulate */ |
simon | 0:1014af42efd9 | 118 | sum0 = __SMLAD(x0, c0, sum0); |
simon | 0:1014af42efd9 | 119 | |
simon | 0:1014af42efd9 | 120 | /* Read the b[numTaps-3] and b[numTaps-4] coefficient */ |
simon | 0:1014af42efd9 | 121 | c0 = *__SIMD32(pb)++; |
simon | 0:1014af42efd9 | 122 | |
simon | 0:1014af42efd9 | 123 | /* Read x[n-numTaps-2] and x[n-numTaps-3] sample */ |
simon | 0:1014af42efd9 | 124 | x0 = *__SIMD32(px)++; |
simon | 0:1014af42efd9 | 125 | |
simon | 0:1014af42efd9 | 126 | /* Perform the multiply-accumulate */ |
simon | 0:1014af42efd9 | 127 | sum0 = __SMLAD(x0, c0, sum0); |
simon | 0:1014af42efd9 | 128 | |
simon | 0:1014af42efd9 | 129 | /* Decrement the loop counter */ |
simon | 0:1014af42efd9 | 130 | tapCnt--; |
simon | 0:1014af42efd9 | 131 | } |
simon | 0:1014af42efd9 | 132 | |
simon | 0:1014af42efd9 | 133 | /* If the filter length is not a multiple of 4, compute the remaining filter taps */ |
simon | 0:1014af42efd9 | 134 | tapCnt = numTaps % 0x4u; |
simon | 0:1014af42efd9 | 135 | |
simon | 0:1014af42efd9 | 136 | while(tapCnt > 0u) |
simon | 0:1014af42efd9 | 137 | { |
simon | 0:1014af42efd9 | 138 | /* Read coefficients */ |
simon | 0:1014af42efd9 | 139 | c0 = *pb++; |
simon | 0:1014af42efd9 | 140 | |
simon | 0:1014af42efd9 | 141 | /* Fetch 1 state variable */ |
simon | 0:1014af42efd9 | 142 | x0 = *px++; |
simon | 0:1014af42efd9 | 143 | |
simon | 0:1014af42efd9 | 144 | /* Perform the multiply-accumulate */ |
simon | 0:1014af42efd9 | 145 | sum0 = __SMLAD(x0, c0, sum0); |
simon | 0:1014af42efd9 | 146 | |
simon | 0:1014af42efd9 | 147 | /* Decrement the loop counter */ |
simon | 0:1014af42efd9 | 148 | tapCnt--; |
simon | 0:1014af42efd9 | 149 | } |
simon | 0:1014af42efd9 | 150 | |
simon | 0:1014af42efd9 | 151 | /* Advance the state pointer by the decimation factor |
simon | 0:1014af42efd9 | 152 | * to process the next group of decimation factor number samples */ |
simon | 0:1014af42efd9 | 153 | pState = pState + S->M; |
simon | 0:1014af42efd9 | 154 | |
simon | 0:1014af42efd9 | 155 | /* Store filter output , smlad returns the values in 2.14 format */ |
simon | 0:1014af42efd9 | 156 | /* so downsacle by 15 to get output in 1.15 */ |
simon | 0:1014af42efd9 | 157 | *pDst++ = (q15_t) ((sum0 >> 15)); |
simon | 0:1014af42efd9 | 158 | |
simon | 0:1014af42efd9 | 159 | /* Decrement the loop counter */ |
simon | 0:1014af42efd9 | 160 | blkCnt--; |
simon | 0:1014af42efd9 | 161 | } |
simon | 0:1014af42efd9 | 162 | |
simon | 0:1014af42efd9 | 163 | /* Processing is complete. |
simon | 0:1014af42efd9 | 164 | ** Now copy the last numTaps - 1 samples to the satrt of the state buffer. |
simon | 0:1014af42efd9 | 165 | ** This prepares the state buffer for the next function call. */ |
simon | 0:1014af42efd9 | 166 | |
simon | 0:1014af42efd9 | 167 | /* Points to the start of the state buffer */ |
simon | 0:1014af42efd9 | 168 | pStateCurnt = S->pState; |
simon | 0:1014af42efd9 | 169 | |
simon | 0:1014af42efd9 | 170 | i = (numTaps - 1u) >> 2u; |
simon | 0:1014af42efd9 | 171 | |
simon | 0:1014af42efd9 | 172 | /* copy data */ |
simon | 0:1014af42efd9 | 173 | while(i > 0u) |
simon | 0:1014af42efd9 | 174 | { |
simon | 0:1014af42efd9 | 175 | *__SIMD32(pStateCurnt)++ = *__SIMD32(pState)++; |
simon | 0:1014af42efd9 | 176 | *__SIMD32(pStateCurnt)++ = *__SIMD32(pState)++; |
simon | 0:1014af42efd9 | 177 | |
simon | 0:1014af42efd9 | 178 | /* Decrement the loop counter */ |
simon | 0:1014af42efd9 | 179 | i--; |
simon | 0:1014af42efd9 | 180 | } |
simon | 0:1014af42efd9 | 181 | |
simon | 0:1014af42efd9 | 182 | i = (numTaps - 1u) % 0x04u; |
simon | 0:1014af42efd9 | 183 | |
simon | 0:1014af42efd9 | 184 | /* copy data */ |
simon | 0:1014af42efd9 | 185 | while(i > 0u) |
simon | 0:1014af42efd9 | 186 | { |
simon | 0:1014af42efd9 | 187 | *pStateCurnt++ = *pState++; |
simon | 0:1014af42efd9 | 188 | |
simon | 0:1014af42efd9 | 189 | /* Decrement the loop counter */ |
simon | 0:1014af42efd9 | 190 | i--; |
simon | 0:1014af42efd9 | 191 | } |
simon | 0:1014af42efd9 | 192 | } |
simon | 0:1014af42efd9 | 193 | |
simon | 0:1014af42efd9 | 194 | /** |
simon | 0:1014af42efd9 | 195 | * @} end of FIR_decimate group |
simon | 0:1014af42efd9 | 196 | */ |