mbed official / mbed-dsp

CMSIS DSP library

Dependents:   performance_timer Surfboard_ gps2rtty Capstone ... more

Legacy Warning

This is an mbed 2 library. To learn more about mbed OS 5, visit the docs.

cmsis_dsp/TransformFunctions/arm_cfft_radix2_f32.c@1:fdd22bb7aa52, 2012-11-28 (annotated)

Committer:
emilmont
Date:
Wed Nov 28 12:30:09 2012 +0000
Revision:
1:fdd22bb7aa52
Child:
2:da51fb522205
DSP library code

Who changed what in which revision?

UserRevisionLine numberNew contents of line
emilmont 1:fdd22bb7aa52 1 /* ----------------------------------------------------------------------
emilmont 1:fdd22bb7aa52 2 * Copyright (C) 2010 ARM Limited. All rights reserved.
emilmont 1:fdd22bb7aa52 3 *
emilmont 1:fdd22bb7aa52 4 * $Date: 15. February 2012
emilmont 1:fdd22bb7aa52 5 * $Revision: V1.1.0
emilmont 1:fdd22bb7aa52 6 *
emilmont 1:fdd22bb7aa52 7 * Project: CMSIS DSP Library
emilmont 1:fdd22bb7aa52 8 * Title: arm_cfft_radix2_f32.c
emilmont 1:fdd22bb7aa52 9 *
emilmont 1:fdd22bb7aa52 10 * Description: Radix-2 Decimation in Frequency CFFT & CIFFT Floating point processing function
emilmont 1:fdd22bb7aa52 11 *
emilmont 1:fdd22bb7aa52 12 *
emilmont 1:fdd22bb7aa52 13 * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
emilmont 1:fdd22bb7aa52 14 *
emilmont 1:fdd22bb7aa52 15 * Version 1.1.0 2012/02/15
emilmont 1:fdd22bb7aa52 16 * Updated with more optimizations, bug fixes and minor API changes.
emilmont 1:fdd22bb7aa52 17 *
emilmont 1:fdd22bb7aa52 18 * Version 1.0.3 2010/11/29
emilmont 1:fdd22bb7aa52 19 * Initial version
emilmont 1:fdd22bb7aa52 20 * -------------------------------------------------------------------- */
emilmont 1:fdd22bb7aa52 21
emilmont 1:fdd22bb7aa52 22 #include "arm_math.h"
emilmont 1:fdd22bb7aa52 23
emilmont 1:fdd22bb7aa52 24 /**
emilmont 1:fdd22bb7aa52 25 * @ingroup groupTransforms
emilmont 1:fdd22bb7aa52 26 */
emilmont 1:fdd22bb7aa52 27
emilmont 1:fdd22bb7aa52 28 /**
emilmont 1:fdd22bb7aa52 29 * @defgroup Radix2_CFFT_CIFFT Radix-2 Complex FFT Functions
emilmont 1:fdd22bb7aa52 30 *
emilmont 1:fdd22bb7aa52 31 * \par
emilmont 1:fdd22bb7aa52 32 * Complex Fast Fourier Transform(CFFT) and Complex Inverse Fast Fourier Transform(CIFFT) is an efficient algorithm to compute Discrete Fourier Transform(DFT) and Inverse Discrete Fourier Transform(IDFT).
emilmont 1:fdd22bb7aa52 33 * Computational complexity of CFFT reduces drastically when compared to DFT.
emilmont 1:fdd22bb7aa52 34 * \par
emilmont 1:fdd22bb7aa52 35 * This set of functions implements CFFT/CIFFT
emilmont 1:fdd22bb7aa52 36 * for Q15, Q31, and floating-point data types. The functions operates on in-place buffer which uses same buffer for input and output.
emilmont 1:fdd22bb7aa52 37 * Complex input is stored in input buffer in an interleaved fashion.
emilmont 1:fdd22bb7aa52 38 *
emilmont 1:fdd22bb7aa52 39 * \par
emilmont 1:fdd22bb7aa52 40 * The functions operate on blocks of input and output data and each call to the function processes
emilmont 1:fdd22bb7aa52 41 * <code>2*fftLen</code> samples through the transform. <code>pSrc</code> points to In-place arrays containing <code>2*fftLen</code> values.
emilmont 1:fdd22bb7aa52 42 * \par
emilmont 1:fdd22bb7aa52 43 * The <code>pSrc</code> points to the array of in-place buffer of size <code>2*fftLen</code> and inputs and outputs are stored in an interleaved fashion as shown below.
emilmont 1:fdd22bb7aa52 44 * <pre> {real[0], imag[0], real[1], imag[1],..} </pre>
emilmont 1:fdd22bb7aa52 45 *
emilmont 1:fdd22bb7aa52 46 * \par Lengths supported by the transform:
emilmont 1:fdd22bb7aa52 47 * \par
emilmont 1:fdd22bb7aa52 48 * Internally, the function utilize a radix-2 decimation in frequency(DIF) algorithm
emilmont 1:fdd22bb7aa52 49 * and the size of the FFT supported are of the lengths [16, 32, 64, 128, 256, 512, 1024, 2048, 4096].
emilmont 1:fdd22bb7aa52 50 *
emilmont 1:fdd22bb7aa52 51 *
emilmont 1:fdd22bb7aa52 52 * \par Algorithm:
emilmont 1:fdd22bb7aa52 53 *
emilmont 1:fdd22bb7aa52 54 * <b>Complex Fast Fourier Transform:</b>
emilmont 1:fdd22bb7aa52 55 * \par
emilmont 1:fdd22bb7aa52 56 * Input real and imaginary data:
emilmont 1:fdd22bb7aa52 57 * <pre>
emilmont 1:fdd22bb7aa52 58 * x(n) = xa + j * ya
emilmont 1:fdd22bb7aa52 59 * x(n+N/2 ) = xb + j * yb
emilmont 1:fdd22bb7aa52 60 * </pre>
emilmont 1:fdd22bb7aa52 61 * where N is length of FFT
emilmont 1:fdd22bb7aa52 62 * \par
emilmont 1:fdd22bb7aa52 63 * Output real and imaginary data:
emilmont 1:fdd22bb7aa52 64 * <pre>
emilmont 1:fdd22bb7aa52 65 * X(2r) = xa'+ j * ya'
emilmont 1:fdd22bb7aa52 66 * X(2r+1) = xb'+ j * yb'
emilmont 1:fdd22bb7aa52 67 * </pre>
emilmont 1:fdd22bb7aa52 68 * \par
emilmont 1:fdd22bb7aa52 69 * Twiddle factors for radix-2 FFT:
emilmont 1:fdd22bb7aa52 70 * <pre>
emilmont 1:fdd22bb7aa52 71 * Wn = cosVal + j * (- sinVal)
emilmont 1:fdd22bb7aa52 72 * </pre>
emilmont 1:fdd22bb7aa52 73 *
emilmont 1:fdd22bb7aa52 74 * \par
emilmont 1:fdd22bb7aa52 75 * \image html CFFT_Radix2.gif "Radix-2 Decimation-in Frequency Complex Fast Fourier Transform"
emilmont 1:fdd22bb7aa52 76 *
emilmont 1:fdd22bb7aa52 77 * \par
emilmont 1:fdd22bb7aa52 78 * Output from Radix-2 CFFT Results in Digit reversal order. Interchange middle two branches of every butterfly results in Bit reversed output.
emilmont 1:fdd22bb7aa52 79 * \par
emilmont 1:fdd22bb7aa52 80 * <b> Butterfly CFFT equations:</b>
emilmont 1:fdd22bb7aa52 81 * <pre>
emilmont 1:fdd22bb7aa52 82 * xa' = xa + xb
emilmont 1:fdd22bb7aa52 83 * ya' = ya + yb
emilmont 1:fdd22bb7aa52 84 * xb' = (xa-xb)* cosVal + (ya-yb) * sinVal
emilmont 1:fdd22bb7aa52 85 * yb' = (ya-yb)* cosVal - (xa-xb) * sinVal
emilmont 1:fdd22bb7aa52 86 * </pre>
emilmont 1:fdd22bb7aa52 87 *
emilmont 1:fdd22bb7aa52 88 *
emilmont 1:fdd22bb7aa52 89 * <b>Complex Inverse Fast Fourier Transform:</b>
emilmont 1:fdd22bb7aa52 90 * \par
emilmont 1:fdd22bb7aa52 91 * CIFFT uses same twiddle factor table as CFFT with modifications in the design equation as shown below.
emilmont 1:fdd22bb7aa52 92 *
emilmont 1:fdd22bb7aa52 93 * \par
emilmont 1:fdd22bb7aa52 94 * <b> Modified Butterfly CIFFT equations:</b>
emilmont 1:fdd22bb7aa52 95 * <pre>
emilmont 1:fdd22bb7aa52 96 * xa' = xa + xb
emilmont 1:fdd22bb7aa52 97 * ya' = ya + yb
emilmont 1:fdd22bb7aa52 98 * xb' = (xa-xb)* cosVal - (ya-yb) * sinVal
emilmont 1:fdd22bb7aa52 99 * yb' = (ya-yb)* cosVal + (xa-xb) * sinVal
emilmont 1:fdd22bb7aa52 100 * </pre>
emilmont 1:fdd22bb7aa52 101 *
emilmont 1:fdd22bb7aa52 102 * \par Instance Structure
emilmont 1:fdd22bb7aa52 103 * A separate instance structure must be defined for each Instance but the twiddle factors and bit reversal tables can be reused.
emilmont 1:fdd22bb7aa52 104 * There are separate instance structure declarations for each of the 3 supported data types.
emilmont 1:fdd22bb7aa52 105 *
emilmont 1:fdd22bb7aa52 106 * \par Initialization Functions
emilmont 1:fdd22bb7aa52 107 * There is also an associated initialization function for each data type.
emilmont 1:fdd22bb7aa52 108 * The initialization function performs the following operations:
emilmont 1:fdd22bb7aa52 109 * - Sets the values of the internal structure fields.
emilmont 1:fdd22bb7aa52 110 * - Initializes twiddle factor table and bit reversal table pointers
emilmont 1:fdd22bb7aa52 111 * \par
emilmont 1:fdd22bb7aa52 112 * Use of the initialization function is optional.
emilmont 1:fdd22bb7aa52 113 * However, if the initialization function is used, then the instance structure cannot be placed into a const data section.
emilmont 1:fdd22bb7aa52 114 * To place an instance structure into a const data section, the instance structure must be manually initialized.
emilmont 1:fdd22bb7aa52 115 * Manually initialize the instance structure as follows:
emilmont 1:fdd22bb7aa52 116 * <pre>
emilmont 1:fdd22bb7aa52 117 *arm_cfft_radix2_instance_f32 S = {fftLen, ifftFlag, bitReverseFlag, pTwiddle, pBitRevTable, twidCoefModifier, bitRevFactor, onebyfftLen};
emilmont 1:fdd22bb7aa52 118 *arm_cfft_radix2_instance_q31 S = {fftLen, ifftFlag, bitReverseFlag, pTwiddle, pBitRevTable, twidCoefModifier, bitRevFactor};
emilmont 1:fdd22bb7aa52 119 *arm_cfft_radix2_instance_q15 S = {fftLen, ifftFlag, bitReverseFlag, pTwiddle, pBitRevTable, twidCoefModifier, bitRevFactor};
emilmont 1:fdd22bb7aa52 120 * </pre>
emilmont 1:fdd22bb7aa52 121 * \par
emilmont 1:fdd22bb7aa52 122 * where <code>fftLen</code> length of CFFT/CIFFT; <code>ifftFlag</code> Flag for selection of CFFT or CIFFT(Set ifftFlag to calculate CIFFT otherwise calculates CFFT);
emilmont 1:fdd22bb7aa52 123 * <code>bitReverseFlag</code> Flag for selection of output order(Set bitReverseFlag to output in normal order otherwise output in bit reversed order);
emilmont 1:fdd22bb7aa52 124 * <code>pTwiddle</code>points to array of twiddle coefficients; <code>pBitRevTable</code> points to the array of bit reversal table.
emilmont 1:fdd22bb7aa52 125 * <code>twidCoefModifier</code> modifier for twiddle factor table which supports all FFT lengths with same table;
emilmont 1:fdd22bb7aa52 126 * <code>pBitRevTable</code> modifier for bit reversal table which supports all FFT lengths with same table.
emilmont 1:fdd22bb7aa52 127 * <code>onebyfftLen</code> value of 1/fftLen to calculate CIFFT;
emilmont 1:fdd22bb7aa52 128 *
emilmont 1:fdd22bb7aa52 129 * \par Fixed-Point Behavior
emilmont 1:fdd22bb7aa52 130 * Care must be taken when using the fixed-point versions of the CFFT/CIFFT function.
emilmont 1:fdd22bb7aa52 131 * Refer to the function specific documentation below for usage guidelines.
emilmont 1:fdd22bb7aa52 132 */
emilmont 1:fdd22bb7aa52 133
emilmont 1:fdd22bb7aa52 134
emilmont 1:fdd22bb7aa52 135 /**
emilmont 1:fdd22bb7aa52 136 * @addtogroup Radix2_CFFT_CIFFT
emilmont 1:fdd22bb7aa52 137 * @{
emilmont 1:fdd22bb7aa52 138 */
emilmont 1:fdd22bb7aa52 139
emilmont 1:fdd22bb7aa52 140 /**
emilmont 1:fdd22bb7aa52 141 * @details
emilmont 1:fdd22bb7aa52 142 * @brief Processing function for the floating-point Radix-2 CFFT/CIFFT.
emilmont 1:fdd22bb7aa52 143 * @param[in] *S points to an instance of the floating-point Radix-2 CFFT/CIFFT structure.
emilmont 1:fdd22bb7aa52 144 * @param[in, out] *pSrc points to the complex data buffer of size <code>2*fftLen</code>. Processing occurs in-place.
emilmont 1:fdd22bb7aa52 145 * @return none.
emilmont 1:fdd22bb7aa52 146 */
emilmont 1:fdd22bb7aa52 147
emilmont 1:fdd22bb7aa52 148 void arm_cfft_radix2_f32(
emilmont 1:fdd22bb7aa52 149 const arm_cfft_radix2_instance_f32 * S,
emilmont 1:fdd22bb7aa52 150 float32_t * pSrc)
emilmont 1:fdd22bb7aa52 151 {
emilmont 1:fdd22bb7aa52 152
emilmont 1:fdd22bb7aa52 153 if(S->ifftFlag == 1u)
emilmont 1:fdd22bb7aa52 154 {
emilmont 1:fdd22bb7aa52 155 /* Complex IFFT radix-2 */
emilmont 1:fdd22bb7aa52 156 arm_radix2_butterfly_inverse_f32(pSrc, S->fftLen, S->pTwiddle,
emilmont 1:fdd22bb7aa52 157 S->twidCoefModifier, S->onebyfftLen);
emilmont 1:fdd22bb7aa52 158 }
emilmont 1:fdd22bb7aa52 159 else
emilmont 1:fdd22bb7aa52 160 {
emilmont 1:fdd22bb7aa52 161 /* Complex FFT radix-2 */
emilmont 1:fdd22bb7aa52 162 arm_radix2_butterfly_f32(pSrc, S->fftLen, S->pTwiddle,
emilmont 1:fdd22bb7aa52 163 S->twidCoefModifier);
emilmont 1:fdd22bb7aa52 164 }
emilmont 1:fdd22bb7aa52 165
emilmont 1:fdd22bb7aa52 166 if(S->bitReverseFlag == 1u)
emilmont 1:fdd22bb7aa52 167 {
emilmont 1:fdd22bb7aa52 168 /* Bit Reversal */
emilmont 1:fdd22bb7aa52 169 arm_bitreversal_f32(pSrc, S->fftLen, S->bitRevFactor, S->pBitRevTable);
emilmont 1:fdd22bb7aa52 170 }
emilmont 1:fdd22bb7aa52 171
emilmont 1:fdd22bb7aa52 172 }
emilmont 1:fdd22bb7aa52 173
emilmont 1:fdd22bb7aa52 174
emilmont 1:fdd22bb7aa52 175 /**
emilmont 1:fdd22bb7aa52 176 * @} end of Radix2_CFFT_CIFFT group
emilmont 1:fdd22bb7aa52 177 */
emilmont 1:fdd22bb7aa52 178
emilmont 1:fdd22bb7aa52 179
emilmont 1:fdd22bb7aa52 180
emilmont 1:fdd22bb7aa52 181 /* ----------------------------------------------------------------------
emilmont 1:fdd22bb7aa52 182 ** Internal helper function used by the FFTs
emilmont 1:fdd22bb7aa52 183 ** ------------------------------------------------------------------- */
emilmont 1:fdd22bb7aa52 184
emilmont 1:fdd22bb7aa52 185 /*
emilmont 1:fdd22bb7aa52 186 * @brief Core function for the floating-point CFFT butterfly process.
emilmont 1:fdd22bb7aa52 187 * @param[in, out] *pSrc points to the in-place buffer of floating-point data type.
emilmont 1:fdd22bb7aa52 188 * @param[in] fftLen length of the FFT.
emilmont 1:fdd22bb7aa52 189 * @param[in] *pCoef points to the twiddle coefficient buffer.
emilmont 1:fdd22bb7aa52 190 * @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.
emilmont 1:fdd22bb7aa52 191 * @return none.
emilmont 1:fdd22bb7aa52 192 */
emilmont 1:fdd22bb7aa52 193
emilmont 1:fdd22bb7aa52 194 void arm_radix2_butterfly_f32(
emilmont 1:fdd22bb7aa52 195 float32_t * pSrc,
emilmont 1:fdd22bb7aa52 196 uint32_t fftLen,
emilmont 1:fdd22bb7aa52 197 float32_t * pCoef,
emilmont 1:fdd22bb7aa52 198 uint16_t twidCoefModifier)
emilmont 1:fdd22bb7aa52 199 {
emilmont 1:fdd22bb7aa52 200
emilmont 1:fdd22bb7aa52 201 int i, j, k, l;
emilmont 1:fdd22bb7aa52 202 int n1, n2, ia;
emilmont 1:fdd22bb7aa52 203 float32_t xt, yt, cosVal, sinVal;
emilmont 1:fdd22bb7aa52 204
emilmont 1:fdd22bb7aa52 205 #ifndef ARM_MATH_CM0
emilmont 1:fdd22bb7aa52 206
emilmont 1:fdd22bb7aa52 207 /* Initializations for the first stage */
emilmont 1:fdd22bb7aa52 208 n2 = fftLen;
emilmont 1:fdd22bb7aa52 209
emilmont 1:fdd22bb7aa52 210 n1 = n2;
emilmont 1:fdd22bb7aa52 211 n2 = n2 >> 1;
emilmont 1:fdd22bb7aa52 212 ia = 0;
emilmont 1:fdd22bb7aa52 213
emilmont 1:fdd22bb7aa52 214 // loop for groups
emilmont 1:fdd22bb7aa52 215 for (i = 0; i < n2; i++)
emilmont 1:fdd22bb7aa52 216 {
emilmont 1:fdd22bb7aa52 217 cosVal = pCoef[ia * 2];
emilmont 1:fdd22bb7aa52 218 sinVal = pCoef[(ia * 2) + 1];
emilmont 1:fdd22bb7aa52 219
emilmont 1:fdd22bb7aa52 220 /* Twiddle coefficients index modifier */
emilmont 1:fdd22bb7aa52 221 ia = ia + twidCoefModifier;
emilmont 1:fdd22bb7aa52 222
emilmont 1:fdd22bb7aa52 223 /* index calculation for the input as, */
emilmont 1:fdd22bb7aa52 224 /* pSrc[i + 0], pSrc[i + fftLen/1] */
emilmont 1:fdd22bb7aa52 225 l = i + n2;
emilmont 1:fdd22bb7aa52 226
emilmont 1:fdd22bb7aa52 227 /* Butterfly implementation */
emilmont 1:fdd22bb7aa52 228 xt = pSrc[2 * i] - pSrc[2 * l];
emilmont 1:fdd22bb7aa52 229 pSrc[2 * i] = pSrc[2 * i] + pSrc[2 * l];
emilmont 1:fdd22bb7aa52 230
emilmont 1:fdd22bb7aa52 231 yt = pSrc[2 * i + 1] - pSrc[2 * l + 1];
emilmont 1:fdd22bb7aa52 232 pSrc[2 * i + 1] = pSrc[2 * l + 1] + pSrc[2 * i + 1];
emilmont 1:fdd22bb7aa52 233
emilmont 1:fdd22bb7aa52 234 pSrc[2u * l] = xt * cosVal + yt * sinVal;
emilmont 1:fdd22bb7aa52 235
emilmont 1:fdd22bb7aa52 236 pSrc[2u * l + 1u] = yt * cosVal - xt * sinVal;
emilmont 1:fdd22bb7aa52 237
emilmont 1:fdd22bb7aa52 238 } // groups loop end
emilmont 1:fdd22bb7aa52 239
emilmont 1:fdd22bb7aa52 240 twidCoefModifier = twidCoefModifier << 1u;
emilmont 1:fdd22bb7aa52 241
emilmont 1:fdd22bb7aa52 242 // loop for stage
emilmont 1:fdd22bb7aa52 243 for (k = fftLen / 2; k > 2; k = k >> 1)
emilmont 1:fdd22bb7aa52 244 {
emilmont 1:fdd22bb7aa52 245 n1 = n2;
emilmont 1:fdd22bb7aa52 246 n2 = n2 >> 1;
emilmont 1:fdd22bb7aa52 247 ia = 0;
emilmont 1:fdd22bb7aa52 248
emilmont 1:fdd22bb7aa52 249 // loop for groups
emilmont 1:fdd22bb7aa52 250 for (j = 0; j < n2; j++)
emilmont 1:fdd22bb7aa52 251 {
emilmont 1:fdd22bb7aa52 252 cosVal = pCoef[ia * 2];
emilmont 1:fdd22bb7aa52 253 sinVal = pCoef[(ia * 2) + 1];
emilmont 1:fdd22bb7aa52 254 ia = ia + twidCoefModifier;
emilmont 1:fdd22bb7aa52 255
emilmont 1:fdd22bb7aa52 256 // loop for butterfly
emilmont 1:fdd22bb7aa52 257 for (i = j; i < fftLen; i += n1)
emilmont 1:fdd22bb7aa52 258 {
emilmont 1:fdd22bb7aa52 259 l = i + n2;
emilmont 1:fdd22bb7aa52 260 xt = pSrc[2 * i] - pSrc[2 * l];
emilmont 1:fdd22bb7aa52 261 pSrc[2 * i] = pSrc[2 * i] + pSrc[2 * l];
emilmont 1:fdd22bb7aa52 262
emilmont 1:fdd22bb7aa52 263 yt = pSrc[2 * i + 1] - pSrc[2 * l + 1];
emilmont 1:fdd22bb7aa52 264 pSrc[2 * i + 1] = pSrc[2 * l + 1] + pSrc[2 * i + 1];
emilmont 1:fdd22bb7aa52 265
emilmont 1:fdd22bb7aa52 266 pSrc[2u * l] = xt * cosVal + yt * sinVal;
emilmont 1:fdd22bb7aa52 267
emilmont 1:fdd22bb7aa52 268 pSrc[2u * l + 1u] = yt * cosVal - xt * sinVal;
emilmont 1:fdd22bb7aa52 269
emilmont 1:fdd22bb7aa52 270 } // butterfly loop end
emilmont 1:fdd22bb7aa52 271
emilmont 1:fdd22bb7aa52 272 } // groups loop end
emilmont 1:fdd22bb7aa52 273
emilmont 1:fdd22bb7aa52 274 twidCoefModifier = twidCoefModifier << 1u;
emilmont 1:fdd22bb7aa52 275 } // stages loop end
emilmont 1:fdd22bb7aa52 276
emilmont 1:fdd22bb7aa52 277 n1 = n2;
emilmont 1:fdd22bb7aa52 278 n2 = n2 >> 1;
emilmont 1:fdd22bb7aa52 279 ia = 0;
emilmont 1:fdd22bb7aa52 280
emilmont 1:fdd22bb7aa52 281 cosVal = pCoef[ia * 2];
emilmont 1:fdd22bb7aa52 282 sinVal = pCoef[(ia * 2) + 1];
emilmont 1:fdd22bb7aa52 283 ia = ia + twidCoefModifier;
emilmont 1:fdd22bb7aa52 284
emilmont 1:fdd22bb7aa52 285 // loop for butterfly
emilmont 1:fdd22bb7aa52 286 for (i = 0; i < fftLen; i += n1)
emilmont 1:fdd22bb7aa52 287 {
emilmont 1:fdd22bb7aa52 288 l = i + n2;
emilmont 1:fdd22bb7aa52 289 xt = pSrc[2 * i] - pSrc[2 * l];
emilmont 1:fdd22bb7aa52 290 pSrc[2 * i] = (pSrc[2 * i] + pSrc[2 * l]);
emilmont 1:fdd22bb7aa52 291
emilmont 1:fdd22bb7aa52 292 yt = pSrc[2 * i + 1] - pSrc[2 * l + 1];
emilmont 1:fdd22bb7aa52 293 pSrc[2 * i + 1] = (pSrc[2 * l + 1] + pSrc[2 * i + 1]);
emilmont 1:fdd22bb7aa52 294
emilmont 1:fdd22bb7aa52 295 pSrc[2u * l] = xt;
emilmont 1:fdd22bb7aa52 296
emilmont 1:fdd22bb7aa52 297 pSrc[2u * l + 1u] = yt;
emilmont 1:fdd22bb7aa52 298
emilmont 1:fdd22bb7aa52 299 } // groups loop end
emilmont 1:fdd22bb7aa52 300
emilmont 1:fdd22bb7aa52 301 #else
emilmont 1:fdd22bb7aa52 302
emilmont 1:fdd22bb7aa52 303 //N = fftLen;
emilmont 1:fdd22bb7aa52 304 n2 = fftLen;
emilmont 1:fdd22bb7aa52 305
emilmont 1:fdd22bb7aa52 306 // loop for stage
emilmont 1:fdd22bb7aa52 307 for (k = fftLen; k > 1; k = k >> 1)
emilmont 1:fdd22bb7aa52 308 {
emilmont 1:fdd22bb7aa52 309 n1 = n2;
emilmont 1:fdd22bb7aa52 310 n2 = n2 >> 1;
emilmont 1:fdd22bb7aa52 311 ia = 0;
emilmont 1:fdd22bb7aa52 312
emilmont 1:fdd22bb7aa52 313 // loop for groups
emilmont 1:fdd22bb7aa52 314 for (j = 0; j < n2; j++)
emilmont 1:fdd22bb7aa52 315 {
emilmont 1:fdd22bb7aa52 316 cosVal = pCoef[ia * 2];
emilmont 1:fdd22bb7aa52 317 sinVal = pCoef[(ia * 2) + 1];
emilmont 1:fdd22bb7aa52 318 ia = ia + twidCoefModifier;
emilmont 1:fdd22bb7aa52 319
emilmont 1:fdd22bb7aa52 320 // loop for butterfly
emilmont 1:fdd22bb7aa52 321 for (i = j; i < fftLen; i += n1)
emilmont 1:fdd22bb7aa52 322 {
emilmont 1:fdd22bb7aa52 323 l = i + n2;
emilmont 1:fdd22bb7aa52 324 xt = pSrc[2 * i] - pSrc[2 * l];
emilmont 1:fdd22bb7aa52 325 pSrc[2 * i] = pSrc[2 * i] + pSrc[2 * l];
emilmont 1:fdd22bb7aa52 326
emilmont 1:fdd22bb7aa52 327 yt = pSrc[2 * i + 1] - pSrc[2 * l + 1];
emilmont 1:fdd22bb7aa52 328 pSrc[2 * i + 1] = pSrc[2 * l + 1] + pSrc[2 * i + 1];
emilmont 1:fdd22bb7aa52 329
emilmont 1:fdd22bb7aa52 330 pSrc[2 * l] = (cosVal * xt + sinVal * yt); // >> 15;
emilmont 1:fdd22bb7aa52 331 pSrc[2 * l + 1] = (cosVal * yt - sinVal * xt); // >> 15;
emilmont 1:fdd22bb7aa52 332
emilmont 1:fdd22bb7aa52 333 }
emilmont 1:fdd22bb7aa52 334 }
emilmont 1:fdd22bb7aa52 335 twidCoefModifier = twidCoefModifier << 1u;
emilmont 1:fdd22bb7aa52 336 }
emilmont 1:fdd22bb7aa52 337
emilmont 1:fdd22bb7aa52 338 #endif // #ifndef ARM_MATH_CM0
emilmont 1:fdd22bb7aa52 339
emilmont 1:fdd22bb7aa52 340 }
emilmont 1:fdd22bb7aa52 341
emilmont 1:fdd22bb7aa52 342
emilmont 1:fdd22bb7aa52 343 void arm_radix2_butterfly_inverse_f32(
emilmont 1:fdd22bb7aa52 344 float32_t * pSrc,
emilmont 1:fdd22bb7aa52 345 uint32_t fftLen,
emilmont 1:fdd22bb7aa52 346 float32_t * pCoef,
emilmont 1:fdd22bb7aa52 347 uint16_t twidCoefModifier,
emilmont 1:fdd22bb7aa52 348 float32_t onebyfftLen)
emilmont 1:fdd22bb7aa52 349 {
emilmont 1:fdd22bb7aa52 350
emilmont 1:fdd22bb7aa52 351 int i, j, k, l;
emilmont 1:fdd22bb7aa52 352 int n1, n2, ia;
emilmont 1:fdd22bb7aa52 353 float32_t xt, yt, cosVal, sinVal;
emilmont 1:fdd22bb7aa52 354
emilmont 1:fdd22bb7aa52 355 #ifndef ARM_MATH_CM0
emilmont 1:fdd22bb7aa52 356
emilmont 1:fdd22bb7aa52 357 //N = fftLen;
emilmont 1:fdd22bb7aa52 358 n2 = fftLen;
emilmont 1:fdd22bb7aa52 359
emilmont 1:fdd22bb7aa52 360 n1 = n2;
emilmont 1:fdd22bb7aa52 361 n2 = n2 >> 1;
emilmont 1:fdd22bb7aa52 362 ia = 0;
emilmont 1:fdd22bb7aa52 363
emilmont 1:fdd22bb7aa52 364 // loop for groups
emilmont 1:fdd22bb7aa52 365 for (i = 0; i < n2; i++)
emilmont 1:fdd22bb7aa52 366 {
emilmont 1:fdd22bb7aa52 367 cosVal = pCoef[ia * 2];
emilmont 1:fdd22bb7aa52 368 sinVal = pCoef[(ia * 2) + 1];
emilmont 1:fdd22bb7aa52 369 ia = ia + twidCoefModifier;
emilmont 1:fdd22bb7aa52 370
emilmont 1:fdd22bb7aa52 371 l = i + n2;
emilmont 1:fdd22bb7aa52 372 xt = pSrc[2 * i] - pSrc[2 * l];
emilmont 1:fdd22bb7aa52 373 pSrc[2 * i] = pSrc[2 * i] + pSrc[2 * l];
emilmont 1:fdd22bb7aa52 374
emilmont 1:fdd22bb7aa52 375 yt = pSrc[2 * i + 1] - pSrc[2 * l + 1];
emilmont 1:fdd22bb7aa52 376 pSrc[2 * i + 1] = pSrc[2 * l + 1] + pSrc[2 * i + 1];
emilmont 1:fdd22bb7aa52 377
emilmont 1:fdd22bb7aa52 378 pSrc[2u * l] = xt * cosVal - yt * sinVal;
emilmont 1:fdd22bb7aa52 379
emilmont 1:fdd22bb7aa52 380 pSrc[2u * l + 1u] = yt * cosVal + xt * sinVal;
emilmont 1:fdd22bb7aa52 381
emilmont 1:fdd22bb7aa52 382 } // groups loop end
emilmont 1:fdd22bb7aa52 383
emilmont 1:fdd22bb7aa52 384 twidCoefModifier = twidCoefModifier << 1u;
emilmont 1:fdd22bb7aa52 385
emilmont 1:fdd22bb7aa52 386 // loop for stage
emilmont 1:fdd22bb7aa52 387 for (k = fftLen / 2; k > 2; k = k >> 1)
emilmont 1:fdd22bb7aa52 388 {
emilmont 1:fdd22bb7aa52 389 n1 = n2;
emilmont 1:fdd22bb7aa52 390 n2 = n2 >> 1;
emilmont 1:fdd22bb7aa52 391 ia = 0;
emilmont 1:fdd22bb7aa52 392
emilmont 1:fdd22bb7aa52 393 // loop for groups
emilmont 1:fdd22bb7aa52 394 for (j = 0; j < n2; j++)
emilmont 1:fdd22bb7aa52 395 {
emilmont 1:fdd22bb7aa52 396 cosVal = pCoef[ia * 2];
emilmont 1:fdd22bb7aa52 397 sinVal = pCoef[(ia * 2) + 1];
emilmont 1:fdd22bb7aa52 398 ia = ia + twidCoefModifier;
emilmont 1:fdd22bb7aa52 399
emilmont 1:fdd22bb7aa52 400 // loop for butterfly
emilmont 1:fdd22bb7aa52 401 for (i = j; i < fftLen; i += n1)
emilmont 1:fdd22bb7aa52 402 {
emilmont 1:fdd22bb7aa52 403 l = i + n2;
emilmont 1:fdd22bb7aa52 404 xt = pSrc[2 * i] - pSrc[2 * l];
emilmont 1:fdd22bb7aa52 405 pSrc[2 * i] = pSrc[2 * i] + pSrc[2 * l];
emilmont 1:fdd22bb7aa52 406
emilmont 1:fdd22bb7aa52 407 yt = pSrc[2 * i + 1] - pSrc[2 * l + 1];
emilmont 1:fdd22bb7aa52 408 pSrc[2 * i + 1] = pSrc[2 * l + 1] + pSrc[2 * i + 1];
emilmont 1:fdd22bb7aa52 409
emilmont 1:fdd22bb7aa52 410 pSrc[2u * l] = xt * cosVal - yt * sinVal;
emilmont 1:fdd22bb7aa52 411
emilmont 1:fdd22bb7aa52 412 pSrc[2u * l + 1u] = yt * cosVal + xt * sinVal;
emilmont 1:fdd22bb7aa52 413
emilmont 1:fdd22bb7aa52 414 } // butterfly loop end
emilmont 1:fdd22bb7aa52 415
emilmont 1:fdd22bb7aa52 416 } // groups loop end
emilmont 1:fdd22bb7aa52 417
emilmont 1:fdd22bb7aa52 418 twidCoefModifier = twidCoefModifier << 1u;
emilmont 1:fdd22bb7aa52 419 } // stages loop end
emilmont 1:fdd22bb7aa52 420
emilmont 1:fdd22bb7aa52 421 n1 = n2;
emilmont 1:fdd22bb7aa52 422 n2 = n2 >> 1;
emilmont 1:fdd22bb7aa52 423 ia = 0;
emilmont 1:fdd22bb7aa52 424
emilmont 1:fdd22bb7aa52 425 cosVal = pCoef[ia * 2];
emilmont 1:fdd22bb7aa52 426 sinVal = pCoef[(ia * 2) + 1];
emilmont 1:fdd22bb7aa52 427 ia = ia + twidCoefModifier;
emilmont 1:fdd22bb7aa52 428
emilmont 1:fdd22bb7aa52 429 // loop for butterfly
emilmont 1:fdd22bb7aa52 430 for (i = 0; i < fftLen; i += n1)
emilmont 1:fdd22bb7aa52 431 {
emilmont 1:fdd22bb7aa52 432 l = i + n2;
emilmont 1:fdd22bb7aa52 433 xt = pSrc[2 * i] - pSrc[2 * l];
emilmont 1:fdd22bb7aa52 434 pSrc[2 * i] = (pSrc[2 * i] + pSrc[2 * l]) * onebyfftLen;
emilmont 1:fdd22bb7aa52 435
emilmont 1:fdd22bb7aa52 436 yt = pSrc[2 * i + 1] - pSrc[2 * l + 1];
emilmont 1:fdd22bb7aa52 437 pSrc[2 * i + 1] = (pSrc[2 * l + 1] + pSrc[2 * i + 1]) * onebyfftLen;
emilmont 1:fdd22bb7aa52 438
emilmont 1:fdd22bb7aa52 439 pSrc[2u * l] = xt * onebyfftLen;
emilmont 1:fdd22bb7aa52 440
emilmont 1:fdd22bb7aa52 441 pSrc[2u * l + 1u] = yt * onebyfftLen;
emilmont 1:fdd22bb7aa52 442
emilmont 1:fdd22bb7aa52 443 } // butterfly loop end
emilmont 1:fdd22bb7aa52 444
emilmont 1:fdd22bb7aa52 445 #else
emilmont 1:fdd22bb7aa52 446
emilmont 1:fdd22bb7aa52 447 //N = fftLen;
emilmont 1:fdd22bb7aa52 448 n2 = fftLen;
emilmont 1:fdd22bb7aa52 449
emilmont 1:fdd22bb7aa52 450 // loop for stage
emilmont 1:fdd22bb7aa52 451 for (k = fftLen; k > 2; k = k >> 1)
emilmont 1:fdd22bb7aa52 452 {
emilmont 1:fdd22bb7aa52 453 n1 = n2;
emilmont 1:fdd22bb7aa52 454 n2 = n2 >> 1;
emilmont 1:fdd22bb7aa52 455 ia = 0;
emilmont 1:fdd22bb7aa52 456
emilmont 1:fdd22bb7aa52 457 // loop for groups
emilmont 1:fdd22bb7aa52 458 for (j = 0; j < n2; j++)
emilmont 1:fdd22bb7aa52 459 {
emilmont 1:fdd22bb7aa52 460 cosVal = pCoef[ia * 2];
emilmont 1:fdd22bb7aa52 461 sinVal = pCoef[(ia * 2) + 1];
emilmont 1:fdd22bb7aa52 462 ia = ia + twidCoefModifier;
emilmont 1:fdd22bb7aa52 463
emilmont 1:fdd22bb7aa52 464 // loop for butterfly
emilmont 1:fdd22bb7aa52 465 for (i = j; i < fftLen; i += n1)
emilmont 1:fdd22bb7aa52 466 {
emilmont 1:fdd22bb7aa52 467 l = i + n2;
emilmont 1:fdd22bb7aa52 468 xt = pSrc[2 * i] - pSrc[2 * l];
emilmont 1:fdd22bb7aa52 469 pSrc[2 * i] = pSrc[2 * i] + pSrc[2 * l];
emilmont 1:fdd22bb7aa52 470
emilmont 1:fdd22bb7aa52 471 yt = pSrc[2 * i + 1] - pSrc[2 * l + 1];
emilmont 1:fdd22bb7aa52 472 pSrc[2 * i + 1] = pSrc[2 * l + 1] + pSrc[2 * i + 1];
emilmont 1:fdd22bb7aa52 473
emilmont 1:fdd22bb7aa52 474 pSrc[2u * l] = xt * cosVal - yt * sinVal;
emilmont 1:fdd22bb7aa52 475
emilmont 1:fdd22bb7aa52 476 pSrc[2u * l + 1u] = yt * cosVal + xt * sinVal;
emilmont 1:fdd22bb7aa52 477
emilmont 1:fdd22bb7aa52 478 } // butterfly loop end
emilmont 1:fdd22bb7aa52 479
emilmont 1:fdd22bb7aa52 480 } // groups loop end
emilmont 1:fdd22bb7aa52 481
emilmont 1:fdd22bb7aa52 482 twidCoefModifier = twidCoefModifier << 1u;
emilmont 1:fdd22bb7aa52 483 } // stages loop end
emilmont 1:fdd22bb7aa52 484
emilmont 1:fdd22bb7aa52 485 n1 = n2;
emilmont 1:fdd22bb7aa52 486 n2 = n2 >> 1;
emilmont 1:fdd22bb7aa52 487 ia = 0;
emilmont 1:fdd22bb7aa52 488
emilmont 1:fdd22bb7aa52 489 cosVal = pCoef[ia * 2];
emilmont 1:fdd22bb7aa52 490 sinVal = pCoef[(ia * 2) + 1];
emilmont 1:fdd22bb7aa52 491 ia = ia + twidCoefModifier;
emilmont 1:fdd22bb7aa52 492
emilmont 1:fdd22bb7aa52 493 // loop for butterfly
emilmont 1:fdd22bb7aa52 494 for (i = 0; i < fftLen; i += n1)
emilmont 1:fdd22bb7aa52 495 {
emilmont 1:fdd22bb7aa52 496 l = i + n2;
emilmont 1:fdd22bb7aa52 497 xt = pSrc[2 * i] - pSrc[2 * l];
emilmont 1:fdd22bb7aa52 498 pSrc[2 * i] = (pSrc[2 * i] + pSrc[2 * l]) * onebyfftLen;
emilmont 1:fdd22bb7aa52 499
emilmont 1:fdd22bb7aa52 500 yt = pSrc[2 * i + 1] - pSrc[2 * l + 1];
emilmont 1:fdd22bb7aa52 501 pSrc[2 * i + 1] = (pSrc[2 * l + 1] + pSrc[2 * i + 1]) * onebyfftLen;
emilmont 1:fdd22bb7aa52 502
emilmont 1:fdd22bb7aa52 503 pSrc[2u * l] = xt * onebyfftLen;
emilmont 1:fdd22bb7aa52 504
emilmont 1:fdd22bb7aa52 505 pSrc[2u * l + 1u] = yt * onebyfftLen;
emilmont 1:fdd22bb7aa52 506
emilmont 1:fdd22bb7aa52 507 } // butterfly loop end
emilmont 1:fdd22bb7aa52 508
emilmont 1:fdd22bb7aa52 509 #endif // #ifndef ARM_MATH_CM0
emilmont 1:fdd22bb7aa52 510
emilmont 1:fdd22bb7aa52 511 }