mbed-os 6.10 versione
cmsis_dsp/TransformFunctions/arm_cfft_radix2_q15.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?
User | Revision | Line number | New 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_q15.c |
emilmont | 1:fdd22bb7aa52 | 9 | * |
emilmont | 1:fdd22bb7aa52 | 10 | * Description: Radix-2 Decimation in Frequency CFFT & CIFFT Fixed 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 0.0.3 2010/03/10 |
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 | */ |
emilmont | 1:fdd22bb7aa52 | 35 | |
emilmont | 1:fdd22bb7aa52 | 36 | |
emilmont | 1:fdd22bb7aa52 | 37 | /** |
emilmont | 1:fdd22bb7aa52 | 38 | * @addtogroup Radix2_CFFT_CIFFT |
emilmont | 1:fdd22bb7aa52 | 39 | * @{ |
emilmont | 1:fdd22bb7aa52 | 40 | */ |
emilmont | 1:fdd22bb7aa52 | 41 | |
emilmont | 1:fdd22bb7aa52 | 42 | /** |
emilmont | 1:fdd22bb7aa52 | 43 | * @details |
emilmont | 1:fdd22bb7aa52 | 44 | * @brief Processing function for the fixed-point CFFT/CIFFT. |
emilmont | 1:fdd22bb7aa52 | 45 | * @param[in] *S points to an instance of the fixed-point CFFT/CIFFT structure. |
emilmont | 1:fdd22bb7aa52 | 46 | * @param[in, out] *pSrc points to the complex data buffer of size <code>2*fftLen</code>. Processing occurs in-place. |
emilmont | 1:fdd22bb7aa52 | 47 | * @return none. |
emilmont | 1:fdd22bb7aa52 | 48 | */ |
emilmont | 1:fdd22bb7aa52 | 49 | |
emilmont | 1:fdd22bb7aa52 | 50 | void arm_cfft_radix2_q15( |
emilmont | 1:fdd22bb7aa52 | 51 | const arm_cfft_radix2_instance_q15 * S, |
emilmont | 1:fdd22bb7aa52 | 52 | q15_t * pSrc) |
emilmont | 1:fdd22bb7aa52 | 53 | { |
emilmont | 1:fdd22bb7aa52 | 54 | |
emilmont | 1:fdd22bb7aa52 | 55 | if(S->ifftFlag == 1u) |
emilmont | 1:fdd22bb7aa52 | 56 | { |
emilmont | 1:fdd22bb7aa52 | 57 | arm_radix2_butterfly_inverse_q15(pSrc, S->fftLen, |
emilmont | 1:fdd22bb7aa52 | 58 | S->pTwiddle, S->twidCoefModifier); |
emilmont | 1:fdd22bb7aa52 | 59 | } |
emilmont | 1:fdd22bb7aa52 | 60 | else |
emilmont | 1:fdd22bb7aa52 | 61 | { |
emilmont | 1:fdd22bb7aa52 | 62 | arm_radix2_butterfly_q15(pSrc, S->fftLen, |
emilmont | 1:fdd22bb7aa52 | 63 | S->pTwiddle, S->twidCoefModifier); |
emilmont | 1:fdd22bb7aa52 | 64 | } |
emilmont | 1:fdd22bb7aa52 | 65 | |
emilmont | 1:fdd22bb7aa52 | 66 | arm_bitreversal_q15(pSrc, S->fftLen, S->bitRevFactor, S->pBitRevTable); |
emilmont | 1:fdd22bb7aa52 | 67 | } |
emilmont | 1:fdd22bb7aa52 | 68 | |
emilmont | 1:fdd22bb7aa52 | 69 | /** |
emilmont | 1:fdd22bb7aa52 | 70 | * @} end of Radix2_CFFT_CIFFT group |
emilmont | 1:fdd22bb7aa52 | 71 | */ |
emilmont | 1:fdd22bb7aa52 | 72 | |
emilmont | 1:fdd22bb7aa52 | 73 | void arm_radix2_butterfly_q15( |
emilmont | 1:fdd22bb7aa52 | 74 | q15_t * pSrc, |
emilmont | 1:fdd22bb7aa52 | 75 | uint32_t fftLen, |
emilmont | 1:fdd22bb7aa52 | 76 | q15_t * pCoef, |
emilmont | 1:fdd22bb7aa52 | 77 | uint16_t twidCoefModifier) |
emilmont | 1:fdd22bb7aa52 | 78 | { |
emilmont | 1:fdd22bb7aa52 | 79 | #ifndef ARM_MATH_CM0 |
emilmont | 1:fdd22bb7aa52 | 80 | |
emilmont | 1:fdd22bb7aa52 | 81 | int i, j, k, l; |
emilmont | 1:fdd22bb7aa52 | 82 | int n1, n2, ia; |
emilmont | 1:fdd22bb7aa52 | 83 | q15_t in; |
emilmont | 1:fdd22bb7aa52 | 84 | q31_t T, S, R; |
emilmont | 1:fdd22bb7aa52 | 85 | q31_t coeff, out1, out2; |
emilmont | 1:fdd22bb7aa52 | 86 | |
emilmont | 1:fdd22bb7aa52 | 87 | //N = fftLen; |
emilmont | 1:fdd22bb7aa52 | 88 | n2 = fftLen; |
emilmont | 1:fdd22bb7aa52 | 89 | |
emilmont | 1:fdd22bb7aa52 | 90 | n1 = n2; |
emilmont | 1:fdd22bb7aa52 | 91 | n2 = n2 >> 1; |
emilmont | 1:fdd22bb7aa52 | 92 | ia = 0; |
emilmont | 1:fdd22bb7aa52 | 93 | |
emilmont | 1:fdd22bb7aa52 | 94 | // loop for groups |
emilmont | 1:fdd22bb7aa52 | 95 | for (i = 0; i < n2; i++) |
emilmont | 1:fdd22bb7aa52 | 96 | { |
emilmont | 1:fdd22bb7aa52 | 97 | coeff = _SIMD32_OFFSET(pCoef + (ia * 2u)); |
emilmont | 1:fdd22bb7aa52 | 98 | |
emilmont | 1:fdd22bb7aa52 | 99 | ia = ia + twidCoefModifier; |
emilmont | 1:fdd22bb7aa52 | 100 | |
emilmont | 1:fdd22bb7aa52 | 101 | l = i + n2; |
emilmont | 1:fdd22bb7aa52 | 102 | |
emilmont | 1:fdd22bb7aa52 | 103 | T = _SIMD32_OFFSET(pSrc + (2 * i)); |
emilmont | 1:fdd22bb7aa52 | 104 | in = ((int16_t) (T & 0xFFFF)) >> 2; |
emilmont | 1:fdd22bb7aa52 | 105 | T = ((T >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
emilmont | 1:fdd22bb7aa52 | 106 | |
emilmont | 1:fdd22bb7aa52 | 107 | S = _SIMD32_OFFSET(pSrc + (2 * l)); |
emilmont | 1:fdd22bb7aa52 | 108 | in = ((int16_t) (S & 0xFFFF)) >> 2; |
emilmont | 1:fdd22bb7aa52 | 109 | S = ((S >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
emilmont | 1:fdd22bb7aa52 | 110 | |
emilmont | 1:fdd22bb7aa52 | 111 | R = __QSUB16(T, S); |
emilmont | 1:fdd22bb7aa52 | 112 | |
emilmont | 1:fdd22bb7aa52 | 113 | _SIMD32_OFFSET(pSrc + (2 * i)) = __SHADD16(T, S); |
emilmont | 1:fdd22bb7aa52 | 114 | |
emilmont | 1:fdd22bb7aa52 | 115 | #ifndef ARM_MATH_BIG_ENDIAN |
emilmont | 1:fdd22bb7aa52 | 116 | |
emilmont | 1:fdd22bb7aa52 | 117 | out1 = __SMUAD(coeff, R) >> 16; |
emilmont | 1:fdd22bb7aa52 | 118 | out2 = __SMUSDX(coeff, R); |
emilmont | 1:fdd22bb7aa52 | 119 | |
emilmont | 1:fdd22bb7aa52 | 120 | #else |
emilmont | 1:fdd22bb7aa52 | 121 | |
emilmont | 1:fdd22bb7aa52 | 122 | out1 = __SMUSDX(R, coeff) >> 16u; |
emilmont | 1:fdd22bb7aa52 | 123 | out2 = __SMUAD(coeff, R); |
emilmont | 1:fdd22bb7aa52 | 124 | |
emilmont | 1:fdd22bb7aa52 | 125 | #endif // #ifndef ARM_MATH_BIG_ENDIAN |
emilmont | 1:fdd22bb7aa52 | 126 | |
emilmont | 1:fdd22bb7aa52 | 127 | _SIMD32_OFFSET(pSrc + (2u * l)) = |
emilmont | 1:fdd22bb7aa52 | 128 | (q31_t) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
emilmont | 1:fdd22bb7aa52 | 129 | |
emilmont | 1:fdd22bb7aa52 | 130 | coeff = _SIMD32_OFFSET(pCoef + (ia * 2u)); |
emilmont | 1:fdd22bb7aa52 | 131 | |
emilmont | 1:fdd22bb7aa52 | 132 | ia = ia + twidCoefModifier; |
emilmont | 1:fdd22bb7aa52 | 133 | |
emilmont | 1:fdd22bb7aa52 | 134 | // loop for butterfly |
emilmont | 1:fdd22bb7aa52 | 135 | i++; |
emilmont | 1:fdd22bb7aa52 | 136 | l++; |
emilmont | 1:fdd22bb7aa52 | 137 | |
emilmont | 1:fdd22bb7aa52 | 138 | T = _SIMD32_OFFSET(pSrc + (2 * i)); |
emilmont | 1:fdd22bb7aa52 | 139 | in = ((int16_t) (T & 0xFFFF)) >> 2; |
emilmont | 1:fdd22bb7aa52 | 140 | T = ((T >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
emilmont | 1:fdd22bb7aa52 | 141 | |
emilmont | 1:fdd22bb7aa52 | 142 | S = _SIMD32_OFFSET(pSrc + (2 * l)); |
emilmont | 1:fdd22bb7aa52 | 143 | in = ((int16_t) (S & 0xFFFF)) >> 2; |
emilmont | 1:fdd22bb7aa52 | 144 | S = ((S >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
emilmont | 1:fdd22bb7aa52 | 145 | |
emilmont | 1:fdd22bb7aa52 | 146 | R = __QSUB16(T, S); |
emilmont | 1:fdd22bb7aa52 | 147 | |
emilmont | 1:fdd22bb7aa52 | 148 | _SIMD32_OFFSET(pSrc + (2 * i)) = __SHADD16(T, S); |
emilmont | 1:fdd22bb7aa52 | 149 | |
emilmont | 1:fdd22bb7aa52 | 150 | #ifndef ARM_MATH_BIG_ENDIAN |
emilmont | 1:fdd22bb7aa52 | 151 | |
emilmont | 1:fdd22bb7aa52 | 152 | out1 = __SMUAD(coeff, R) >> 16; |
emilmont | 1:fdd22bb7aa52 | 153 | out2 = __SMUSDX(coeff, R); |
emilmont | 1:fdd22bb7aa52 | 154 | |
emilmont | 1:fdd22bb7aa52 | 155 | #else |
emilmont | 1:fdd22bb7aa52 | 156 | |
emilmont | 1:fdd22bb7aa52 | 157 | out1 = __SMUSDX(R, coeff) >> 16u; |
emilmont | 1:fdd22bb7aa52 | 158 | out2 = __SMUAD(coeff, R); |
emilmont | 1:fdd22bb7aa52 | 159 | |
emilmont | 1:fdd22bb7aa52 | 160 | #endif // #ifndef ARM_MATH_BIG_ENDIAN |
emilmont | 1:fdd22bb7aa52 | 161 | |
emilmont | 1:fdd22bb7aa52 | 162 | _SIMD32_OFFSET(pSrc + (2u * l)) = |
emilmont | 1:fdd22bb7aa52 | 163 | (q31_t) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
emilmont | 1:fdd22bb7aa52 | 164 | |
emilmont | 1:fdd22bb7aa52 | 165 | } // groups loop end |
emilmont | 1:fdd22bb7aa52 | 166 | |
emilmont | 1:fdd22bb7aa52 | 167 | twidCoefModifier = twidCoefModifier << 1u; |
emilmont | 1:fdd22bb7aa52 | 168 | |
emilmont | 1:fdd22bb7aa52 | 169 | // loop for stage |
emilmont | 1:fdd22bb7aa52 | 170 | for (k = fftLen / 2; k > 2; k = k >> 1) |
emilmont | 1:fdd22bb7aa52 | 171 | { |
emilmont | 1:fdd22bb7aa52 | 172 | n1 = n2; |
emilmont | 1:fdd22bb7aa52 | 173 | n2 = n2 >> 1; |
emilmont | 1:fdd22bb7aa52 | 174 | ia = 0; |
emilmont | 1:fdd22bb7aa52 | 175 | |
emilmont | 1:fdd22bb7aa52 | 176 | // loop for groups |
emilmont | 1:fdd22bb7aa52 | 177 | for (j = 0; j < n2; j++) |
emilmont | 1:fdd22bb7aa52 | 178 | { |
emilmont | 1:fdd22bb7aa52 | 179 | coeff = _SIMD32_OFFSET(pCoef + (ia * 2u)); |
emilmont | 1:fdd22bb7aa52 | 180 | |
emilmont | 1:fdd22bb7aa52 | 181 | ia = ia + twidCoefModifier; |
emilmont | 1:fdd22bb7aa52 | 182 | |
emilmont | 1:fdd22bb7aa52 | 183 | // loop for butterfly |
emilmont | 1:fdd22bb7aa52 | 184 | for (i = j; i < fftLen; i += n1) |
emilmont | 1:fdd22bb7aa52 | 185 | { |
emilmont | 1:fdd22bb7aa52 | 186 | l = i + n2; |
emilmont | 1:fdd22bb7aa52 | 187 | |
emilmont | 1:fdd22bb7aa52 | 188 | T = _SIMD32_OFFSET(pSrc + (2 * i)); |
emilmont | 1:fdd22bb7aa52 | 189 | |
emilmont | 1:fdd22bb7aa52 | 190 | S = _SIMD32_OFFSET(pSrc + (2 * l)); |
emilmont | 1:fdd22bb7aa52 | 191 | |
emilmont | 1:fdd22bb7aa52 | 192 | R = __QSUB16(T, S); |
emilmont | 1:fdd22bb7aa52 | 193 | |
emilmont | 1:fdd22bb7aa52 | 194 | _SIMD32_OFFSET(pSrc + (2 * i)) = __SHADD16(T, S); |
emilmont | 1:fdd22bb7aa52 | 195 | |
emilmont | 1:fdd22bb7aa52 | 196 | #ifndef ARM_MATH_BIG_ENDIAN |
emilmont | 1:fdd22bb7aa52 | 197 | |
emilmont | 1:fdd22bb7aa52 | 198 | out1 = __SMUAD(coeff, R) >> 16; |
emilmont | 1:fdd22bb7aa52 | 199 | out2 = __SMUSDX(coeff, R); |
emilmont | 1:fdd22bb7aa52 | 200 | |
emilmont | 1:fdd22bb7aa52 | 201 | #else |
emilmont | 1:fdd22bb7aa52 | 202 | |
emilmont | 1:fdd22bb7aa52 | 203 | out1 = __SMUSDX(R, coeff) >> 16u; |
emilmont | 1:fdd22bb7aa52 | 204 | out2 = __SMUAD(coeff, R); |
emilmont | 1:fdd22bb7aa52 | 205 | |
emilmont | 1:fdd22bb7aa52 | 206 | #endif // #ifndef ARM_MATH_BIG_ENDIAN |
emilmont | 1:fdd22bb7aa52 | 207 | |
emilmont | 1:fdd22bb7aa52 | 208 | _SIMD32_OFFSET(pSrc + (2u * l)) = |
emilmont | 1:fdd22bb7aa52 | 209 | (q31_t) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
emilmont | 1:fdd22bb7aa52 | 210 | |
emilmont | 1:fdd22bb7aa52 | 211 | i += n1; |
emilmont | 1:fdd22bb7aa52 | 212 | |
emilmont | 1:fdd22bb7aa52 | 213 | l = i + n2; |
emilmont | 1:fdd22bb7aa52 | 214 | |
emilmont | 1:fdd22bb7aa52 | 215 | T = _SIMD32_OFFSET(pSrc + (2 * i)); |
emilmont | 1:fdd22bb7aa52 | 216 | |
emilmont | 1:fdd22bb7aa52 | 217 | S = _SIMD32_OFFSET(pSrc + (2 * l)); |
emilmont | 1:fdd22bb7aa52 | 218 | |
emilmont | 1:fdd22bb7aa52 | 219 | R = __QSUB16(T, S); |
emilmont | 1:fdd22bb7aa52 | 220 | |
emilmont | 1:fdd22bb7aa52 | 221 | _SIMD32_OFFSET(pSrc + (2 * i)) = __SHADD16(T, S); |
emilmont | 1:fdd22bb7aa52 | 222 | |
emilmont | 1:fdd22bb7aa52 | 223 | #ifndef ARM_MATH_BIG_ENDIAN |
emilmont | 1:fdd22bb7aa52 | 224 | |
emilmont | 1:fdd22bb7aa52 | 225 | out1 = __SMUAD(coeff, R) >> 16; |
emilmont | 1:fdd22bb7aa52 | 226 | out2 = __SMUSDX(coeff, R); |
emilmont | 1:fdd22bb7aa52 | 227 | |
emilmont | 1:fdd22bb7aa52 | 228 | #else |
emilmont | 1:fdd22bb7aa52 | 229 | |
emilmont | 1:fdd22bb7aa52 | 230 | out1 = __SMUSDX(R, coeff) >> 16u; |
emilmont | 1:fdd22bb7aa52 | 231 | out2 = __SMUAD(coeff, R); |
emilmont | 1:fdd22bb7aa52 | 232 | |
emilmont | 1:fdd22bb7aa52 | 233 | #endif // #ifndef ARM_MATH_BIG_ENDIAN |
emilmont | 1:fdd22bb7aa52 | 234 | |
emilmont | 1:fdd22bb7aa52 | 235 | _SIMD32_OFFSET(pSrc + (2u * l)) = |
emilmont | 1:fdd22bb7aa52 | 236 | (q31_t) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
emilmont | 1:fdd22bb7aa52 | 237 | |
emilmont | 1:fdd22bb7aa52 | 238 | } // butterfly loop end |
emilmont | 1:fdd22bb7aa52 | 239 | |
emilmont | 1:fdd22bb7aa52 | 240 | } // groups loop end |
emilmont | 1:fdd22bb7aa52 | 241 | |
emilmont | 1:fdd22bb7aa52 | 242 | twidCoefModifier = twidCoefModifier << 1u; |
emilmont | 1:fdd22bb7aa52 | 243 | } // stages loop end |
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 | coeff = _SIMD32_OFFSET(pCoef + (ia * 2u)); |
emilmont | 1:fdd22bb7aa52 | 250 | |
emilmont | 1:fdd22bb7aa52 | 251 | ia = ia + twidCoefModifier; |
emilmont | 1:fdd22bb7aa52 | 252 | |
emilmont | 1:fdd22bb7aa52 | 253 | // loop for butterfly |
emilmont | 1:fdd22bb7aa52 | 254 | for (i = 0; i < fftLen; i += n1) |
emilmont | 1:fdd22bb7aa52 | 255 | { |
emilmont | 1:fdd22bb7aa52 | 256 | l = i + n2; |
emilmont | 1:fdd22bb7aa52 | 257 | |
emilmont | 1:fdd22bb7aa52 | 258 | T = _SIMD32_OFFSET(pSrc + (2 * i)); |
emilmont | 1:fdd22bb7aa52 | 259 | |
emilmont | 1:fdd22bb7aa52 | 260 | S = _SIMD32_OFFSET(pSrc + (2 * l)); |
emilmont | 1:fdd22bb7aa52 | 261 | |
emilmont | 1:fdd22bb7aa52 | 262 | R = __QSUB16(T, S); |
emilmont | 1:fdd22bb7aa52 | 263 | |
emilmont | 1:fdd22bb7aa52 | 264 | _SIMD32_OFFSET(pSrc + (2 * i)) = __QADD16(T, S); |
emilmont | 1:fdd22bb7aa52 | 265 | |
emilmont | 1:fdd22bb7aa52 | 266 | _SIMD32_OFFSET(pSrc + (2u * l)) = R; |
emilmont | 1:fdd22bb7aa52 | 267 | |
emilmont | 1:fdd22bb7aa52 | 268 | i += n1; |
emilmont | 1:fdd22bb7aa52 | 269 | l = i + n2; |
emilmont | 1:fdd22bb7aa52 | 270 | |
emilmont | 1:fdd22bb7aa52 | 271 | T = _SIMD32_OFFSET(pSrc + (2 * i)); |
emilmont | 1:fdd22bb7aa52 | 272 | |
emilmont | 1:fdd22bb7aa52 | 273 | S = _SIMD32_OFFSET(pSrc + (2 * l)); |
emilmont | 1:fdd22bb7aa52 | 274 | |
emilmont | 1:fdd22bb7aa52 | 275 | R = __QSUB16(T, S); |
emilmont | 1:fdd22bb7aa52 | 276 | |
emilmont | 1:fdd22bb7aa52 | 277 | _SIMD32_OFFSET(pSrc + (2 * i)) = __QADD16(T, S); |
emilmont | 1:fdd22bb7aa52 | 278 | |
emilmont | 1:fdd22bb7aa52 | 279 | _SIMD32_OFFSET(pSrc + (2u * l)) = R; |
emilmont | 1:fdd22bb7aa52 | 280 | |
emilmont | 1:fdd22bb7aa52 | 281 | } // groups loop end |
emilmont | 1:fdd22bb7aa52 | 282 | |
emilmont | 1:fdd22bb7aa52 | 283 | |
emilmont | 1:fdd22bb7aa52 | 284 | #else |
emilmont | 1:fdd22bb7aa52 | 285 | |
emilmont | 1:fdd22bb7aa52 | 286 | int i, j, k, l; |
emilmont | 1:fdd22bb7aa52 | 287 | int n1, n2, ia; |
emilmont | 1:fdd22bb7aa52 | 288 | q15_t xt, yt, cosVal, sinVal; |
emilmont | 1:fdd22bb7aa52 | 289 | |
emilmont | 1:fdd22bb7aa52 | 290 | |
emilmont | 1:fdd22bb7aa52 | 291 | //N = fftLen; |
emilmont | 1:fdd22bb7aa52 | 292 | n2 = fftLen; |
emilmont | 1:fdd22bb7aa52 | 293 | |
emilmont | 1:fdd22bb7aa52 | 294 | n1 = n2; |
emilmont | 1:fdd22bb7aa52 | 295 | n2 = n2 >> 1; |
emilmont | 1:fdd22bb7aa52 | 296 | ia = 0; |
emilmont | 1:fdd22bb7aa52 | 297 | |
emilmont | 1:fdd22bb7aa52 | 298 | // loop for groups |
emilmont | 1:fdd22bb7aa52 | 299 | for (j = 0; j < n2; j++) |
emilmont | 1:fdd22bb7aa52 | 300 | { |
emilmont | 1:fdd22bb7aa52 | 301 | cosVal = pCoef[ia * 2]; |
emilmont | 1:fdd22bb7aa52 | 302 | sinVal = pCoef[(ia * 2) + 1]; |
emilmont | 1:fdd22bb7aa52 | 303 | ia = ia + twidCoefModifier; |
emilmont | 1:fdd22bb7aa52 | 304 | |
emilmont | 1:fdd22bb7aa52 | 305 | // loop for butterfly |
emilmont | 1:fdd22bb7aa52 | 306 | for (i = j; i < fftLen; i += n1) |
emilmont | 1:fdd22bb7aa52 | 307 | { |
emilmont | 1:fdd22bb7aa52 | 308 | l = i + n2; |
emilmont | 1:fdd22bb7aa52 | 309 | xt = (pSrc[2 * i] >> 2u) - (pSrc[2 * l] >> 2u); |
emilmont | 1:fdd22bb7aa52 | 310 | pSrc[2 * i] = ((pSrc[2 * i] >> 2u) + (pSrc[2 * l] >> 2u)) >> 1u; |
emilmont | 1:fdd22bb7aa52 | 311 | |
emilmont | 1:fdd22bb7aa52 | 312 | yt = (pSrc[2 * i + 1] >> 2u) - (pSrc[2 * l + 1] >> 2u); |
emilmont | 1:fdd22bb7aa52 | 313 | pSrc[2 * i + 1] = |
emilmont | 1:fdd22bb7aa52 | 314 | ((pSrc[2 * l + 1] >> 2u) + (pSrc[2 * i + 1] >> 2u)) >> 1u; |
emilmont | 1:fdd22bb7aa52 | 315 | |
emilmont | 1:fdd22bb7aa52 | 316 | pSrc[2u * l] = (((int16_t) (((q31_t) xt * cosVal) >> 16)) + |
emilmont | 1:fdd22bb7aa52 | 317 | ((int16_t) (((q31_t) yt * sinVal) >> 16))); |
emilmont | 1:fdd22bb7aa52 | 318 | |
emilmont | 1:fdd22bb7aa52 | 319 | pSrc[2u * l + 1u] = (((int16_t) (((q31_t) yt * cosVal) >> 16)) - |
emilmont | 1:fdd22bb7aa52 | 320 | ((int16_t) (((q31_t) xt * sinVal) >> 16))); |
emilmont | 1:fdd22bb7aa52 | 321 | |
emilmont | 1:fdd22bb7aa52 | 322 | } // butterfly loop end |
emilmont | 1:fdd22bb7aa52 | 323 | |
emilmont | 1:fdd22bb7aa52 | 324 | } // groups loop end |
emilmont | 1:fdd22bb7aa52 | 325 | |
emilmont | 1:fdd22bb7aa52 | 326 | twidCoefModifier = twidCoefModifier << 1u; |
emilmont | 1:fdd22bb7aa52 | 327 | |
emilmont | 1:fdd22bb7aa52 | 328 | // loop for stage |
emilmont | 1:fdd22bb7aa52 | 329 | for (k = fftLen / 2; k > 2; k = k >> 1) |
emilmont | 1:fdd22bb7aa52 | 330 | { |
emilmont | 1:fdd22bb7aa52 | 331 | n1 = n2; |
emilmont | 1:fdd22bb7aa52 | 332 | n2 = n2 >> 1; |
emilmont | 1:fdd22bb7aa52 | 333 | ia = 0; |
emilmont | 1:fdd22bb7aa52 | 334 | |
emilmont | 1:fdd22bb7aa52 | 335 | // loop for groups |
emilmont | 1:fdd22bb7aa52 | 336 | for (j = 0; j < n2; j++) |
emilmont | 1:fdd22bb7aa52 | 337 | { |
emilmont | 1:fdd22bb7aa52 | 338 | cosVal = pCoef[ia * 2]; |
emilmont | 1:fdd22bb7aa52 | 339 | sinVal = pCoef[(ia * 2) + 1]; |
emilmont | 1:fdd22bb7aa52 | 340 | ia = ia + twidCoefModifier; |
emilmont | 1:fdd22bb7aa52 | 341 | |
emilmont | 1:fdd22bb7aa52 | 342 | // loop for butterfly |
emilmont | 1:fdd22bb7aa52 | 343 | for (i = j; i < fftLen; i += n1) |
emilmont | 1:fdd22bb7aa52 | 344 | { |
emilmont | 1:fdd22bb7aa52 | 345 | l = i + n2; |
emilmont | 1:fdd22bb7aa52 | 346 | xt = pSrc[2 * i] - pSrc[2 * l]; |
emilmont | 1:fdd22bb7aa52 | 347 | pSrc[2 * i] = (pSrc[2 * i] + pSrc[2 * l]) >> 1u; |
emilmont | 1:fdd22bb7aa52 | 348 | |
emilmont | 1:fdd22bb7aa52 | 349 | yt = pSrc[2 * i + 1] - pSrc[2 * l + 1]; |
emilmont | 1:fdd22bb7aa52 | 350 | pSrc[2 * i + 1] = (pSrc[2 * l + 1] + pSrc[2 * i + 1]) >> 1u; |
emilmont | 1:fdd22bb7aa52 | 351 | |
emilmont | 1:fdd22bb7aa52 | 352 | pSrc[2u * l] = (((int16_t) (((q31_t) xt * cosVal) >> 16)) + |
emilmont | 1:fdd22bb7aa52 | 353 | ((int16_t) (((q31_t) yt * sinVal) >> 16))); |
emilmont | 1:fdd22bb7aa52 | 354 | |
emilmont | 1:fdd22bb7aa52 | 355 | pSrc[2u * l + 1u] = (((int16_t) (((q31_t) yt * cosVal) >> 16)) - |
emilmont | 1:fdd22bb7aa52 | 356 | ((int16_t) (((q31_t) xt * sinVal) >> 16))); |
emilmont | 1:fdd22bb7aa52 | 357 | |
emilmont | 1:fdd22bb7aa52 | 358 | } // butterfly loop end |
emilmont | 1:fdd22bb7aa52 | 359 | |
emilmont | 1:fdd22bb7aa52 | 360 | } // groups loop end |
emilmont | 1:fdd22bb7aa52 | 361 | |
emilmont | 1:fdd22bb7aa52 | 362 | twidCoefModifier = twidCoefModifier << 1u; |
emilmont | 1:fdd22bb7aa52 | 363 | } // stages loop end |
emilmont | 1:fdd22bb7aa52 | 364 | |
emilmont | 1:fdd22bb7aa52 | 365 | n1 = n2; |
emilmont | 1:fdd22bb7aa52 | 366 | n2 = n2 >> 1; |
emilmont | 1:fdd22bb7aa52 | 367 | ia = 0; |
emilmont | 1:fdd22bb7aa52 | 368 | |
emilmont | 1:fdd22bb7aa52 | 369 | // loop for groups |
emilmont | 1:fdd22bb7aa52 | 370 | for (j = 0; j < n2; j++) |
emilmont | 1:fdd22bb7aa52 | 371 | { |
emilmont | 1:fdd22bb7aa52 | 372 | cosVal = pCoef[ia * 2]; |
emilmont | 1:fdd22bb7aa52 | 373 | sinVal = pCoef[(ia * 2) + 1]; |
emilmont | 1:fdd22bb7aa52 | 374 | |
emilmont | 1:fdd22bb7aa52 | 375 | ia = ia + twidCoefModifier; |
emilmont | 1:fdd22bb7aa52 | 376 | |
emilmont | 1:fdd22bb7aa52 | 377 | // loop for butterfly |
emilmont | 1:fdd22bb7aa52 | 378 | for (i = j; i < fftLen; i += n1) |
emilmont | 1:fdd22bb7aa52 | 379 | { |
emilmont | 1:fdd22bb7aa52 | 380 | l = i + n2; |
emilmont | 1:fdd22bb7aa52 | 381 | xt = pSrc[2 * i] - pSrc[2 * l]; |
emilmont | 1:fdd22bb7aa52 | 382 | pSrc[2 * i] = (pSrc[2 * i] + pSrc[2 * l]); |
emilmont | 1:fdd22bb7aa52 | 383 | |
emilmont | 1:fdd22bb7aa52 | 384 | yt = pSrc[2 * i + 1] - pSrc[2 * l + 1]; |
emilmont | 1:fdd22bb7aa52 | 385 | pSrc[2 * i + 1] = (pSrc[2 * l + 1] + pSrc[2 * i + 1]); |
emilmont | 1:fdd22bb7aa52 | 386 | |
emilmont | 1:fdd22bb7aa52 | 387 | pSrc[2u * l] = xt; |
emilmont | 1:fdd22bb7aa52 | 388 | |
emilmont | 1:fdd22bb7aa52 | 389 | pSrc[2u * l + 1u] = yt; |
emilmont | 1:fdd22bb7aa52 | 390 | |
emilmont | 1:fdd22bb7aa52 | 391 | } // butterfly loop end |
emilmont | 1:fdd22bb7aa52 | 392 | |
emilmont | 1:fdd22bb7aa52 | 393 | } // groups loop end |
emilmont | 1:fdd22bb7aa52 | 394 | |
emilmont | 1:fdd22bb7aa52 | 395 | twidCoefModifier = twidCoefModifier << 1u; |
emilmont | 1:fdd22bb7aa52 | 396 | |
emilmont | 1:fdd22bb7aa52 | 397 | #endif // #ifndef ARM_MATH_CM0 |
emilmont | 1:fdd22bb7aa52 | 398 | |
emilmont | 1:fdd22bb7aa52 | 399 | } |
emilmont | 1:fdd22bb7aa52 | 400 | |
emilmont | 1:fdd22bb7aa52 | 401 | |
emilmont | 1:fdd22bb7aa52 | 402 | void arm_radix2_butterfly_inverse_q15( |
emilmont | 1:fdd22bb7aa52 | 403 | q15_t * pSrc, |
emilmont | 1:fdd22bb7aa52 | 404 | uint32_t fftLen, |
emilmont | 1:fdd22bb7aa52 | 405 | q15_t * pCoef, |
emilmont | 1:fdd22bb7aa52 | 406 | uint16_t twidCoefModifier) |
emilmont | 1:fdd22bb7aa52 | 407 | { |
emilmont | 1:fdd22bb7aa52 | 408 | #ifndef ARM_MATH_CM0 |
emilmont | 1:fdd22bb7aa52 | 409 | |
emilmont | 1:fdd22bb7aa52 | 410 | int i, j, k, l; |
emilmont | 1:fdd22bb7aa52 | 411 | int n1, n2, ia; |
emilmont | 1:fdd22bb7aa52 | 412 | q15_t in; |
emilmont | 1:fdd22bb7aa52 | 413 | q31_t T, S, R; |
emilmont | 1:fdd22bb7aa52 | 414 | q31_t coeff, out1, out2; |
emilmont | 1:fdd22bb7aa52 | 415 | |
emilmont | 1:fdd22bb7aa52 | 416 | //N = fftLen; |
emilmont | 1:fdd22bb7aa52 | 417 | n2 = fftLen; |
emilmont | 1:fdd22bb7aa52 | 418 | |
emilmont | 1:fdd22bb7aa52 | 419 | n1 = n2; |
emilmont | 1:fdd22bb7aa52 | 420 | n2 = n2 >> 1; |
emilmont | 1:fdd22bb7aa52 | 421 | ia = 0; |
emilmont | 1:fdd22bb7aa52 | 422 | |
emilmont | 1:fdd22bb7aa52 | 423 | // loop for groups |
emilmont | 1:fdd22bb7aa52 | 424 | for (i = 0; i < n2; i++) |
emilmont | 1:fdd22bb7aa52 | 425 | { |
emilmont | 1:fdd22bb7aa52 | 426 | coeff = _SIMD32_OFFSET(pCoef + (ia * 2u)); |
emilmont | 1:fdd22bb7aa52 | 427 | |
emilmont | 1:fdd22bb7aa52 | 428 | ia = ia + twidCoefModifier; |
emilmont | 1:fdd22bb7aa52 | 429 | |
emilmont | 1:fdd22bb7aa52 | 430 | l = i + n2; |
emilmont | 1:fdd22bb7aa52 | 431 | |
emilmont | 1:fdd22bb7aa52 | 432 | T = _SIMD32_OFFSET(pSrc + (2 * i)); |
emilmont | 1:fdd22bb7aa52 | 433 | in = ((int16_t) (T & 0xFFFF)) >> 2; |
emilmont | 1:fdd22bb7aa52 | 434 | T = ((T >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
emilmont | 1:fdd22bb7aa52 | 435 | |
emilmont | 1:fdd22bb7aa52 | 436 | S = _SIMD32_OFFSET(pSrc + (2 * l)); |
emilmont | 1:fdd22bb7aa52 | 437 | in = ((int16_t) (S & 0xFFFF)) >> 2; |
emilmont | 1:fdd22bb7aa52 | 438 | S = ((S >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
emilmont | 1:fdd22bb7aa52 | 439 | |
emilmont | 1:fdd22bb7aa52 | 440 | R = __QSUB16(T, S); |
emilmont | 1:fdd22bb7aa52 | 441 | |
emilmont | 1:fdd22bb7aa52 | 442 | _SIMD32_OFFSET(pSrc + (2 * i)) = __SHADD16(T, S); |
emilmont | 1:fdd22bb7aa52 | 443 | |
emilmont | 1:fdd22bb7aa52 | 444 | #ifndef ARM_MATH_BIG_ENDIAN |
emilmont | 1:fdd22bb7aa52 | 445 | |
emilmont | 1:fdd22bb7aa52 | 446 | out1 = __SMUSD(coeff, R) >> 16; |
emilmont | 1:fdd22bb7aa52 | 447 | out2 = __SMUADX(coeff, R); |
emilmont | 1:fdd22bb7aa52 | 448 | #else |
emilmont | 1:fdd22bb7aa52 | 449 | |
emilmont | 1:fdd22bb7aa52 | 450 | out1 = __SMUADX(R, coeff) >> 16u; |
emilmont | 1:fdd22bb7aa52 | 451 | out2 = __SMUSD(__QSUB(0, coeff), R); |
emilmont | 1:fdd22bb7aa52 | 452 | |
emilmont | 1:fdd22bb7aa52 | 453 | #endif // #ifndef ARM_MATH_BIG_ENDIAN |
emilmont | 1:fdd22bb7aa52 | 454 | |
emilmont | 1:fdd22bb7aa52 | 455 | _SIMD32_OFFSET(pSrc + (2u * l)) = |
emilmont | 1:fdd22bb7aa52 | 456 | (q31_t) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
emilmont | 1:fdd22bb7aa52 | 457 | |
emilmont | 1:fdd22bb7aa52 | 458 | coeff = _SIMD32_OFFSET(pCoef + (ia * 2u)); |
emilmont | 1:fdd22bb7aa52 | 459 | |
emilmont | 1:fdd22bb7aa52 | 460 | ia = ia + twidCoefModifier; |
emilmont | 1:fdd22bb7aa52 | 461 | |
emilmont | 1:fdd22bb7aa52 | 462 | // loop for butterfly |
emilmont | 1:fdd22bb7aa52 | 463 | i++; |
emilmont | 1:fdd22bb7aa52 | 464 | l++; |
emilmont | 1:fdd22bb7aa52 | 465 | |
emilmont | 1:fdd22bb7aa52 | 466 | T = _SIMD32_OFFSET(pSrc + (2 * i)); |
emilmont | 1:fdd22bb7aa52 | 467 | in = ((int16_t) (T & 0xFFFF)) >> 2; |
emilmont | 1:fdd22bb7aa52 | 468 | T = ((T >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
emilmont | 1:fdd22bb7aa52 | 469 | |
emilmont | 1:fdd22bb7aa52 | 470 | S = _SIMD32_OFFSET(pSrc + (2 * l)); |
emilmont | 1:fdd22bb7aa52 | 471 | in = ((int16_t) (S & 0xFFFF)) >> 2; |
emilmont | 1:fdd22bb7aa52 | 472 | S = ((S >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
emilmont | 1:fdd22bb7aa52 | 473 | |
emilmont | 1:fdd22bb7aa52 | 474 | R = __QSUB16(T, S); |
emilmont | 1:fdd22bb7aa52 | 475 | |
emilmont | 1:fdd22bb7aa52 | 476 | _SIMD32_OFFSET(pSrc + (2 * i)) = __SHADD16(T, S); |
emilmont | 1:fdd22bb7aa52 | 477 | |
emilmont | 1:fdd22bb7aa52 | 478 | #ifndef ARM_MATH_BIG_ENDIAN |
emilmont | 1:fdd22bb7aa52 | 479 | |
emilmont | 1:fdd22bb7aa52 | 480 | out1 = __SMUSD(coeff, R) >> 16; |
emilmont | 1:fdd22bb7aa52 | 481 | out2 = __SMUADX(coeff, R); |
emilmont | 1:fdd22bb7aa52 | 482 | #else |
emilmont | 1:fdd22bb7aa52 | 483 | |
emilmont | 1:fdd22bb7aa52 | 484 | out1 = __SMUADX(R, coeff) >> 16u; |
emilmont | 1:fdd22bb7aa52 | 485 | out2 = __SMUSD(__QSUB(0, coeff), R); |
emilmont | 1:fdd22bb7aa52 | 486 | |
emilmont | 1:fdd22bb7aa52 | 487 | #endif // #ifndef ARM_MATH_BIG_ENDIAN |
emilmont | 1:fdd22bb7aa52 | 488 | |
emilmont | 1:fdd22bb7aa52 | 489 | _SIMD32_OFFSET(pSrc + (2u * l)) = |
emilmont | 1:fdd22bb7aa52 | 490 | (q31_t) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
emilmont | 1:fdd22bb7aa52 | 491 | |
emilmont | 1:fdd22bb7aa52 | 492 | } // groups loop end |
emilmont | 1:fdd22bb7aa52 | 493 | |
emilmont | 1:fdd22bb7aa52 | 494 | twidCoefModifier = twidCoefModifier << 1u; |
emilmont | 1:fdd22bb7aa52 | 495 | |
emilmont | 1:fdd22bb7aa52 | 496 | // loop for stage |
emilmont | 1:fdd22bb7aa52 | 497 | for (k = fftLen / 2; k > 2; k = k >> 1) |
emilmont | 1:fdd22bb7aa52 | 498 | { |
emilmont | 1:fdd22bb7aa52 | 499 | n1 = n2; |
emilmont | 1:fdd22bb7aa52 | 500 | n2 = n2 >> 1; |
emilmont | 1:fdd22bb7aa52 | 501 | ia = 0; |
emilmont | 1:fdd22bb7aa52 | 502 | |
emilmont | 1:fdd22bb7aa52 | 503 | // loop for groups |
emilmont | 1:fdd22bb7aa52 | 504 | for (j = 0; j < n2; j++) |
emilmont | 1:fdd22bb7aa52 | 505 | { |
emilmont | 1:fdd22bb7aa52 | 506 | coeff = _SIMD32_OFFSET(pCoef + (ia * 2u)); |
emilmont | 1:fdd22bb7aa52 | 507 | |
emilmont | 1:fdd22bb7aa52 | 508 | ia = ia + twidCoefModifier; |
emilmont | 1:fdd22bb7aa52 | 509 | |
emilmont | 1:fdd22bb7aa52 | 510 | // loop for butterfly |
emilmont | 1:fdd22bb7aa52 | 511 | for (i = j; i < fftLen; i += n1) |
emilmont | 1:fdd22bb7aa52 | 512 | { |
emilmont | 1:fdd22bb7aa52 | 513 | l = i + n2; |
emilmont | 1:fdd22bb7aa52 | 514 | |
emilmont | 1:fdd22bb7aa52 | 515 | T = _SIMD32_OFFSET(pSrc + (2 * i)); |
emilmont | 1:fdd22bb7aa52 | 516 | |
emilmont | 1:fdd22bb7aa52 | 517 | S = _SIMD32_OFFSET(pSrc + (2 * l)); |
emilmont | 1:fdd22bb7aa52 | 518 | |
emilmont | 1:fdd22bb7aa52 | 519 | R = __QSUB16(T, S); |
emilmont | 1:fdd22bb7aa52 | 520 | |
emilmont | 1:fdd22bb7aa52 | 521 | _SIMD32_OFFSET(pSrc + (2 * i)) = __SHADD16(T, S); |
emilmont | 1:fdd22bb7aa52 | 522 | |
emilmont | 1:fdd22bb7aa52 | 523 | #ifndef ARM_MATH_BIG_ENDIAN |
emilmont | 1:fdd22bb7aa52 | 524 | |
emilmont | 1:fdd22bb7aa52 | 525 | out1 = __SMUSD(coeff, R) >> 16; |
emilmont | 1:fdd22bb7aa52 | 526 | out2 = __SMUADX(coeff, R); |
emilmont | 1:fdd22bb7aa52 | 527 | |
emilmont | 1:fdd22bb7aa52 | 528 | #else |
emilmont | 1:fdd22bb7aa52 | 529 | |
emilmont | 1:fdd22bb7aa52 | 530 | out1 = __SMUADX(R, coeff) >> 16u; |
emilmont | 1:fdd22bb7aa52 | 531 | out2 = __SMUSD(__QSUB(0, coeff), R); |
emilmont | 1:fdd22bb7aa52 | 532 | |
emilmont | 1:fdd22bb7aa52 | 533 | #endif // #ifndef ARM_MATH_BIG_ENDIAN |
emilmont | 1:fdd22bb7aa52 | 534 | |
emilmont | 1:fdd22bb7aa52 | 535 | _SIMD32_OFFSET(pSrc + (2u * l)) = |
emilmont | 1:fdd22bb7aa52 | 536 | (q31_t) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
emilmont | 1:fdd22bb7aa52 | 537 | |
emilmont | 1:fdd22bb7aa52 | 538 | i += n1; |
emilmont | 1:fdd22bb7aa52 | 539 | |
emilmont | 1:fdd22bb7aa52 | 540 | l = i + n2; |
emilmont | 1:fdd22bb7aa52 | 541 | |
emilmont | 1:fdd22bb7aa52 | 542 | T = _SIMD32_OFFSET(pSrc + (2 * i)); |
emilmont | 1:fdd22bb7aa52 | 543 | |
emilmont | 1:fdd22bb7aa52 | 544 | S = _SIMD32_OFFSET(pSrc + (2 * l)); |
emilmont | 1:fdd22bb7aa52 | 545 | |
emilmont | 1:fdd22bb7aa52 | 546 | R = __QSUB16(T, S); |
emilmont | 1:fdd22bb7aa52 | 547 | |
emilmont | 1:fdd22bb7aa52 | 548 | _SIMD32_OFFSET(pSrc + (2 * i)) = __SHADD16(T, S); |
emilmont | 1:fdd22bb7aa52 | 549 | |
emilmont | 1:fdd22bb7aa52 | 550 | #ifndef ARM_MATH_BIG_ENDIAN |
emilmont | 1:fdd22bb7aa52 | 551 | |
emilmont | 1:fdd22bb7aa52 | 552 | out1 = __SMUSD(coeff, R) >> 16; |
emilmont | 1:fdd22bb7aa52 | 553 | out2 = __SMUADX(coeff, R); |
emilmont | 1:fdd22bb7aa52 | 554 | #else |
emilmont | 1:fdd22bb7aa52 | 555 | |
emilmont | 1:fdd22bb7aa52 | 556 | out1 = __SMUADX(R, coeff) >> 16u; |
emilmont | 1:fdd22bb7aa52 | 557 | out2 = __SMUSD(__QSUB(0, coeff), R); |
emilmont | 1:fdd22bb7aa52 | 558 | |
emilmont | 1:fdd22bb7aa52 | 559 | #endif // #ifndef ARM_MATH_BIG_ENDIAN |
emilmont | 1:fdd22bb7aa52 | 560 | |
emilmont | 1:fdd22bb7aa52 | 561 | _SIMD32_OFFSET(pSrc + (2u * l)) = |
emilmont | 1:fdd22bb7aa52 | 562 | (q31_t) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
emilmont | 1:fdd22bb7aa52 | 563 | |
emilmont | 1:fdd22bb7aa52 | 564 | } // butterfly loop end |
emilmont | 1:fdd22bb7aa52 | 565 | |
emilmont | 1:fdd22bb7aa52 | 566 | } // groups loop end |
emilmont | 1:fdd22bb7aa52 | 567 | |
emilmont | 1:fdd22bb7aa52 | 568 | twidCoefModifier = twidCoefModifier << 1u; |
emilmont | 1:fdd22bb7aa52 | 569 | } // stages loop end |
emilmont | 1:fdd22bb7aa52 | 570 | |
emilmont | 1:fdd22bb7aa52 | 571 | n1 = n2; |
emilmont | 1:fdd22bb7aa52 | 572 | n2 = n2 >> 1; |
emilmont | 1:fdd22bb7aa52 | 573 | ia = 0; |
emilmont | 1:fdd22bb7aa52 | 574 | |
emilmont | 1:fdd22bb7aa52 | 575 | // loop for groups |
emilmont | 1:fdd22bb7aa52 | 576 | for (j = 0; j < n2; j++) |
emilmont | 1:fdd22bb7aa52 | 577 | { |
emilmont | 1:fdd22bb7aa52 | 578 | coeff = _SIMD32_OFFSET(pCoef + (ia * 2u)); |
emilmont | 1:fdd22bb7aa52 | 579 | |
emilmont | 1:fdd22bb7aa52 | 580 | ia = ia + twidCoefModifier; |
emilmont | 1:fdd22bb7aa52 | 581 | |
emilmont | 1:fdd22bb7aa52 | 582 | // loop for butterfly |
emilmont | 1:fdd22bb7aa52 | 583 | for (i = j; i < fftLen; i += n1) |
emilmont | 1:fdd22bb7aa52 | 584 | { |
emilmont | 1:fdd22bb7aa52 | 585 | l = i + n2; |
emilmont | 1:fdd22bb7aa52 | 586 | |
emilmont | 1:fdd22bb7aa52 | 587 | T = _SIMD32_OFFSET(pSrc + (2 * i)); |
emilmont | 1:fdd22bb7aa52 | 588 | |
emilmont | 1:fdd22bb7aa52 | 589 | S = _SIMD32_OFFSET(pSrc + (2 * l)); |
emilmont | 1:fdd22bb7aa52 | 590 | |
emilmont | 1:fdd22bb7aa52 | 591 | R = __QSUB16(T, S); |
emilmont | 1:fdd22bb7aa52 | 592 | |
emilmont | 1:fdd22bb7aa52 | 593 | _SIMD32_OFFSET(pSrc + (2 * i)) = __QADD16(T, S); |
emilmont | 1:fdd22bb7aa52 | 594 | |
emilmont | 1:fdd22bb7aa52 | 595 | _SIMD32_OFFSET(pSrc + (2u * l)) = R; |
emilmont | 1:fdd22bb7aa52 | 596 | |
emilmont | 1:fdd22bb7aa52 | 597 | } // butterfly loop end |
emilmont | 1:fdd22bb7aa52 | 598 | |
emilmont | 1:fdd22bb7aa52 | 599 | } // groups loop end |
emilmont | 1:fdd22bb7aa52 | 600 | |
emilmont | 1:fdd22bb7aa52 | 601 | twidCoefModifier = twidCoefModifier << 1u; |
emilmont | 1:fdd22bb7aa52 | 602 | |
emilmont | 1:fdd22bb7aa52 | 603 | #else |
emilmont | 1:fdd22bb7aa52 | 604 | |
emilmont | 1:fdd22bb7aa52 | 605 | |
emilmont | 1:fdd22bb7aa52 | 606 | int i, j, k, l; |
emilmont | 1:fdd22bb7aa52 | 607 | int n1, n2, ia; |
emilmont | 1:fdd22bb7aa52 | 608 | q15_t xt, yt, cosVal, sinVal; |
emilmont | 1:fdd22bb7aa52 | 609 | |
emilmont | 1:fdd22bb7aa52 | 610 | //N = fftLen; |
emilmont | 1:fdd22bb7aa52 | 611 | n2 = fftLen; |
emilmont | 1:fdd22bb7aa52 | 612 | |
emilmont | 1:fdd22bb7aa52 | 613 | n1 = n2; |
emilmont | 1:fdd22bb7aa52 | 614 | n2 = n2 >> 1; |
emilmont | 1:fdd22bb7aa52 | 615 | ia = 0; |
emilmont | 1:fdd22bb7aa52 | 616 | |
emilmont | 1:fdd22bb7aa52 | 617 | // loop for groups |
emilmont | 1:fdd22bb7aa52 | 618 | for (j = 0; j < n2; j++) |
emilmont | 1:fdd22bb7aa52 | 619 | { |
emilmont | 1:fdd22bb7aa52 | 620 | cosVal = pCoef[ia * 2]; |
emilmont | 1:fdd22bb7aa52 | 621 | sinVal = pCoef[(ia * 2) + 1]; |
emilmont | 1:fdd22bb7aa52 | 622 | ia = ia + twidCoefModifier; |
emilmont | 1:fdd22bb7aa52 | 623 | |
emilmont | 1:fdd22bb7aa52 | 624 | // loop for butterfly |
emilmont | 1:fdd22bb7aa52 | 625 | for (i = j; i < fftLen; i += n1) |
emilmont | 1:fdd22bb7aa52 | 626 | { |
emilmont | 1:fdd22bb7aa52 | 627 | l = i + n2; |
emilmont | 1:fdd22bb7aa52 | 628 | xt = (pSrc[2 * i] >> 2u) - (pSrc[2 * l] >> 2u); |
emilmont | 1:fdd22bb7aa52 | 629 | pSrc[2 * i] = ((pSrc[2 * i] >> 2u) + (pSrc[2 * l] >> 2u)) >> 1u; |
emilmont | 1:fdd22bb7aa52 | 630 | |
emilmont | 1:fdd22bb7aa52 | 631 | yt = (pSrc[2 * i + 1] >> 2u) - (pSrc[2 * l + 1] >> 2u); |
emilmont | 1:fdd22bb7aa52 | 632 | pSrc[2 * i + 1] = |
emilmont | 1:fdd22bb7aa52 | 633 | ((pSrc[2 * l + 1] >> 2u) + (pSrc[2 * i + 1] >> 2u)) >> 1u; |
emilmont | 1:fdd22bb7aa52 | 634 | |
emilmont | 1:fdd22bb7aa52 | 635 | pSrc[2u * l] = (((int16_t) (((q31_t) xt * cosVal) >> 16)) - |
emilmont | 1:fdd22bb7aa52 | 636 | ((int16_t) (((q31_t) yt * sinVal) >> 16))); |
emilmont | 1:fdd22bb7aa52 | 637 | |
emilmont | 1:fdd22bb7aa52 | 638 | pSrc[2u * l + 1u] = (((int16_t) (((q31_t) yt * cosVal) >> 16)) + |
emilmont | 1:fdd22bb7aa52 | 639 | ((int16_t) (((q31_t) xt * sinVal) >> 16))); |
emilmont | 1:fdd22bb7aa52 | 640 | |
emilmont | 1:fdd22bb7aa52 | 641 | } // butterfly loop end |
emilmont | 1:fdd22bb7aa52 | 642 | |
emilmont | 1:fdd22bb7aa52 | 643 | } // groups loop end |
emilmont | 1:fdd22bb7aa52 | 644 | |
emilmont | 1:fdd22bb7aa52 | 645 | twidCoefModifier = twidCoefModifier << 1u; |
emilmont | 1:fdd22bb7aa52 | 646 | |
emilmont | 1:fdd22bb7aa52 | 647 | // loop for stage |
emilmont | 1:fdd22bb7aa52 | 648 | for (k = fftLen / 2; k > 2; k = k >> 1) |
emilmont | 1:fdd22bb7aa52 | 649 | { |
emilmont | 1:fdd22bb7aa52 | 650 | n1 = n2; |
emilmont | 1:fdd22bb7aa52 | 651 | n2 = n2 >> 1; |
emilmont | 1:fdd22bb7aa52 | 652 | ia = 0; |
emilmont | 1:fdd22bb7aa52 | 653 | |
emilmont | 1:fdd22bb7aa52 | 654 | // loop for groups |
emilmont | 1:fdd22bb7aa52 | 655 | for (j = 0; j < n2; j++) |
emilmont | 1:fdd22bb7aa52 | 656 | { |
emilmont | 1:fdd22bb7aa52 | 657 | cosVal = pCoef[ia * 2]; |
emilmont | 1:fdd22bb7aa52 | 658 | sinVal = pCoef[(ia * 2) + 1]; |
emilmont | 1:fdd22bb7aa52 | 659 | ia = ia + twidCoefModifier; |
emilmont | 1:fdd22bb7aa52 | 660 | |
emilmont | 1:fdd22bb7aa52 | 661 | // loop for butterfly |
emilmont | 1:fdd22bb7aa52 | 662 | for (i = j; i < fftLen; i += n1) |
emilmont | 1:fdd22bb7aa52 | 663 | { |
emilmont | 1:fdd22bb7aa52 | 664 | l = i + n2; |
emilmont | 1:fdd22bb7aa52 | 665 | xt = pSrc[2 * i] - pSrc[2 * l]; |
emilmont | 1:fdd22bb7aa52 | 666 | pSrc[2 * i] = (pSrc[2 * i] + pSrc[2 * l]) >> 1u; |
emilmont | 1:fdd22bb7aa52 | 667 | |
emilmont | 1:fdd22bb7aa52 | 668 | yt = pSrc[2 * i + 1] - pSrc[2 * l + 1]; |
emilmont | 1:fdd22bb7aa52 | 669 | pSrc[2 * i + 1] = (pSrc[2 * l + 1] + pSrc[2 * i + 1]) >> 1u; |
emilmont | 1:fdd22bb7aa52 | 670 | |
emilmont | 1:fdd22bb7aa52 | 671 | pSrc[2u * l] = (((int16_t) (((q31_t) xt * cosVal) >> 16)) - |
emilmont | 1:fdd22bb7aa52 | 672 | ((int16_t) (((q31_t) yt * sinVal) >> 16))); |
emilmont | 1:fdd22bb7aa52 | 673 | |
emilmont | 1:fdd22bb7aa52 | 674 | pSrc[2u * l + 1u] = (((int16_t) (((q31_t) yt * cosVal) >> 16)) + |
emilmont | 1:fdd22bb7aa52 | 675 | ((int16_t) (((q31_t) xt * sinVal) >> 16))); |
emilmont | 1:fdd22bb7aa52 | 676 | |
emilmont | 1:fdd22bb7aa52 | 677 | } // butterfly loop end |
emilmont | 1:fdd22bb7aa52 | 678 | |
emilmont | 1:fdd22bb7aa52 | 679 | } // groups loop end |
emilmont | 1:fdd22bb7aa52 | 680 | |
emilmont | 1:fdd22bb7aa52 | 681 | twidCoefModifier = twidCoefModifier << 1u; |
emilmont | 1:fdd22bb7aa52 | 682 | } // stages loop end |
emilmont | 1:fdd22bb7aa52 | 683 | |
emilmont | 1:fdd22bb7aa52 | 684 | n1 = n2; |
emilmont | 1:fdd22bb7aa52 | 685 | n2 = n2 >> 1; |
emilmont | 1:fdd22bb7aa52 | 686 | ia = 0; |
emilmont | 1:fdd22bb7aa52 | 687 | |
emilmont | 1:fdd22bb7aa52 | 688 | cosVal = pCoef[ia * 2]; |
emilmont | 1:fdd22bb7aa52 | 689 | sinVal = pCoef[(ia * 2) + 1]; |
emilmont | 1:fdd22bb7aa52 | 690 | |
emilmont | 1:fdd22bb7aa52 | 691 | ia = ia + twidCoefModifier; |
emilmont | 1:fdd22bb7aa52 | 692 | |
emilmont | 1:fdd22bb7aa52 | 693 | // loop for butterfly |
emilmont | 1:fdd22bb7aa52 | 694 | for (i = 0; i < fftLen; i += n1) |
emilmont | 1:fdd22bb7aa52 | 695 | { |
emilmont | 1:fdd22bb7aa52 | 696 | l = i + n2; |
emilmont | 1:fdd22bb7aa52 | 697 | xt = pSrc[2 * i] - pSrc[2 * l]; |
emilmont | 1:fdd22bb7aa52 | 698 | pSrc[2 * i] = (pSrc[2 * i] + pSrc[2 * l]); |
emilmont | 1:fdd22bb7aa52 | 699 | |
emilmont | 1:fdd22bb7aa52 | 700 | yt = pSrc[2 * i + 1] - pSrc[2 * l + 1]; |
emilmont | 1:fdd22bb7aa52 | 701 | pSrc[2 * i + 1] = (pSrc[2 * l + 1] + pSrc[2 * i + 1]); |
emilmont | 1:fdd22bb7aa52 | 702 | |
emilmont | 1:fdd22bb7aa52 | 703 | pSrc[2u * l] = xt; |
emilmont | 1:fdd22bb7aa52 | 704 | |
emilmont | 1:fdd22bb7aa52 | 705 | pSrc[2u * l + 1u] = yt; |
emilmont | 1:fdd22bb7aa52 | 706 | |
emilmont | 1:fdd22bb7aa52 | 707 | } // groups loop end |
emilmont | 1:fdd22bb7aa52 | 708 | |
emilmont | 1:fdd22bb7aa52 | 709 | |
emilmont | 1:fdd22bb7aa52 | 710 | #endif // #ifndef ARM_MATH_CM0 |
emilmont | 1:fdd22bb7aa52 | 711 | |
emilmont | 1:fdd22bb7aa52 | 712 | } |