Eli Hughes / CMSIS_DSP_401

V4.0.1 of the ARM CMSIS DSP libraries. Note that arm_bitreversal2.s, arm_cfft_f32.c and arm_rfft_fast_f32.c had to be removed. arm_bitreversal2.s will not assemble with the online tools. So, the fast f32 FFT functions are not yet available. All the other FFT functions are available.

Dependents:   MPU9150_Example fir_f32 fir_f32 MPU9150_nucleo_noni2cdev ... more

FilteringFunctions/arm_biquad_cascade_df1_fast_q15.c@0:3d9c67d97d6f, 2014-07-28 (annotated)

Committer:
emh203
Date:
Mon Jul 28 15:03:15 2014 +0000
Revision:
0:3d9c67d97d6f
1st working commit.   Had to remove arm_bitreversal2.s     arm_cfft_f32.c and arm_rfft_fast_f32.c.    The .s will not assemble.      For now I removed these functions so we could at least have a library for the other functions.

Who changed what in which revision?

UserRevisionLine numberNew contents of line
emh203 0:3d9c67d97d6f 1 /* ----------------------------------------------------------------------
emh203 0:3d9c67d97d6f 2 * Copyright (C) 2010-2014 ARM Limited. All rights reserved.
emh203 0:3d9c67d97d6f 3 *
emh203 0:3d9c67d97d6f 4 * $Date: 12. March 2014
emh203 0:3d9c67d97d6f 5 * $Revision: V1.4.3
emh203 0:3d9c67d97d6f 6 *
emh203 0:3d9c67d97d6f 7 * Project: CMSIS DSP Library
emh203 0:3d9c67d97d6f 8 * Title: arm_biquad_cascade_df1_fast_q15.c
emh203 0:3d9c67d97d6f 9 *
emh203 0:3d9c67d97d6f 10 * Description: Fast processing function for the
emh203 0:3d9c67d97d6f 11 * Q15 Biquad cascade filter.
emh203 0:3d9c67d97d6f 12 *
emh203 0:3d9c67d97d6f 13 * Target Processor: Cortex-M4/Cortex-M3
emh203 0:3d9c67d97d6f 14 *
emh203 0:3d9c67d97d6f 15 * Redistribution and use in source and binary forms, with or without
emh203 0:3d9c67d97d6f 16 * modification, are permitted provided that the following conditions
emh203 0:3d9c67d97d6f 17 * are met:
emh203 0:3d9c67d97d6f 18 * - Redistributions of source code must retain the above copyright
emh203 0:3d9c67d97d6f 19 * notice, this list of conditions and the following disclaimer.
emh203 0:3d9c67d97d6f 20 * - Redistributions in binary form must reproduce the above copyright
emh203 0:3d9c67d97d6f 21 * notice, this list of conditions and the following disclaimer in
emh203 0:3d9c67d97d6f 22 * the documentation and/or other materials provided with the
emh203 0:3d9c67d97d6f 23 * distribution.
emh203 0:3d9c67d97d6f 24 * - Neither the name of ARM LIMITED nor the names of its contributors
emh203 0:3d9c67d97d6f 25 * may be used to endorse or promote products derived from this
emh203 0:3d9c67d97d6f 26 * software without specific prior written permission.
emh203 0:3d9c67d97d6f 27 *
emh203 0:3d9c67d97d6f 28 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
emh203 0:3d9c67d97d6f 29 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
emh203 0:3d9c67d97d6f 30 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
emh203 0:3d9c67d97d6f 31 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
emh203 0:3d9c67d97d6f 32 * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
emh203 0:3d9c67d97d6f 33 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
emh203 0:3d9c67d97d6f 34 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
emh203 0:3d9c67d97d6f 35 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
emh203 0:3d9c67d97d6f 36 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
emh203 0:3d9c67d97d6f 37 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
emh203 0:3d9c67d97d6f 38 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
emh203 0:3d9c67d97d6f 39 * POSSIBILITY OF SUCH DAMAGE.
emh203 0:3d9c67d97d6f 40 * -------------------------------------------------------------------- */
emh203 0:3d9c67d97d6f 41
emh203 0:3d9c67d97d6f 42 #include "arm_math.h"
emh203 0:3d9c67d97d6f 43
emh203 0:3d9c67d97d6f 44 /**
emh203 0:3d9c67d97d6f 45 * @ingroup groupFilters
emh203 0:3d9c67d97d6f 46 */
emh203 0:3d9c67d97d6f 47
emh203 0:3d9c67d97d6f 48 /**
emh203 0:3d9c67d97d6f 49 * @addtogroup BiquadCascadeDF1
emh203 0:3d9c67d97d6f 50 * @{
emh203 0:3d9c67d97d6f 51 */
emh203 0:3d9c67d97d6f 52
emh203 0:3d9c67d97d6f 53 /**
emh203 0:3d9c67d97d6f 54 * @details
emh203 0:3d9c67d97d6f 55 * @param[in] *S points to an instance of the Q15 Biquad cascade structure.
emh203 0:3d9c67d97d6f 56 * @param[in] *pSrc points to the block of input data.
emh203 0:3d9c67d97d6f 57 * @param[out] *pDst points to the block of output data.
emh203 0:3d9c67d97d6f 58 * @param[in] blockSize number of samples to process per call.
emh203 0:3d9c67d97d6f 59 * @return none.
emh203 0:3d9c67d97d6f 60 *
emh203 0:3d9c67d97d6f 61 * <b>Scaling and Overflow Behavior:</b>
emh203 0:3d9c67d97d6f 62 * \par
emh203 0:3d9c67d97d6f 63 * This fast version uses a 32-bit accumulator with 2.30 format.
emh203 0:3d9c67d97d6f 64 * The accumulator maintains full precision of the intermediate multiplication results but provides only a single guard bit.
emh203 0:3d9c67d97d6f 65 * Thus, if the accumulator result overflows it wraps around and distorts the result.
emh203 0:3d9c67d97d6f 66 * In order to avoid overflows completely the input signal must be scaled down by two bits and lie in the range [-0.25 +0.25).
emh203 0:3d9c67d97d6f 67 * The 2.30 accumulator is then shifted by <code>postShift</code> bits and the result truncated to 1.15 format by discarding the low 16 bits.
emh203 0:3d9c67d97d6f 68 *
emh203 0:3d9c67d97d6f 69 * \par
emh203 0:3d9c67d97d6f 70 * Refer to the function <code>arm_biquad_cascade_df1_q15()</code> for a slower implementation of this function which uses 64-bit accumulation to avoid wrap around distortion. Both the slow and the fast versions use the same instance structure.
emh203 0:3d9c67d97d6f 71 * Use the function <code>arm_biquad_cascade_df1_init_q15()</code> to initialize the filter structure.
emh203 0:3d9c67d97d6f 72 *
emh203 0:3d9c67d97d6f 73 */
emh203 0:3d9c67d97d6f 74
emh203 0:3d9c67d97d6f 75 void arm_biquad_cascade_df1_fast_q15(
emh203 0:3d9c67d97d6f 76 const arm_biquad_casd_df1_inst_q15 * S,
emh203 0:3d9c67d97d6f 77 q15_t * pSrc,
emh203 0:3d9c67d97d6f 78 q15_t * pDst,
emh203 0:3d9c67d97d6f 79 uint32_t blockSize)
emh203 0:3d9c67d97d6f 80 {
emh203 0:3d9c67d97d6f 81 q15_t *pIn = pSrc; /* Source pointer */
emh203 0:3d9c67d97d6f 82 q15_t *pOut = pDst; /* Destination pointer */
emh203 0:3d9c67d97d6f 83 q31_t in; /* Temporary variable to hold input value */
emh203 0:3d9c67d97d6f 84 q31_t out; /* Temporary variable to hold output value */
emh203 0:3d9c67d97d6f 85 q31_t b0; /* Temporary variable to hold bo value */
emh203 0:3d9c67d97d6f 86 q31_t b1, a1; /* Filter coefficients */
emh203 0:3d9c67d97d6f 87 q31_t state_in, state_out; /* Filter state variables */
emh203 0:3d9c67d97d6f 88 q31_t acc; /* Accumulator */
emh203 0:3d9c67d97d6f 89 int32_t shift = (int32_t) (15 - S->postShift); /* Post shift */
emh203 0:3d9c67d97d6f 90 q15_t *pState = S->pState; /* State pointer */
emh203 0:3d9c67d97d6f 91 q15_t *pCoeffs = S->pCoeffs; /* Coefficient pointer */
emh203 0:3d9c67d97d6f 92 uint32_t sample, stage = S->numStages; /* Stage loop counter */
emh203 0:3d9c67d97d6f 93
emh203 0:3d9c67d97d6f 94
emh203 0:3d9c67d97d6f 95
emh203 0:3d9c67d97d6f 96 do
emh203 0:3d9c67d97d6f 97 {
emh203 0:3d9c67d97d6f 98
emh203 0:3d9c67d97d6f 99 /* Read the b0 and 0 coefficients using SIMD */
emh203 0:3d9c67d97d6f 100 b0 = *__SIMD32(pCoeffs)++;
emh203 0:3d9c67d97d6f 101
emh203 0:3d9c67d97d6f 102 /* Read the b1 and b2 coefficients using SIMD */
emh203 0:3d9c67d97d6f 103 b1 = *__SIMD32(pCoeffs)++;
emh203 0:3d9c67d97d6f 104
emh203 0:3d9c67d97d6f 105 /* Read the a1 and a2 coefficients using SIMD */
emh203 0:3d9c67d97d6f 106 a1 = *__SIMD32(pCoeffs)++;
emh203 0:3d9c67d97d6f 107
emh203 0:3d9c67d97d6f 108 /* Read the input state values from the state buffer: x[n-1], x[n-2] */
emh203 0:3d9c67d97d6f 109 state_in = *__SIMD32(pState)++;
emh203 0:3d9c67d97d6f 110
emh203 0:3d9c67d97d6f 111 /* Read the output state values from the state buffer: y[n-1], y[n-2] */
emh203 0:3d9c67d97d6f 112 state_out = *__SIMD32(pState)--;
emh203 0:3d9c67d97d6f 113
emh203 0:3d9c67d97d6f 114 /* Apply loop unrolling and compute 2 output values simultaneously. */
emh203 0:3d9c67d97d6f 115 /* The variable acc hold output values that are being computed:
emh203 0:3d9c67d97d6f 116 *
emh203 0:3d9c67d97d6f 117 * acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2]
emh203 0:3d9c67d97d6f 118 * acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2]
emh203 0:3d9c67d97d6f 119 */
emh203 0:3d9c67d97d6f 120 sample = blockSize >> 1u;
emh203 0:3d9c67d97d6f 121
emh203 0:3d9c67d97d6f 122 /* First part of the processing with loop unrolling. Compute 2 outputs at a time.
emh203 0:3d9c67d97d6f 123 ** a second loop below computes the remaining 1 sample. */
emh203 0:3d9c67d97d6f 124 while(sample > 0u)
emh203 0:3d9c67d97d6f 125 {
emh203 0:3d9c67d97d6f 126
emh203 0:3d9c67d97d6f 127 /* Read the input */
emh203 0:3d9c67d97d6f 128 in = *__SIMD32(pIn)++;
emh203 0:3d9c67d97d6f 129
emh203 0:3d9c67d97d6f 130 /* out = b0 * x[n] + 0 * 0 */
emh203 0:3d9c67d97d6f 131 out = __SMUAD(b0, in);
emh203 0:3d9c67d97d6f 132 /* acc = b1 * x[n-1] + acc += b2 * x[n-2] + out */
emh203 0:3d9c67d97d6f 133 acc = __SMLAD(b1, state_in, out);
emh203 0:3d9c67d97d6f 134 /* acc += a1 * y[n-1] + acc += a2 * y[n-2] */
emh203 0:3d9c67d97d6f 135 acc = __SMLAD(a1, state_out, acc);
emh203 0:3d9c67d97d6f 136
emh203 0:3d9c67d97d6f 137 /* The result is converted from 3.29 to 1.31 and then saturation is applied */
emh203 0:3d9c67d97d6f 138 out = __SSAT((acc >> shift), 16);
emh203 0:3d9c67d97d6f 139
emh203 0:3d9c67d97d6f 140 /* Every time after the output is computed state should be updated. */
emh203 0:3d9c67d97d6f 141 /* The states should be updated as: */
emh203 0:3d9c67d97d6f 142 /* Xn2 = Xn1 */
emh203 0:3d9c67d97d6f 143 /* Xn1 = Xn */
emh203 0:3d9c67d97d6f 144 /* Yn2 = Yn1 */
emh203 0:3d9c67d97d6f 145 /* Yn1 = acc */
emh203 0:3d9c67d97d6f 146 /* x[n-N], x[n-N-1] are packed together to make state_in of type q31 */
emh203 0:3d9c67d97d6f 147 /* y[n-N], y[n-N-1] are packed together to make state_out of type q31 */
emh203 0:3d9c67d97d6f 148
emh203 0:3d9c67d97d6f 149 #ifndef ARM_MATH_BIG_ENDIAN
emh203 0:3d9c67d97d6f 150
emh203 0:3d9c67d97d6f 151 state_in = __PKHBT(in, state_in, 16);
emh203 0:3d9c67d97d6f 152 state_out = __PKHBT(out, state_out, 16);
emh203 0:3d9c67d97d6f 153
emh203 0:3d9c67d97d6f 154 #else
emh203 0:3d9c67d97d6f 155
emh203 0:3d9c67d97d6f 156 state_in = __PKHBT(state_in >> 16, (in >> 16), 16);
emh203 0:3d9c67d97d6f 157 state_out = __PKHBT(state_out >> 16, (out), 16);
emh203 0:3d9c67d97d6f 158
emh203 0:3d9c67d97d6f 159 #endif /* #ifndef ARM_MATH_BIG_ENDIAN */
emh203 0:3d9c67d97d6f 160
emh203 0:3d9c67d97d6f 161 /* out = b0 * x[n] + 0 * 0 */
emh203 0:3d9c67d97d6f 162 out = __SMUADX(b0, in);
emh203 0:3d9c67d97d6f 163 /* acc0 = b1 * x[n-1] , acc0 += b2 * x[n-2] + out */
emh203 0:3d9c67d97d6f 164 acc = __SMLAD(b1, state_in, out);
emh203 0:3d9c67d97d6f 165 /* acc += a1 * y[n-1] + acc += a2 * y[n-2] */
emh203 0:3d9c67d97d6f 166 acc = __SMLAD(a1, state_out, acc);
emh203 0:3d9c67d97d6f 167
emh203 0:3d9c67d97d6f 168 /* The result is converted from 3.29 to 1.31 and then saturation is applied */
emh203 0:3d9c67d97d6f 169 out = __SSAT((acc >> shift), 16);
emh203 0:3d9c67d97d6f 170
emh203 0:3d9c67d97d6f 171
emh203 0:3d9c67d97d6f 172 /* Store the output in the destination buffer. */
emh203 0:3d9c67d97d6f 173
emh203 0:3d9c67d97d6f 174 #ifndef ARM_MATH_BIG_ENDIAN
emh203 0:3d9c67d97d6f 175
emh203 0:3d9c67d97d6f 176 *__SIMD32(pOut)++ = __PKHBT(state_out, out, 16);
emh203 0:3d9c67d97d6f 177
emh203 0:3d9c67d97d6f 178 #else
emh203 0:3d9c67d97d6f 179
emh203 0:3d9c67d97d6f 180 *__SIMD32(pOut)++ = __PKHBT(out, state_out >> 16, 16);
emh203 0:3d9c67d97d6f 181
emh203 0:3d9c67d97d6f 182 #endif /* #ifndef ARM_MATH_BIG_ENDIAN */
emh203 0:3d9c67d97d6f 183
emh203 0:3d9c67d97d6f 184 /* Every time after the output is computed state should be updated. */
emh203 0:3d9c67d97d6f 185 /* The states should be updated as: */
emh203 0:3d9c67d97d6f 186 /* Xn2 = Xn1 */
emh203 0:3d9c67d97d6f 187 /* Xn1 = Xn */
emh203 0:3d9c67d97d6f 188 /* Yn2 = Yn1 */
emh203 0:3d9c67d97d6f 189 /* Yn1 = acc */
emh203 0:3d9c67d97d6f 190 /* x[n-N], x[n-N-1] are packed together to make state_in of type q31 */
emh203 0:3d9c67d97d6f 191 /* y[n-N], y[n-N-1] are packed together to make state_out of type q31 */
emh203 0:3d9c67d97d6f 192
emh203 0:3d9c67d97d6f 193 #ifndef ARM_MATH_BIG_ENDIAN
emh203 0:3d9c67d97d6f 194
emh203 0:3d9c67d97d6f 195 state_in = __PKHBT(in >> 16, state_in, 16);
emh203 0:3d9c67d97d6f 196 state_out = __PKHBT(out, state_out, 16);
emh203 0:3d9c67d97d6f 197
emh203 0:3d9c67d97d6f 198 #else
emh203 0:3d9c67d97d6f 199
emh203 0:3d9c67d97d6f 200 state_in = __PKHBT(state_in >> 16, in, 16);
emh203 0:3d9c67d97d6f 201 state_out = __PKHBT(state_out >> 16, out, 16);
emh203 0:3d9c67d97d6f 202
emh203 0:3d9c67d97d6f 203 #endif /* #ifndef ARM_MATH_BIG_ENDIAN */
emh203 0:3d9c67d97d6f 204
emh203 0:3d9c67d97d6f 205
emh203 0:3d9c67d97d6f 206 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 207 sample--;
emh203 0:3d9c67d97d6f 208
emh203 0:3d9c67d97d6f 209 }
emh203 0:3d9c67d97d6f 210
emh203 0:3d9c67d97d6f 211 /* If the blockSize is not a multiple of 2, compute any remaining output samples here.
emh203 0:3d9c67d97d6f 212 ** No loop unrolling is used. */
emh203 0:3d9c67d97d6f 213
emh203 0:3d9c67d97d6f 214 if((blockSize & 0x1u) != 0u)
emh203 0:3d9c67d97d6f 215 {
emh203 0:3d9c67d97d6f 216 /* Read the input */
emh203 0:3d9c67d97d6f 217 in = *pIn++;
emh203 0:3d9c67d97d6f 218
emh203 0:3d9c67d97d6f 219 /* out = b0 * x[n] + 0 * 0 */
emh203 0:3d9c67d97d6f 220
emh203 0:3d9c67d97d6f 221 #ifndef ARM_MATH_BIG_ENDIAN
emh203 0:3d9c67d97d6f 222
emh203 0:3d9c67d97d6f 223 out = __SMUAD(b0, in);
emh203 0:3d9c67d97d6f 224
emh203 0:3d9c67d97d6f 225 #else
emh203 0:3d9c67d97d6f 226
emh203 0:3d9c67d97d6f 227 out = __SMUADX(b0, in);
emh203 0:3d9c67d97d6f 228
emh203 0:3d9c67d97d6f 229 #endif /* #ifndef ARM_MATH_BIG_ENDIAN */
emh203 0:3d9c67d97d6f 230
emh203 0:3d9c67d97d6f 231 /* acc = b1 * x[n-1], acc += b2 * x[n-2] + out */
emh203 0:3d9c67d97d6f 232 acc = __SMLAD(b1, state_in, out);
emh203 0:3d9c67d97d6f 233 /* acc += a1 * y[n-1] + acc += a2 * y[n-2] */
emh203 0:3d9c67d97d6f 234 acc = __SMLAD(a1, state_out, acc);
emh203 0:3d9c67d97d6f 235
emh203 0:3d9c67d97d6f 236 /* The result is converted from 3.29 to 1.31 and then saturation is applied */
emh203 0:3d9c67d97d6f 237 out = __SSAT((acc >> shift), 16);
emh203 0:3d9c67d97d6f 238
emh203 0:3d9c67d97d6f 239 /* Store the output in the destination buffer. */
emh203 0:3d9c67d97d6f 240 *pOut++ = (q15_t) out;
emh203 0:3d9c67d97d6f 241
emh203 0:3d9c67d97d6f 242 /* Every time after the output is computed state should be updated. */
emh203 0:3d9c67d97d6f 243 /* The states should be updated as: */
emh203 0:3d9c67d97d6f 244 /* Xn2 = Xn1 */
emh203 0:3d9c67d97d6f 245 /* Xn1 = Xn */
emh203 0:3d9c67d97d6f 246 /* Yn2 = Yn1 */
emh203 0:3d9c67d97d6f 247 /* Yn1 = acc */
emh203 0:3d9c67d97d6f 248 /* x[n-N], x[n-N-1] are packed together to make state_in of type q31 */
emh203 0:3d9c67d97d6f 249 /* y[n-N], y[n-N-1] are packed together to make state_out of type q31 */
emh203 0:3d9c67d97d6f 250
emh203 0:3d9c67d97d6f 251 #ifndef ARM_MATH_BIG_ENDIAN
emh203 0:3d9c67d97d6f 252
emh203 0:3d9c67d97d6f 253 state_in = __PKHBT(in, state_in, 16);
emh203 0:3d9c67d97d6f 254 state_out = __PKHBT(out, state_out, 16);
emh203 0:3d9c67d97d6f 255
emh203 0:3d9c67d97d6f 256 #else
emh203 0:3d9c67d97d6f 257
emh203 0:3d9c67d97d6f 258 state_in = __PKHBT(state_in >> 16, in, 16);
emh203 0:3d9c67d97d6f 259 state_out = __PKHBT(state_out >> 16, out, 16);
emh203 0:3d9c67d97d6f 260
emh203 0:3d9c67d97d6f 261 #endif /* #ifndef ARM_MATH_BIG_ENDIAN */
emh203 0:3d9c67d97d6f 262
emh203 0:3d9c67d97d6f 263 }
emh203 0:3d9c67d97d6f 264
emh203 0:3d9c67d97d6f 265 /* The first stage goes from the input buffer to the output buffer. */
emh203 0:3d9c67d97d6f 266 /* Subsequent (numStages - 1) occur in-place in the output buffer */
emh203 0:3d9c67d97d6f 267 pIn = pDst;
emh203 0:3d9c67d97d6f 268
emh203 0:3d9c67d97d6f 269 /* Reset the output pointer */
emh203 0:3d9c67d97d6f 270 pOut = pDst;
emh203 0:3d9c67d97d6f 271
emh203 0:3d9c67d97d6f 272 /* Store the updated state variables back into the state array */
emh203 0:3d9c67d97d6f 273 *__SIMD32(pState)++ = state_in;
emh203 0:3d9c67d97d6f 274 *__SIMD32(pState)++ = state_out;
emh203 0:3d9c67d97d6f 275
emh203 0:3d9c67d97d6f 276
emh203 0:3d9c67d97d6f 277 /* Decrement the loop counter */
emh203 0:3d9c67d97d6f 278 stage--;
emh203 0:3d9c67d97d6f 279
emh203 0:3d9c67d97d6f 280 } while(stage > 0u);
emh203 0:3d9c67d97d6f 281 }
emh203 0:3d9c67d97d6f 282
emh203 0:3d9c67d97d6f 283
emh203 0:3d9c67d97d6f 284 /**
emh203 0:3d9c67d97d6f 285 * @} end of BiquadCascadeDF1 group
emh203 0:3d9c67d97d6f 286 */