mbed official
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_radix4_q15.c@5:3762170b6d4d, 2015-11-20 (annotated)
- Committer:
- mbed_official
- Date:
- Fri Nov 20 08:45:18 2015 +0000
- Revision:
- 5:3762170b6d4d
- Parent:
- 3:7a284390b0ce
Synchronized with git revision 2eb940b9a73af188d3004a2575fdfbb05febe62b
Full URL: https://github.com/mbedmicro/mbed/commit/2eb940b9a73af188d3004a2575fdfbb05febe62b/
Added option to build rpc library. closes #1426
Who changed what in which revision?
| User | Revision | Line number | New contents of line |
|---|---|---|---|
| emilmont | 1:fdd22bb7aa52 | 1 | /* ---------------------------------------------------------------------- |
| mbed_official | 5:3762170b6d4d | 2 | * Copyright (C) 2010-2014 ARM Limited. All rights reserved. |
| emilmont | 1:fdd22bb7aa52 | 3 | * |
| mbed_official | 5:3762170b6d4d | 4 | * $Date: 19. March 2015 |
| mbed_official | 5:3762170b6d4d | 5 | * $Revision: V.1.4.5 |
| emilmont | 1:fdd22bb7aa52 | 6 | * |
| emilmont | 2:da51fb522205 | 7 | * Project: CMSIS DSP Library |
| emilmont | 2:da51fb522205 | 8 | * Title: arm_cfft_radix4_q15.c |
| emilmont | 1:fdd22bb7aa52 | 9 | * |
| emilmont | 2:da51fb522205 | 10 | * Description: This file has function definition of Radix-4 FFT & IFFT function and |
| emilmont | 2:da51fb522205 | 11 | * In-place bit reversal using bit reversal table |
| emilmont | 1:fdd22bb7aa52 | 12 | * |
| emilmont | 1:fdd22bb7aa52 | 13 | * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0 |
| emilmont | 1:fdd22bb7aa52 | 14 | * |
| mbed_official | 3:7a284390b0ce | 15 | * Redistribution and use in source and binary forms, with or without |
| mbed_official | 3:7a284390b0ce | 16 | * modification, are permitted provided that the following conditions |
| mbed_official | 3:7a284390b0ce | 17 | * are met: |
| mbed_official | 3:7a284390b0ce | 18 | * - Redistributions of source code must retain the above copyright |
| mbed_official | 3:7a284390b0ce | 19 | * notice, this list of conditions and the following disclaimer. |
| mbed_official | 3:7a284390b0ce | 20 | * - Redistributions in binary form must reproduce the above copyright |
| mbed_official | 3:7a284390b0ce | 21 | * notice, this list of conditions and the following disclaimer in |
| mbed_official | 3:7a284390b0ce | 22 | * the documentation and/or other materials provided with the |
| mbed_official | 3:7a284390b0ce | 23 | * distribution. |
| mbed_official | 3:7a284390b0ce | 24 | * - Neither the name of ARM LIMITED nor the names of its contributors |
| mbed_official | 3:7a284390b0ce | 25 | * may be used to endorse or promote products derived from this |
| mbed_official | 3:7a284390b0ce | 26 | * software without specific prior written permission. |
| mbed_official | 3:7a284390b0ce | 27 | * |
| mbed_official | 3:7a284390b0ce | 28 | * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
| mbed_official | 3:7a284390b0ce | 29 | * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
| mbed_official | 3:7a284390b0ce | 30 | * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS |
| mbed_official | 3:7a284390b0ce | 31 | * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE |
| mbed_official | 3:7a284390b0ce | 32 | * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, |
| mbed_official | 3:7a284390b0ce | 33 | * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, |
| mbed_official | 3:7a284390b0ce | 34 | * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; |
| mbed_official | 3:7a284390b0ce | 35 | * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER |
| mbed_official | 3:7a284390b0ce | 36 | * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
| mbed_official | 3:7a284390b0ce | 37 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN |
| mbed_official | 3:7a284390b0ce | 38 | * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE |
| mbed_official | 3:7a284390b0ce | 39 | * POSSIBILITY OF SUCH DAMAGE. |
| emilmont | 1:fdd22bb7aa52 | 40 | * -------------------------------------------------------------------- */ |
| emilmont | 1:fdd22bb7aa52 | 41 | |
| emilmont | 1:fdd22bb7aa52 | 42 | #include "arm_math.h" |
| emilmont | 1:fdd22bb7aa52 | 43 | |
| mbed_official | 3:7a284390b0ce | 44 | |
| mbed_official | 3:7a284390b0ce | 45 | void arm_radix4_butterfly_q15( |
| mbed_official | 3:7a284390b0ce | 46 | q15_t * pSrc16, |
| mbed_official | 3:7a284390b0ce | 47 | uint32_t fftLen, |
| mbed_official | 3:7a284390b0ce | 48 | q15_t * pCoef16, |
| mbed_official | 3:7a284390b0ce | 49 | uint32_t twidCoefModifier); |
| mbed_official | 3:7a284390b0ce | 50 | |
| mbed_official | 3:7a284390b0ce | 51 | void arm_radix4_butterfly_inverse_q15( |
| mbed_official | 3:7a284390b0ce | 52 | q15_t * pSrc16, |
| mbed_official | 3:7a284390b0ce | 53 | uint32_t fftLen, |
| mbed_official | 3:7a284390b0ce | 54 | q15_t * pCoef16, |
| mbed_official | 3:7a284390b0ce | 55 | uint32_t twidCoefModifier); |
| mbed_official | 3:7a284390b0ce | 56 | |
| mbed_official | 3:7a284390b0ce | 57 | void arm_bitreversal_q15( |
| mbed_official | 3:7a284390b0ce | 58 | q15_t * pSrc, |
| mbed_official | 3:7a284390b0ce | 59 | uint32_t fftLen, |
| mbed_official | 3:7a284390b0ce | 60 | uint16_t bitRevFactor, |
| mbed_official | 3:7a284390b0ce | 61 | uint16_t * pBitRevTab); |
| mbed_official | 3:7a284390b0ce | 62 | |
| emilmont | 1:fdd22bb7aa52 | 63 | /** |
| emilmont | 1:fdd22bb7aa52 | 64 | * @ingroup groupTransforms |
| emilmont | 1:fdd22bb7aa52 | 65 | */ |
| emilmont | 1:fdd22bb7aa52 | 66 | |
| emilmont | 1:fdd22bb7aa52 | 67 | /** |
| mbed_official | 3:7a284390b0ce | 68 | * @addtogroup ComplexFFT |
| emilmont | 1:fdd22bb7aa52 | 69 | * @{ |
| emilmont | 1:fdd22bb7aa52 | 70 | */ |
| emilmont | 1:fdd22bb7aa52 | 71 | |
| emilmont | 1:fdd22bb7aa52 | 72 | |
| emilmont | 1:fdd22bb7aa52 | 73 | /** |
| emilmont | 1:fdd22bb7aa52 | 74 | * @details |
| emilmont | 1:fdd22bb7aa52 | 75 | * @brief Processing function for the Q15 CFFT/CIFFT. |
| mbed_official | 5:3762170b6d4d | 76 | * @deprecated Do not use this function. It has been superseded by \ref arm_cfft_q15 and will be removed |
| emilmont | 1:fdd22bb7aa52 | 77 | * @param[in] *S points to an instance of the Q15 CFFT/CIFFT structure. |
| emilmont | 1:fdd22bb7aa52 | 78 | * @param[in, out] *pSrc points to the complex data buffer. Processing occurs in-place. |
| emilmont | 1:fdd22bb7aa52 | 79 | * @return none. |
| emilmont | 1:fdd22bb7aa52 | 80 | * |
| emilmont | 1:fdd22bb7aa52 | 81 | * \par Input and output formats: |
| emilmont | 1:fdd22bb7aa52 | 82 | * \par |
| emilmont | 1:fdd22bb7aa52 | 83 | * Internally input is downscaled by 2 for every stage to avoid saturations inside CFFT/CIFFT process. |
| emilmont | 1:fdd22bb7aa52 | 84 | * Hence the output format is different for different FFT sizes. |
| emilmont | 1:fdd22bb7aa52 | 85 | * The input and output formats for different FFT sizes and number of bits to upscale are mentioned in the tables below for CFFT and CIFFT: |
| emilmont | 1:fdd22bb7aa52 | 86 | * \par |
| emilmont | 1:fdd22bb7aa52 | 87 | * \image html CFFTQ15.gif "Input and Output Formats for Q15 CFFT" |
| emilmont | 1:fdd22bb7aa52 | 88 | * \image html CIFFTQ15.gif "Input and Output Formats for Q15 CIFFT" |
| emilmont | 1:fdd22bb7aa52 | 89 | */ |
| emilmont | 1:fdd22bb7aa52 | 90 | |
| emilmont | 1:fdd22bb7aa52 | 91 | void arm_cfft_radix4_q15( |
| emilmont | 1:fdd22bb7aa52 | 92 | const arm_cfft_radix4_instance_q15 * S, |
| emilmont | 1:fdd22bb7aa52 | 93 | q15_t * pSrc) |
| emilmont | 1:fdd22bb7aa52 | 94 | { |
| emilmont | 1:fdd22bb7aa52 | 95 | if(S->ifftFlag == 1u) |
| emilmont | 1:fdd22bb7aa52 | 96 | { |
| emilmont | 1:fdd22bb7aa52 | 97 | /* Complex IFFT radix-4 */ |
| emilmont | 1:fdd22bb7aa52 | 98 | arm_radix4_butterfly_inverse_q15(pSrc, S->fftLen, S->pTwiddle, |
| emilmont | 1:fdd22bb7aa52 | 99 | S->twidCoefModifier); |
| emilmont | 1:fdd22bb7aa52 | 100 | } |
| emilmont | 1:fdd22bb7aa52 | 101 | else |
| emilmont | 1:fdd22bb7aa52 | 102 | { |
| emilmont | 1:fdd22bb7aa52 | 103 | /* Complex FFT radix-4 */ |
| emilmont | 1:fdd22bb7aa52 | 104 | arm_radix4_butterfly_q15(pSrc, S->fftLen, S->pTwiddle, |
| emilmont | 1:fdd22bb7aa52 | 105 | S->twidCoefModifier); |
| emilmont | 1:fdd22bb7aa52 | 106 | } |
| emilmont | 1:fdd22bb7aa52 | 107 | |
| emilmont | 1:fdd22bb7aa52 | 108 | if(S->bitReverseFlag == 1u) |
| emilmont | 1:fdd22bb7aa52 | 109 | { |
| emilmont | 1:fdd22bb7aa52 | 110 | /* Bit Reversal */ |
| emilmont | 1:fdd22bb7aa52 | 111 | arm_bitreversal_q15(pSrc, S->fftLen, S->bitRevFactor, S->pBitRevTable); |
| emilmont | 1:fdd22bb7aa52 | 112 | } |
| emilmont | 1:fdd22bb7aa52 | 113 | |
| emilmont | 1:fdd22bb7aa52 | 114 | } |
| emilmont | 1:fdd22bb7aa52 | 115 | |
| emilmont | 1:fdd22bb7aa52 | 116 | /** |
| mbed_official | 3:7a284390b0ce | 117 | * @} end of ComplexFFT group |
| emilmont | 1:fdd22bb7aa52 | 118 | */ |
| emilmont | 1:fdd22bb7aa52 | 119 | |
| emilmont | 1:fdd22bb7aa52 | 120 | /* |
| emilmont | 1:fdd22bb7aa52 | 121 | * Radix-4 FFT algorithm used is : |
| emilmont | 1:fdd22bb7aa52 | 122 | * |
| emilmont | 1:fdd22bb7aa52 | 123 | * Input real and imaginary data: |
| emilmont | 1:fdd22bb7aa52 | 124 | * x(n) = xa + j * ya |
| emilmont | 1:fdd22bb7aa52 | 125 | * x(n+N/4 ) = xb + j * yb |
| emilmont | 1:fdd22bb7aa52 | 126 | * x(n+N/2 ) = xc + j * yc |
| emilmont | 1:fdd22bb7aa52 | 127 | * x(n+3N 4) = xd + j * yd |
| emilmont | 1:fdd22bb7aa52 | 128 | * |
| emilmont | 1:fdd22bb7aa52 | 129 | * |
| emilmont | 1:fdd22bb7aa52 | 130 | * Output real and imaginary data: |
| emilmont | 1:fdd22bb7aa52 | 131 | * x(4r) = xa'+ j * ya' |
| emilmont | 1:fdd22bb7aa52 | 132 | * x(4r+1) = xb'+ j * yb' |
| emilmont | 1:fdd22bb7aa52 | 133 | * x(4r+2) = xc'+ j * yc' |
| emilmont | 1:fdd22bb7aa52 | 134 | * x(4r+3) = xd'+ j * yd' |
| emilmont | 1:fdd22bb7aa52 | 135 | * |
| emilmont | 1:fdd22bb7aa52 | 136 | * |
| emilmont | 1:fdd22bb7aa52 | 137 | * Twiddle factors for radix-4 FFT: |
| emilmont | 1:fdd22bb7aa52 | 138 | * Wn = co1 + j * (- si1) |
| emilmont | 1:fdd22bb7aa52 | 139 | * W2n = co2 + j * (- si2) |
| emilmont | 1:fdd22bb7aa52 | 140 | * W3n = co3 + j * (- si3) |
| emilmont | 1:fdd22bb7aa52 | 141 | |
| emilmont | 1:fdd22bb7aa52 | 142 | * The real and imaginary output values for the radix-4 butterfly are |
| emilmont | 1:fdd22bb7aa52 | 143 | * xa' = xa + xb + xc + xd |
| emilmont | 1:fdd22bb7aa52 | 144 | * ya' = ya + yb + yc + yd |
| emilmont | 1:fdd22bb7aa52 | 145 | * xb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) |
| emilmont | 1:fdd22bb7aa52 | 146 | * yb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) |
| emilmont | 1:fdd22bb7aa52 | 147 | * xc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) |
| emilmont | 1:fdd22bb7aa52 | 148 | * yc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) |
| emilmont | 1:fdd22bb7aa52 | 149 | * xd' = (xa-yb-xc+yd)* co3 + (ya+xb-yc-xd)* (si3) |
| emilmont | 1:fdd22bb7aa52 | 150 | * yd' = (ya+xb-yc-xd)* co3 - (xa-yb-xc+yd)* (si3) |
| emilmont | 1:fdd22bb7aa52 | 151 | * |
| emilmont | 1:fdd22bb7aa52 | 152 | */ |
| emilmont | 1:fdd22bb7aa52 | 153 | |
| emilmont | 1:fdd22bb7aa52 | 154 | /** |
| emilmont | 1:fdd22bb7aa52 | 155 | * @brief Core function for the Q15 CFFT butterfly process. |
| emilmont | 1:fdd22bb7aa52 | 156 | * @param[in, out] *pSrc16 points to the in-place buffer of Q15 data type. |
| emilmont | 1:fdd22bb7aa52 | 157 | * @param[in] fftLen length of the FFT. |
| emilmont | 1:fdd22bb7aa52 | 158 | * @param[in] *pCoef16 points to twiddle coefficient buffer. |
| emilmont | 1:fdd22bb7aa52 | 159 | * @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table. |
| emilmont | 1:fdd22bb7aa52 | 160 | * @return none. |
| emilmont | 1:fdd22bb7aa52 | 161 | */ |
| emilmont | 1:fdd22bb7aa52 | 162 | |
| emilmont | 1:fdd22bb7aa52 | 163 | void arm_radix4_butterfly_q15( |
| emilmont | 1:fdd22bb7aa52 | 164 | q15_t * pSrc16, |
| emilmont | 1:fdd22bb7aa52 | 165 | uint32_t fftLen, |
| emilmont | 1:fdd22bb7aa52 | 166 | q15_t * pCoef16, |
| emilmont | 1:fdd22bb7aa52 | 167 | uint32_t twidCoefModifier) |
| emilmont | 1:fdd22bb7aa52 | 168 | { |
| emilmont | 1:fdd22bb7aa52 | 169 | |
| mbed_official | 3:7a284390b0ce | 170 | #ifndef ARM_MATH_CM0_FAMILY |
| emilmont | 1:fdd22bb7aa52 | 171 | |
| emilmont | 1:fdd22bb7aa52 | 172 | /* Run the below code for Cortex-M4 and Cortex-M3 */ |
| emilmont | 1:fdd22bb7aa52 | 173 | |
| emilmont | 1:fdd22bb7aa52 | 174 | q31_t R, S, T, U; |
| emilmont | 1:fdd22bb7aa52 | 175 | q31_t C1, C2, C3, out1, out2; |
| mbed_official | 5:3762170b6d4d | 176 | uint32_t n1, n2, ic, i0, j, k; |
| emilmont | 1:fdd22bb7aa52 | 177 | |
| emilmont | 1:fdd22bb7aa52 | 178 | q15_t *ptr1; |
| mbed_official | 5:3762170b6d4d | 179 | q15_t *pSi0; |
| mbed_official | 5:3762170b6d4d | 180 | q15_t *pSi1; |
| mbed_official | 5:3762170b6d4d | 181 | q15_t *pSi2; |
| mbed_official | 5:3762170b6d4d | 182 | q15_t *pSi3; |
| emilmont | 1:fdd22bb7aa52 | 183 | |
| emilmont | 1:fdd22bb7aa52 | 184 | q31_t xaya, xbyb, xcyc, xdyd; |
| emilmont | 1:fdd22bb7aa52 | 185 | |
| emilmont | 1:fdd22bb7aa52 | 186 | /* Total process is divided into three stages */ |
| emilmont | 1:fdd22bb7aa52 | 187 | |
| emilmont | 1:fdd22bb7aa52 | 188 | /* process first stage, middle stages, & last stage */ |
| emilmont | 1:fdd22bb7aa52 | 189 | |
| emilmont | 1:fdd22bb7aa52 | 190 | /* Initializations for the first stage */ |
| emilmont | 1:fdd22bb7aa52 | 191 | n2 = fftLen; |
| emilmont | 1:fdd22bb7aa52 | 192 | n1 = n2; |
| emilmont | 1:fdd22bb7aa52 | 193 | |
| emilmont | 1:fdd22bb7aa52 | 194 | /* n2 = fftLen/4 */ |
| emilmont | 1:fdd22bb7aa52 | 195 | n2 >>= 2u; |
| emilmont | 1:fdd22bb7aa52 | 196 | |
| emilmont | 1:fdd22bb7aa52 | 197 | /* Index for twiddle coefficient */ |
| emilmont | 1:fdd22bb7aa52 | 198 | ic = 0u; |
| emilmont | 1:fdd22bb7aa52 | 199 | |
| emilmont | 1:fdd22bb7aa52 | 200 | /* Index for input read and output write */ |
| emilmont | 1:fdd22bb7aa52 | 201 | j = n2; |
| mbed_official | 5:3762170b6d4d | 202 | |
| mbed_official | 5:3762170b6d4d | 203 | pSi0 = pSrc16; |
| mbed_official | 5:3762170b6d4d | 204 | pSi1 = pSi0 + 2 * n2; |
| mbed_official | 5:3762170b6d4d | 205 | pSi2 = pSi1 + 2 * n2; |
| mbed_official | 5:3762170b6d4d | 206 | pSi3 = pSi2 + 2 * n2; |
| emilmont | 1:fdd22bb7aa52 | 207 | |
| emilmont | 1:fdd22bb7aa52 | 208 | /* Input is in 1.15(q15) format */ |
| emilmont | 1:fdd22bb7aa52 | 209 | |
| emilmont | 1:fdd22bb7aa52 | 210 | /* start of first stage process */ |
| emilmont | 1:fdd22bb7aa52 | 211 | do |
| emilmont | 1:fdd22bb7aa52 | 212 | { |
| emilmont | 1:fdd22bb7aa52 | 213 | /* Butterfly implementation */ |
| emilmont | 1:fdd22bb7aa52 | 214 | |
| emilmont | 1:fdd22bb7aa52 | 215 | /* Reading i0, i0+fftLen/2 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 216 | /* Read ya (real), xa(imag) input */ |
| mbed_official | 5:3762170b6d4d | 217 | T = _SIMD32_OFFSET(pSi0); |
| mbed_official | 5:3762170b6d4d | 218 | T = __SHADD16(T, 0); // this is just a SIMD arithmetic shift right by 1 |
| mbed_official | 5:3762170b6d4d | 219 | T = __SHADD16(T, 0); // it turns out doing this twice is 2 cycles, the alternative takes 3 cycles |
| mbed_official | 5:3762170b6d4d | 220 | //in = ((int16_t) (T & 0xFFFF)) >> 2; // alternative code that takes 3 cycles |
| mbed_official | 5:3762170b6d4d | 221 | //T = ((T >> 2) & 0xFFFF0000) | (in & 0xFFFF); |
| emilmont | 1:fdd22bb7aa52 | 222 | |
| emilmont | 1:fdd22bb7aa52 | 223 | /* Read yc (real), xc(imag) input */ |
| mbed_official | 5:3762170b6d4d | 224 | S = _SIMD32_OFFSET(pSi2); |
| mbed_official | 5:3762170b6d4d | 225 | S = __SHADD16(S, 0); |
| mbed_official | 5:3762170b6d4d | 226 | S = __SHADD16(S, 0); |
| emilmont | 1:fdd22bb7aa52 | 227 | |
| emilmont | 1:fdd22bb7aa52 | 228 | /* R = packed((ya + yc), (xa + xc) ) */ |
| emilmont | 1:fdd22bb7aa52 | 229 | R = __QADD16(T, S); |
| emilmont | 1:fdd22bb7aa52 | 230 | |
| emilmont | 1:fdd22bb7aa52 | 231 | /* S = packed((ya - yc), (xa - xc) ) */ |
| emilmont | 1:fdd22bb7aa52 | 232 | S = __QSUB16(T, S); |
| emilmont | 1:fdd22bb7aa52 | 233 | |
| emilmont | 1:fdd22bb7aa52 | 234 | /* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 235 | /* Read yb (real), xb(imag) input */ |
| mbed_official | 5:3762170b6d4d | 236 | T = _SIMD32_OFFSET(pSi1); |
| mbed_official | 5:3762170b6d4d | 237 | T = __SHADD16(T, 0); |
| mbed_official | 5:3762170b6d4d | 238 | T = __SHADD16(T, 0); |
| emilmont | 1:fdd22bb7aa52 | 239 | |
| emilmont | 1:fdd22bb7aa52 | 240 | /* Read yd (real), xd(imag) input */ |
| mbed_official | 5:3762170b6d4d | 241 | U = _SIMD32_OFFSET(pSi3); |
| mbed_official | 5:3762170b6d4d | 242 | U = __SHADD16(U, 0); |
| mbed_official | 5:3762170b6d4d | 243 | U = __SHADD16(U, 0); |
| emilmont | 1:fdd22bb7aa52 | 244 | |
| emilmont | 1:fdd22bb7aa52 | 245 | /* T = packed((yb + yd), (xb + xd) ) */ |
| emilmont | 1:fdd22bb7aa52 | 246 | T = __QADD16(T, U); |
| emilmont | 1:fdd22bb7aa52 | 247 | |
| emilmont | 1:fdd22bb7aa52 | 248 | /* writing the butterfly processed i0 sample */ |
| emilmont | 1:fdd22bb7aa52 | 249 | /* xa' = xa + xb + xc + xd */ |
| emilmont | 1:fdd22bb7aa52 | 250 | /* ya' = ya + yb + yc + yd */ |
| mbed_official | 5:3762170b6d4d | 251 | _SIMD32_OFFSET(pSi0) = __SHADD16(R, T); |
| mbed_official | 5:3762170b6d4d | 252 | pSi0 += 2; |
| emilmont | 1:fdd22bb7aa52 | 253 | |
| emilmont | 1:fdd22bb7aa52 | 254 | /* R = packed((ya + yc) - (yb + yd), (xa + xc)- (xb + xd)) */ |
| emilmont | 1:fdd22bb7aa52 | 255 | R = __QSUB16(R, T); |
| emilmont | 1:fdd22bb7aa52 | 256 | |
| emilmont | 1:fdd22bb7aa52 | 257 | /* co2 & si2 are read from SIMD Coefficient pointer */ |
| emilmont | 1:fdd22bb7aa52 | 258 | C2 = _SIMD32_OFFSET(pCoef16 + (4u * ic)); |
| emilmont | 1:fdd22bb7aa52 | 259 | |
| emilmont | 1:fdd22bb7aa52 | 260 | #ifndef ARM_MATH_BIG_ENDIAN |
| emilmont | 1:fdd22bb7aa52 | 261 | |
| emilmont | 1:fdd22bb7aa52 | 262 | /* xc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 263 | out1 = __SMUAD(C2, R) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 264 | /* yc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 265 | out2 = __SMUSDX(C2, R); |
| emilmont | 1:fdd22bb7aa52 | 266 | |
| emilmont | 1:fdd22bb7aa52 | 267 | #else |
| emilmont | 1:fdd22bb7aa52 | 268 | |
| emilmont | 1:fdd22bb7aa52 | 269 | /* xc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 270 | out1 = __SMUSDX(R, C2) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 271 | /* yc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 272 | out2 = __SMUAD(C2, R); |
| emilmont | 1:fdd22bb7aa52 | 273 | |
| emilmont | 1:fdd22bb7aa52 | 274 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
| emilmont | 1:fdd22bb7aa52 | 275 | |
| emilmont | 1:fdd22bb7aa52 | 276 | /* Reading i0+fftLen/4 */ |
| emilmont | 1:fdd22bb7aa52 | 277 | /* T = packed(yb, xb) */ |
| mbed_official | 5:3762170b6d4d | 278 | T = _SIMD32_OFFSET(pSi1); |
| mbed_official | 5:3762170b6d4d | 279 | T = __SHADD16(T, 0); |
| mbed_official | 5:3762170b6d4d | 280 | T = __SHADD16(T, 0); |
| emilmont | 1:fdd22bb7aa52 | 281 | |
| emilmont | 1:fdd22bb7aa52 | 282 | /* writing the butterfly processed i0 + fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 283 | /* writing output(xc', yc') in little endian format */ |
| mbed_official | 5:3762170b6d4d | 284 | _SIMD32_OFFSET(pSi1) = |
| emilmont | 1:fdd22bb7aa52 | 285 | (q31_t) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| mbed_official | 5:3762170b6d4d | 286 | pSi1 += 2; |
| emilmont | 1:fdd22bb7aa52 | 287 | |
| emilmont | 1:fdd22bb7aa52 | 288 | /* Butterfly calculations */ |
| emilmont | 1:fdd22bb7aa52 | 289 | /* U = packed(yd, xd) */ |
| mbed_official | 5:3762170b6d4d | 290 | U = _SIMD32_OFFSET(pSi3); |
| mbed_official | 5:3762170b6d4d | 291 | U = __SHADD16(U, 0); |
| mbed_official | 5:3762170b6d4d | 292 | U = __SHADD16(U, 0); |
| emilmont | 1:fdd22bb7aa52 | 293 | |
| emilmont | 1:fdd22bb7aa52 | 294 | /* T = packed(yb-yd, xb-xd) */ |
| emilmont | 1:fdd22bb7aa52 | 295 | T = __QSUB16(T, U); |
| emilmont | 1:fdd22bb7aa52 | 296 | |
| emilmont | 1:fdd22bb7aa52 | 297 | #ifndef ARM_MATH_BIG_ENDIAN |
| emilmont | 1:fdd22bb7aa52 | 298 | |
| emilmont | 1:fdd22bb7aa52 | 299 | /* R = packed((ya-yc) + (xb- xd) , (xa-xc) - (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 300 | R = __QASX(S, T); |
| emilmont | 1:fdd22bb7aa52 | 301 | /* S = packed((ya-yc) - (xb- xd), (xa-xc) + (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 302 | S = __QSAX(S, T); |
| emilmont | 1:fdd22bb7aa52 | 303 | |
| emilmont | 1:fdd22bb7aa52 | 304 | #else |
| emilmont | 1:fdd22bb7aa52 | 305 | |
| emilmont | 1:fdd22bb7aa52 | 306 | /* R = packed((ya-yc) + (xb- xd) , (xa-xc) - (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 307 | R = __QSAX(S, T); |
| emilmont | 1:fdd22bb7aa52 | 308 | /* S = packed((ya-yc) - (xb- xd), (xa-xc) + (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 309 | S = __QASX(S, T); |
| emilmont | 1:fdd22bb7aa52 | 310 | |
| emilmont | 1:fdd22bb7aa52 | 311 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
| emilmont | 1:fdd22bb7aa52 | 312 | |
| emilmont | 1:fdd22bb7aa52 | 313 | /* co1 & si1 are read from SIMD Coefficient pointer */ |
| emilmont | 1:fdd22bb7aa52 | 314 | C1 = _SIMD32_OFFSET(pCoef16 + (2u * ic)); |
| emilmont | 1:fdd22bb7aa52 | 315 | /* Butterfly process for the i0+fftLen/2 sample */ |
| emilmont | 1:fdd22bb7aa52 | 316 | |
| emilmont | 1:fdd22bb7aa52 | 317 | #ifndef ARM_MATH_BIG_ENDIAN |
| emilmont | 1:fdd22bb7aa52 | 318 | |
| emilmont | 1:fdd22bb7aa52 | 319 | /* xb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) */ |
| emilmont | 1:fdd22bb7aa52 | 320 | out1 = __SMUAD(C1, S) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 321 | /* yb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) */ |
| emilmont | 1:fdd22bb7aa52 | 322 | out2 = __SMUSDX(C1, S); |
| emilmont | 1:fdd22bb7aa52 | 323 | |
| emilmont | 1:fdd22bb7aa52 | 324 | #else |
| emilmont | 1:fdd22bb7aa52 | 325 | |
| emilmont | 1:fdd22bb7aa52 | 326 | /* xb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) */ |
| emilmont | 1:fdd22bb7aa52 | 327 | out1 = __SMUSDX(S, C1) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 328 | /* yb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) */ |
| emilmont | 1:fdd22bb7aa52 | 329 | out2 = __SMUAD(C1, S); |
| emilmont | 1:fdd22bb7aa52 | 330 | |
| emilmont | 1:fdd22bb7aa52 | 331 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
| emilmont | 1:fdd22bb7aa52 | 332 | |
| emilmont | 1:fdd22bb7aa52 | 333 | /* writing output(xb', yb') in little endian format */ |
| mbed_official | 5:3762170b6d4d | 334 | _SIMD32_OFFSET(pSi2) = |
| emilmont | 1:fdd22bb7aa52 | 335 | ((out2) & 0xFFFF0000) | ((out1) & 0x0000FFFF); |
| mbed_official | 5:3762170b6d4d | 336 | pSi2 += 2; |
| emilmont | 1:fdd22bb7aa52 | 337 | |
| emilmont | 1:fdd22bb7aa52 | 338 | |
| emilmont | 1:fdd22bb7aa52 | 339 | /* co3 & si3 are read from SIMD Coefficient pointer */ |
| emilmont | 1:fdd22bb7aa52 | 340 | C3 = _SIMD32_OFFSET(pCoef16 + (6u * ic)); |
| emilmont | 1:fdd22bb7aa52 | 341 | /* Butterfly process for the i0+3fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 342 | |
| emilmont | 1:fdd22bb7aa52 | 343 | #ifndef ARM_MATH_BIG_ENDIAN |
| emilmont | 1:fdd22bb7aa52 | 344 | |
| emilmont | 1:fdd22bb7aa52 | 345 | /* xd' = (xa-yb-xc+yd)* co3 + (ya+xb-yc-xd)* (si3) */ |
| emilmont | 1:fdd22bb7aa52 | 346 | out1 = __SMUAD(C3, R) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 347 | /* yd' = (ya+xb-yc-xd)* co3 - (xa-yb-xc+yd)* (si3) */ |
| emilmont | 1:fdd22bb7aa52 | 348 | out2 = __SMUSDX(C3, R); |
| emilmont | 1:fdd22bb7aa52 | 349 | |
| emilmont | 1:fdd22bb7aa52 | 350 | #else |
| emilmont | 1:fdd22bb7aa52 | 351 | |
| emilmont | 1:fdd22bb7aa52 | 352 | /* xd' = (ya+xb-yc-xd)* co3 - (xa-yb-xc+yd)* (si3) */ |
| emilmont | 1:fdd22bb7aa52 | 353 | out1 = __SMUSDX(R, C3) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 354 | /* yd' = (xa-yb-xc+yd)* co3 + (ya+xb-yc-xd)* (si3) */ |
| emilmont | 1:fdd22bb7aa52 | 355 | out2 = __SMUAD(C3, R); |
| emilmont | 1:fdd22bb7aa52 | 356 | |
| emilmont | 1:fdd22bb7aa52 | 357 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
| emilmont | 1:fdd22bb7aa52 | 358 | |
| emilmont | 1:fdd22bb7aa52 | 359 | /* writing output(xd', yd') in little endian format */ |
| mbed_official | 5:3762170b6d4d | 360 | _SIMD32_OFFSET(pSi3) = |
| emilmont | 1:fdd22bb7aa52 | 361 | ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| mbed_official | 5:3762170b6d4d | 362 | pSi3 += 2; |
| emilmont | 1:fdd22bb7aa52 | 363 | |
| emilmont | 1:fdd22bb7aa52 | 364 | /* Twiddle coefficients index modifier */ |
| emilmont | 1:fdd22bb7aa52 | 365 | ic = ic + twidCoefModifier; |
| emilmont | 1:fdd22bb7aa52 | 366 | |
| emilmont | 1:fdd22bb7aa52 | 367 | } while(--j); |
| emilmont | 1:fdd22bb7aa52 | 368 | /* data is in 4.11(q11) format */ |
| emilmont | 1:fdd22bb7aa52 | 369 | |
| emilmont | 1:fdd22bb7aa52 | 370 | /* end of first stage process */ |
| emilmont | 1:fdd22bb7aa52 | 371 | |
| emilmont | 1:fdd22bb7aa52 | 372 | |
| emilmont | 1:fdd22bb7aa52 | 373 | /* start of middle stage process */ |
| emilmont | 1:fdd22bb7aa52 | 374 | |
| emilmont | 1:fdd22bb7aa52 | 375 | /* Twiddle coefficients index modifier */ |
| emilmont | 1:fdd22bb7aa52 | 376 | twidCoefModifier <<= 2u; |
| emilmont | 1:fdd22bb7aa52 | 377 | |
| emilmont | 1:fdd22bb7aa52 | 378 | /* Calculation of Middle stage */ |
| emilmont | 1:fdd22bb7aa52 | 379 | for (k = fftLen / 4u; k > 4u; k >>= 2u) |
| emilmont | 1:fdd22bb7aa52 | 380 | { |
| emilmont | 1:fdd22bb7aa52 | 381 | /* Initializations for the middle stage */ |
| emilmont | 1:fdd22bb7aa52 | 382 | n1 = n2; |
| emilmont | 1:fdd22bb7aa52 | 383 | n2 >>= 2u; |
| emilmont | 1:fdd22bb7aa52 | 384 | ic = 0u; |
| emilmont | 1:fdd22bb7aa52 | 385 | |
| emilmont | 1:fdd22bb7aa52 | 386 | for (j = 0u; j <= (n2 - 1u); j++) |
| emilmont | 1:fdd22bb7aa52 | 387 | { |
| emilmont | 1:fdd22bb7aa52 | 388 | /* index calculation for the coefficients */ |
| emilmont | 1:fdd22bb7aa52 | 389 | C1 = _SIMD32_OFFSET(pCoef16 + (2u * ic)); |
| emilmont | 1:fdd22bb7aa52 | 390 | C2 = _SIMD32_OFFSET(pCoef16 + (4u * ic)); |
| emilmont | 1:fdd22bb7aa52 | 391 | C3 = _SIMD32_OFFSET(pCoef16 + (6u * ic)); |
| emilmont | 1:fdd22bb7aa52 | 392 | |
| emilmont | 1:fdd22bb7aa52 | 393 | /* Twiddle coefficients index modifier */ |
| emilmont | 1:fdd22bb7aa52 | 394 | ic = ic + twidCoefModifier; |
| mbed_official | 5:3762170b6d4d | 395 | |
| mbed_official | 5:3762170b6d4d | 396 | pSi0 = pSrc16 + 2 * j; |
| mbed_official | 5:3762170b6d4d | 397 | pSi1 = pSi0 + 2 * n2; |
| mbed_official | 5:3762170b6d4d | 398 | pSi2 = pSi1 + 2 * n2; |
| mbed_official | 5:3762170b6d4d | 399 | pSi3 = pSi2 + 2 * n2; |
| emilmont | 1:fdd22bb7aa52 | 400 | |
| emilmont | 1:fdd22bb7aa52 | 401 | /* Butterfly implementation */ |
| emilmont | 1:fdd22bb7aa52 | 402 | for (i0 = j; i0 < fftLen; i0 += n1) |
| emilmont | 1:fdd22bb7aa52 | 403 | { |
| emilmont | 1:fdd22bb7aa52 | 404 | /* Reading i0, i0+fftLen/2 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 405 | /* Read ya (real), xa(imag) input */ |
| mbed_official | 5:3762170b6d4d | 406 | T = _SIMD32_OFFSET(pSi0); |
| emilmont | 1:fdd22bb7aa52 | 407 | |
| emilmont | 1:fdd22bb7aa52 | 408 | /* Read yc (real), xc(imag) input */ |
| mbed_official | 5:3762170b6d4d | 409 | S = _SIMD32_OFFSET(pSi2); |
| emilmont | 1:fdd22bb7aa52 | 410 | |
| emilmont | 1:fdd22bb7aa52 | 411 | /* R = packed( (ya + yc), (xa + xc)) */ |
| emilmont | 1:fdd22bb7aa52 | 412 | R = __QADD16(T, S); |
| emilmont | 1:fdd22bb7aa52 | 413 | |
| emilmont | 1:fdd22bb7aa52 | 414 | /* S = packed((ya - yc), (xa - xc)) */ |
| emilmont | 1:fdd22bb7aa52 | 415 | S = __QSUB16(T, S); |
| emilmont | 1:fdd22bb7aa52 | 416 | |
| emilmont | 1:fdd22bb7aa52 | 417 | /* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 418 | /* Read yb (real), xb(imag) input */ |
| mbed_official | 5:3762170b6d4d | 419 | T = _SIMD32_OFFSET(pSi1); |
| emilmont | 1:fdd22bb7aa52 | 420 | |
| emilmont | 1:fdd22bb7aa52 | 421 | /* Read yd (real), xd(imag) input */ |
| mbed_official | 5:3762170b6d4d | 422 | U = _SIMD32_OFFSET(pSi3); |
| emilmont | 1:fdd22bb7aa52 | 423 | |
| emilmont | 1:fdd22bb7aa52 | 424 | /* T = packed( (yb + yd), (xb + xd)) */ |
| emilmont | 1:fdd22bb7aa52 | 425 | T = __QADD16(T, U); |
| emilmont | 1:fdd22bb7aa52 | 426 | |
| emilmont | 1:fdd22bb7aa52 | 427 | /* writing the butterfly processed i0 sample */ |
| emilmont | 1:fdd22bb7aa52 | 428 | |
| emilmont | 1:fdd22bb7aa52 | 429 | /* xa' = xa + xb + xc + xd */ |
| emilmont | 1:fdd22bb7aa52 | 430 | /* ya' = ya + yb + yc + yd */ |
| emilmont | 1:fdd22bb7aa52 | 431 | out1 = __SHADD16(R, T); |
| mbed_official | 5:3762170b6d4d | 432 | out1 = __SHADD16(out1, 0); |
| mbed_official | 5:3762170b6d4d | 433 | _SIMD32_OFFSET(pSi0) = out1; |
| mbed_official | 5:3762170b6d4d | 434 | pSi0 += 2 * n1; |
| emilmont | 1:fdd22bb7aa52 | 435 | |
| emilmont | 1:fdd22bb7aa52 | 436 | /* R = packed( (ya + yc) - (yb + yd), (xa + xc) - (xb + xd)) */ |
| emilmont | 1:fdd22bb7aa52 | 437 | R = __SHSUB16(R, T); |
| emilmont | 1:fdd22bb7aa52 | 438 | |
| emilmont | 1:fdd22bb7aa52 | 439 | #ifndef ARM_MATH_BIG_ENDIAN |
| emilmont | 1:fdd22bb7aa52 | 440 | |
| emilmont | 1:fdd22bb7aa52 | 441 | /* (ya-yb+yc-yd)* (si2) + (xa-xb+xc-xd)* co2 */ |
| emilmont | 1:fdd22bb7aa52 | 442 | out1 = __SMUAD(C2, R) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 443 | |
| emilmont | 1:fdd22bb7aa52 | 444 | /* (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 445 | out2 = __SMUSDX(C2, R); |
| emilmont | 1:fdd22bb7aa52 | 446 | |
| emilmont | 1:fdd22bb7aa52 | 447 | #else |
| emilmont | 1:fdd22bb7aa52 | 448 | |
| emilmont | 1:fdd22bb7aa52 | 449 | /* (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 450 | out1 = __SMUSDX(R, C2) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 451 | |
| emilmont | 1:fdd22bb7aa52 | 452 | /* (ya-yb+yc-yd)* (si2) + (xa-xb+xc-xd)* co2 */ |
| emilmont | 1:fdd22bb7aa52 | 453 | out2 = __SMUAD(C2, R); |
| emilmont | 1:fdd22bb7aa52 | 454 | |
| emilmont | 1:fdd22bb7aa52 | 455 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
| emilmont | 1:fdd22bb7aa52 | 456 | |
| emilmont | 1:fdd22bb7aa52 | 457 | /* Reading i0+3fftLen/4 */ |
| emilmont | 1:fdd22bb7aa52 | 458 | /* Read yb (real), xb(imag) input */ |
| mbed_official | 5:3762170b6d4d | 459 | T = _SIMD32_OFFSET(pSi1); |
| emilmont | 1:fdd22bb7aa52 | 460 | |
| emilmont | 1:fdd22bb7aa52 | 461 | /* writing the butterfly processed i0 + fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 462 | /* xc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 463 | /* yc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */ |
| mbed_official | 5:3762170b6d4d | 464 | _SIMD32_OFFSET(pSi1) = |
| emilmont | 1:fdd22bb7aa52 | 465 | ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| mbed_official | 5:3762170b6d4d | 466 | pSi1 += 2 * n1; |
| emilmont | 1:fdd22bb7aa52 | 467 | |
| emilmont | 1:fdd22bb7aa52 | 468 | /* Butterfly calculations */ |
| emilmont | 1:fdd22bb7aa52 | 469 | |
| emilmont | 1:fdd22bb7aa52 | 470 | /* Read yd (real), xd(imag) input */ |
| mbed_official | 5:3762170b6d4d | 471 | U = _SIMD32_OFFSET(pSi3); |
| emilmont | 1:fdd22bb7aa52 | 472 | |
| emilmont | 1:fdd22bb7aa52 | 473 | /* T = packed(yb-yd, xb-xd) */ |
| emilmont | 1:fdd22bb7aa52 | 474 | T = __QSUB16(T, U); |
| emilmont | 1:fdd22bb7aa52 | 475 | |
| emilmont | 1:fdd22bb7aa52 | 476 | #ifndef ARM_MATH_BIG_ENDIAN |
| emilmont | 1:fdd22bb7aa52 | 477 | |
| emilmont | 1:fdd22bb7aa52 | 478 | /* R = packed((ya-yc) + (xb- xd) , (xa-xc) - (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 479 | R = __SHASX(S, T); |
| emilmont | 1:fdd22bb7aa52 | 480 | |
| emilmont | 1:fdd22bb7aa52 | 481 | /* S = packed((ya-yc) - (xb- xd), (xa-xc) + (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 482 | S = __SHSAX(S, T); |
| emilmont | 1:fdd22bb7aa52 | 483 | |
| emilmont | 1:fdd22bb7aa52 | 484 | |
| emilmont | 1:fdd22bb7aa52 | 485 | /* Butterfly process for the i0+fftLen/2 sample */ |
| emilmont | 1:fdd22bb7aa52 | 486 | out1 = __SMUAD(C1, S) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 487 | out2 = __SMUSDX(C1, S); |
| emilmont | 1:fdd22bb7aa52 | 488 | |
| emilmont | 1:fdd22bb7aa52 | 489 | #else |
| emilmont | 1:fdd22bb7aa52 | 490 | |
| emilmont | 1:fdd22bb7aa52 | 491 | /* R = packed((ya-yc) + (xb- xd) , (xa-xc) - (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 492 | R = __SHSAX(S, T); |
| emilmont | 1:fdd22bb7aa52 | 493 | |
| emilmont | 1:fdd22bb7aa52 | 494 | /* S = packed((ya-yc) - (xb- xd), (xa-xc) + (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 495 | S = __SHASX(S, T); |
| emilmont | 1:fdd22bb7aa52 | 496 | |
| emilmont | 1:fdd22bb7aa52 | 497 | |
| emilmont | 1:fdd22bb7aa52 | 498 | /* Butterfly process for the i0+fftLen/2 sample */ |
| emilmont | 1:fdd22bb7aa52 | 499 | out1 = __SMUSDX(S, C1) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 500 | out2 = __SMUAD(C1, S); |
| emilmont | 1:fdd22bb7aa52 | 501 | |
| emilmont | 1:fdd22bb7aa52 | 502 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
| emilmont | 1:fdd22bb7aa52 | 503 | |
| emilmont | 1:fdd22bb7aa52 | 504 | /* xb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) */ |
| emilmont | 1:fdd22bb7aa52 | 505 | /* yb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) */ |
| mbed_official | 5:3762170b6d4d | 506 | _SIMD32_OFFSET(pSi2) = |
| emilmont | 1:fdd22bb7aa52 | 507 | ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| mbed_official | 5:3762170b6d4d | 508 | pSi2 += 2 * n1; |
| emilmont | 1:fdd22bb7aa52 | 509 | |
| emilmont | 1:fdd22bb7aa52 | 510 | /* Butterfly process for the i0+3fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 511 | |
| emilmont | 1:fdd22bb7aa52 | 512 | #ifndef ARM_MATH_BIG_ENDIAN |
| emilmont | 1:fdd22bb7aa52 | 513 | |
| emilmont | 1:fdd22bb7aa52 | 514 | out1 = __SMUAD(C3, R) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 515 | out2 = __SMUSDX(C3, R); |
| emilmont | 1:fdd22bb7aa52 | 516 | |
| emilmont | 1:fdd22bb7aa52 | 517 | #else |
| emilmont | 1:fdd22bb7aa52 | 518 | |
| emilmont | 1:fdd22bb7aa52 | 519 | out1 = __SMUSDX(R, C3) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 520 | out2 = __SMUAD(C3, R); |
| emilmont | 1:fdd22bb7aa52 | 521 | |
| emilmont | 1:fdd22bb7aa52 | 522 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
| emilmont | 1:fdd22bb7aa52 | 523 | |
| emilmont | 1:fdd22bb7aa52 | 524 | /* xd' = (xa-yb-xc+yd)* co3 + (ya+xb-yc-xd)* (si3) */ |
| emilmont | 1:fdd22bb7aa52 | 525 | /* yd' = (ya+xb-yc-xd)* co3 - (xa-yb-xc+yd)* (si3) */ |
| mbed_official | 5:3762170b6d4d | 526 | _SIMD32_OFFSET(pSi3) = |
| emilmont | 1:fdd22bb7aa52 | 527 | ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| mbed_official | 5:3762170b6d4d | 528 | pSi3 += 2 * n1; |
| emilmont | 1:fdd22bb7aa52 | 529 | } |
| emilmont | 1:fdd22bb7aa52 | 530 | } |
| emilmont | 1:fdd22bb7aa52 | 531 | /* Twiddle coefficients index modifier */ |
| emilmont | 1:fdd22bb7aa52 | 532 | twidCoefModifier <<= 2u; |
| emilmont | 1:fdd22bb7aa52 | 533 | } |
| emilmont | 1:fdd22bb7aa52 | 534 | /* end of middle stage process */ |
| emilmont | 1:fdd22bb7aa52 | 535 | |
| emilmont | 1:fdd22bb7aa52 | 536 | |
| emilmont | 1:fdd22bb7aa52 | 537 | /* data is in 10.6(q6) format for the 1024 point */ |
| emilmont | 1:fdd22bb7aa52 | 538 | /* data is in 8.8(q8) format for the 256 point */ |
| emilmont | 1:fdd22bb7aa52 | 539 | /* data is in 6.10(q10) format for the 64 point */ |
| emilmont | 1:fdd22bb7aa52 | 540 | /* data is in 4.12(q12) format for the 16 point */ |
| emilmont | 1:fdd22bb7aa52 | 541 | |
| emilmont | 1:fdd22bb7aa52 | 542 | /* Initializations for the last stage */ |
| emilmont | 1:fdd22bb7aa52 | 543 | j = fftLen >> 2; |
| emilmont | 1:fdd22bb7aa52 | 544 | |
| emilmont | 1:fdd22bb7aa52 | 545 | ptr1 = &pSrc16[0]; |
| emilmont | 1:fdd22bb7aa52 | 546 | |
| emilmont | 1:fdd22bb7aa52 | 547 | /* start of last stage process */ |
| emilmont | 1:fdd22bb7aa52 | 548 | |
| emilmont | 1:fdd22bb7aa52 | 549 | /* Butterfly implementation */ |
| emilmont | 1:fdd22bb7aa52 | 550 | do |
| emilmont | 1:fdd22bb7aa52 | 551 | { |
| emilmont | 1:fdd22bb7aa52 | 552 | /* Read xa (real), ya(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 553 | xaya = *__SIMD32(ptr1)++; |
| emilmont | 1:fdd22bb7aa52 | 554 | |
| emilmont | 1:fdd22bb7aa52 | 555 | /* Read xb (real), yb(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 556 | xbyb = *__SIMD32(ptr1)++; |
| emilmont | 1:fdd22bb7aa52 | 557 | |
| emilmont | 1:fdd22bb7aa52 | 558 | /* Read xc (real), yc(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 559 | xcyc = *__SIMD32(ptr1)++; |
| emilmont | 1:fdd22bb7aa52 | 560 | |
| emilmont | 1:fdd22bb7aa52 | 561 | /* Read xd (real), yd(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 562 | xdyd = *__SIMD32(ptr1)++; |
| emilmont | 1:fdd22bb7aa52 | 563 | |
| emilmont | 1:fdd22bb7aa52 | 564 | /* R = packed((ya + yc), (xa + xc)) */ |
| emilmont | 1:fdd22bb7aa52 | 565 | R = __QADD16(xaya, xcyc); |
| emilmont | 1:fdd22bb7aa52 | 566 | |
| emilmont | 1:fdd22bb7aa52 | 567 | /* T = packed((yb + yd), (xb + xd)) */ |
| emilmont | 1:fdd22bb7aa52 | 568 | T = __QADD16(xbyb, xdyd); |
| emilmont | 1:fdd22bb7aa52 | 569 | |
| emilmont | 1:fdd22bb7aa52 | 570 | /* pointer updation for writing */ |
| emilmont | 1:fdd22bb7aa52 | 571 | ptr1 = ptr1 - 8u; |
| emilmont | 1:fdd22bb7aa52 | 572 | |
| emilmont | 1:fdd22bb7aa52 | 573 | |
| emilmont | 1:fdd22bb7aa52 | 574 | /* xa' = xa + xb + xc + xd */ |
| emilmont | 1:fdd22bb7aa52 | 575 | /* ya' = ya + yb + yc + yd */ |
| emilmont | 1:fdd22bb7aa52 | 576 | *__SIMD32(ptr1)++ = __SHADD16(R, T); |
| emilmont | 1:fdd22bb7aa52 | 577 | |
| emilmont | 1:fdd22bb7aa52 | 578 | /* T = packed((yb + yd), (xb + xd)) */ |
| emilmont | 1:fdd22bb7aa52 | 579 | T = __QADD16(xbyb, xdyd); |
| emilmont | 1:fdd22bb7aa52 | 580 | |
| emilmont | 1:fdd22bb7aa52 | 581 | /* xc' = (xa-xb+xc-xd) */ |
| emilmont | 1:fdd22bb7aa52 | 582 | /* yc' = (ya-yb+yc-yd) */ |
| emilmont | 1:fdd22bb7aa52 | 583 | *__SIMD32(ptr1)++ = __SHSUB16(R, T); |
| emilmont | 1:fdd22bb7aa52 | 584 | |
| emilmont | 1:fdd22bb7aa52 | 585 | /* S = packed((ya - yc), (xa - xc)) */ |
| emilmont | 1:fdd22bb7aa52 | 586 | S = __QSUB16(xaya, xcyc); |
| emilmont | 1:fdd22bb7aa52 | 587 | |
| emilmont | 1:fdd22bb7aa52 | 588 | /* Read yd (real), xd(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 589 | /* T = packed( (yb - yd), (xb - xd)) */ |
| emilmont | 1:fdd22bb7aa52 | 590 | U = __QSUB16(xbyb, xdyd); |
| emilmont | 1:fdd22bb7aa52 | 591 | |
| emilmont | 1:fdd22bb7aa52 | 592 | #ifndef ARM_MATH_BIG_ENDIAN |
| emilmont | 1:fdd22bb7aa52 | 593 | |
| emilmont | 1:fdd22bb7aa52 | 594 | /* xb' = (xa+yb-xc-yd) */ |
| emilmont | 1:fdd22bb7aa52 | 595 | /* yb' = (ya-xb-yc+xd) */ |
| emilmont | 1:fdd22bb7aa52 | 596 | *__SIMD32(ptr1)++ = __SHSAX(S, U); |
| emilmont | 1:fdd22bb7aa52 | 597 | |
| emilmont | 1:fdd22bb7aa52 | 598 | |
| emilmont | 1:fdd22bb7aa52 | 599 | /* xd' = (xa-yb-xc+yd) */ |
| emilmont | 1:fdd22bb7aa52 | 600 | /* yd' = (ya+xb-yc-xd) */ |
| emilmont | 1:fdd22bb7aa52 | 601 | *__SIMD32(ptr1)++ = __SHASX(S, U); |
| emilmont | 1:fdd22bb7aa52 | 602 | |
| emilmont | 1:fdd22bb7aa52 | 603 | #else |
| emilmont | 1:fdd22bb7aa52 | 604 | |
| emilmont | 1:fdd22bb7aa52 | 605 | /* xb' = (xa+yb-xc-yd) */ |
| emilmont | 1:fdd22bb7aa52 | 606 | /* yb' = (ya-xb-yc+xd) */ |
| emilmont | 1:fdd22bb7aa52 | 607 | *__SIMD32(ptr1)++ = __SHASX(S, U); |
| emilmont | 1:fdd22bb7aa52 | 608 | |
| emilmont | 1:fdd22bb7aa52 | 609 | |
| emilmont | 1:fdd22bb7aa52 | 610 | /* xd' = (xa-yb-xc+yd) */ |
| emilmont | 1:fdd22bb7aa52 | 611 | /* yd' = (ya+xb-yc-xd) */ |
| emilmont | 1:fdd22bb7aa52 | 612 | *__SIMD32(ptr1)++ = __SHSAX(S, U); |
| emilmont | 1:fdd22bb7aa52 | 613 | |
| emilmont | 1:fdd22bb7aa52 | 614 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
| emilmont | 1:fdd22bb7aa52 | 615 | |
| emilmont | 1:fdd22bb7aa52 | 616 | } while(--j); |
| emilmont | 1:fdd22bb7aa52 | 617 | |
| emilmont | 1:fdd22bb7aa52 | 618 | /* end of last stage process */ |
| emilmont | 1:fdd22bb7aa52 | 619 | |
| emilmont | 1:fdd22bb7aa52 | 620 | /* output is in 11.5(q5) format for the 1024 point */ |
| emilmont | 1:fdd22bb7aa52 | 621 | /* output is in 9.7(q7) format for the 256 point */ |
| emilmont | 1:fdd22bb7aa52 | 622 | /* output is in 7.9(q9) format for the 64 point */ |
| emilmont | 1:fdd22bb7aa52 | 623 | /* output is in 5.11(q11) format for the 16 point */ |
| emilmont | 1:fdd22bb7aa52 | 624 | |
| emilmont | 1:fdd22bb7aa52 | 625 | |
| emilmont | 1:fdd22bb7aa52 | 626 | #else |
| emilmont | 1:fdd22bb7aa52 | 627 | |
| emilmont | 1:fdd22bb7aa52 | 628 | /* Run the below code for Cortex-M0 */ |
| emilmont | 1:fdd22bb7aa52 | 629 | |
| emilmont | 1:fdd22bb7aa52 | 630 | q15_t R0, R1, S0, S1, T0, T1, U0, U1; |
| emilmont | 1:fdd22bb7aa52 | 631 | q15_t Co1, Si1, Co2, Si2, Co3, Si3, out1, out2; |
| emilmont | 1:fdd22bb7aa52 | 632 | uint32_t n1, n2, ic, i0, i1, i2, i3, j, k; |
| emilmont | 1:fdd22bb7aa52 | 633 | |
| emilmont | 1:fdd22bb7aa52 | 634 | /* Total process is divided into three stages */ |
| emilmont | 1:fdd22bb7aa52 | 635 | |
| emilmont | 1:fdd22bb7aa52 | 636 | /* process first stage, middle stages, & last stage */ |
| emilmont | 1:fdd22bb7aa52 | 637 | |
| emilmont | 1:fdd22bb7aa52 | 638 | /* Initializations for the first stage */ |
| emilmont | 1:fdd22bb7aa52 | 639 | n2 = fftLen; |
| emilmont | 1:fdd22bb7aa52 | 640 | n1 = n2; |
| emilmont | 1:fdd22bb7aa52 | 641 | |
| emilmont | 1:fdd22bb7aa52 | 642 | /* n2 = fftLen/4 */ |
| emilmont | 1:fdd22bb7aa52 | 643 | n2 >>= 2u; |
| emilmont | 1:fdd22bb7aa52 | 644 | |
| emilmont | 1:fdd22bb7aa52 | 645 | /* Index for twiddle coefficient */ |
| emilmont | 1:fdd22bb7aa52 | 646 | ic = 0u; |
| emilmont | 1:fdd22bb7aa52 | 647 | |
| emilmont | 1:fdd22bb7aa52 | 648 | /* Index for input read and output write */ |
| emilmont | 1:fdd22bb7aa52 | 649 | i0 = 0u; |
| emilmont | 1:fdd22bb7aa52 | 650 | j = n2; |
| emilmont | 1:fdd22bb7aa52 | 651 | |
| emilmont | 1:fdd22bb7aa52 | 652 | /* Input is in 1.15(q15) format */ |
| emilmont | 1:fdd22bb7aa52 | 653 | |
| emilmont | 1:fdd22bb7aa52 | 654 | /* start of first stage process */ |
| emilmont | 1:fdd22bb7aa52 | 655 | do |
| emilmont | 1:fdd22bb7aa52 | 656 | { |
| emilmont | 1:fdd22bb7aa52 | 657 | /* Butterfly implementation */ |
| emilmont | 1:fdd22bb7aa52 | 658 | |
| emilmont | 1:fdd22bb7aa52 | 659 | /* index calculation for the input as, */ |
| emilmont | 1:fdd22bb7aa52 | 660 | /* pSrc16[i0 + 0], pSrc16[i0 + fftLen/4], pSrc16[i0 + fftLen/2], pSrc16[i0 + 3fftLen/4] */ |
| emilmont | 1:fdd22bb7aa52 | 661 | i1 = i0 + n2; |
| emilmont | 1:fdd22bb7aa52 | 662 | i2 = i1 + n2; |
| emilmont | 1:fdd22bb7aa52 | 663 | i3 = i2 + n2; |
| emilmont | 1:fdd22bb7aa52 | 664 | |
| emilmont | 1:fdd22bb7aa52 | 665 | /* Reading i0, i0+fftLen/2 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 666 | |
| emilmont | 1:fdd22bb7aa52 | 667 | /* input is down scale by 4 to avoid overflow */ |
| emilmont | 1:fdd22bb7aa52 | 668 | /* Read ya (real), xa(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 669 | T0 = pSrc16[i0 * 2u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 670 | T1 = pSrc16[(i0 * 2u) + 1u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 671 | |
| emilmont | 1:fdd22bb7aa52 | 672 | /* input is down scale by 4 to avoid overflow */ |
| emilmont | 1:fdd22bb7aa52 | 673 | /* Read yc (real), xc(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 674 | S0 = pSrc16[i2 * 2u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 675 | S1 = pSrc16[(i2 * 2u) + 1u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 676 | |
| emilmont | 1:fdd22bb7aa52 | 677 | /* R0 = (ya + yc) */ |
| emilmont | 1:fdd22bb7aa52 | 678 | R0 = __SSAT(T0 + S0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 679 | /* R1 = (xa + xc) */ |
| emilmont | 1:fdd22bb7aa52 | 680 | R1 = __SSAT(T1 + S1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 681 | |
| emilmont | 1:fdd22bb7aa52 | 682 | /* S0 = (ya - yc) */ |
| emilmont | 1:fdd22bb7aa52 | 683 | S0 = __SSAT(T0 - S0, 16); |
| emilmont | 1:fdd22bb7aa52 | 684 | /* S1 = (xa - xc) */ |
| emilmont | 1:fdd22bb7aa52 | 685 | S1 = __SSAT(T1 - S1, 16); |
| emilmont | 1:fdd22bb7aa52 | 686 | |
| emilmont | 1:fdd22bb7aa52 | 687 | /* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 688 | /* input is down scale by 4 to avoid overflow */ |
| emilmont | 1:fdd22bb7aa52 | 689 | /* Read yb (real), xb(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 690 | T0 = pSrc16[i1 * 2u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 691 | T1 = pSrc16[(i1 * 2u) + 1u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 692 | |
| emilmont | 1:fdd22bb7aa52 | 693 | /* input is down scale by 4 to avoid overflow */ |
| emilmont | 1:fdd22bb7aa52 | 694 | /* Read yd (real), xd(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 695 | U0 = pSrc16[i3 * 2u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 696 | U1 = pSrc16[(i3 * 2u) + 1] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 697 | |
| emilmont | 1:fdd22bb7aa52 | 698 | /* T0 = (yb + yd) */ |
| emilmont | 1:fdd22bb7aa52 | 699 | T0 = __SSAT(T0 + U0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 700 | /* T1 = (xb + xd) */ |
| emilmont | 1:fdd22bb7aa52 | 701 | T1 = __SSAT(T1 + U1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 702 | |
| emilmont | 1:fdd22bb7aa52 | 703 | /* writing the butterfly processed i0 sample */ |
| emilmont | 1:fdd22bb7aa52 | 704 | /* ya' = ya + yb + yc + yd */ |
| emilmont | 1:fdd22bb7aa52 | 705 | /* xa' = xa + xb + xc + xd */ |
| emilmont | 1:fdd22bb7aa52 | 706 | pSrc16[i0 * 2u] = (R0 >> 1u) + (T0 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 707 | pSrc16[(i0 * 2u) + 1u] = (R1 >> 1u) + (T1 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 708 | |
| emilmont | 1:fdd22bb7aa52 | 709 | /* R0 = (ya + yc) - (yb + yd) */ |
| emilmont | 1:fdd22bb7aa52 | 710 | /* R1 = (xa + xc) - (xb + xd) */ |
| emilmont | 1:fdd22bb7aa52 | 711 | R0 = __SSAT(R0 - T0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 712 | R1 = __SSAT(R1 - T1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 713 | |
| emilmont | 1:fdd22bb7aa52 | 714 | /* co2 & si2 are read from Coefficient pointer */ |
| emilmont | 1:fdd22bb7aa52 | 715 | Co2 = pCoef16[2u * ic * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 716 | Si2 = pCoef16[(2u * ic * 2u) + 1]; |
| emilmont | 1:fdd22bb7aa52 | 717 | |
| emilmont | 1:fdd22bb7aa52 | 718 | /* xc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) */ |
| mbed_official | 5:3762170b6d4d | 719 | out1 = (q15_t) ((Co2 * R0 + Si2 * R1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 720 | /* yc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */ |
| mbed_official | 5:3762170b6d4d | 721 | out2 = (q15_t) ((-Si2 * R0 + Co2 * R1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 722 | |
| emilmont | 1:fdd22bb7aa52 | 723 | /* Reading i0+fftLen/4 */ |
| emilmont | 1:fdd22bb7aa52 | 724 | /* input is down scale by 4 to avoid overflow */ |
| emilmont | 1:fdd22bb7aa52 | 725 | /* T0 = yb, T1 = xb */ |
| emilmont | 1:fdd22bb7aa52 | 726 | T0 = pSrc16[i1 * 2u] >> 2; |
| emilmont | 1:fdd22bb7aa52 | 727 | T1 = pSrc16[(i1 * 2u) + 1] >> 2; |
| emilmont | 1:fdd22bb7aa52 | 728 | |
| emilmont | 1:fdd22bb7aa52 | 729 | /* writing the butterfly processed i0 + fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 730 | /* writing output(xc', yc') in little endian format */ |
| emilmont | 1:fdd22bb7aa52 | 731 | pSrc16[i1 * 2u] = out1; |
| emilmont | 1:fdd22bb7aa52 | 732 | pSrc16[(i1 * 2u) + 1] = out2; |
| emilmont | 1:fdd22bb7aa52 | 733 | |
| emilmont | 1:fdd22bb7aa52 | 734 | /* Butterfly calculations */ |
| emilmont | 1:fdd22bb7aa52 | 735 | /* input is down scale by 4 to avoid overflow */ |
| emilmont | 1:fdd22bb7aa52 | 736 | /* U0 = yd, U1 = xd */ |
| emilmont | 1:fdd22bb7aa52 | 737 | U0 = pSrc16[i3 * 2u] >> 2; |
| emilmont | 1:fdd22bb7aa52 | 738 | U1 = pSrc16[(i3 * 2u) + 1] >> 2; |
| emilmont | 1:fdd22bb7aa52 | 739 | /* T0 = yb-yd */ |
| emilmont | 1:fdd22bb7aa52 | 740 | T0 = __SSAT(T0 - U0, 16); |
| emilmont | 1:fdd22bb7aa52 | 741 | /* T1 = xb-xd */ |
| emilmont | 1:fdd22bb7aa52 | 742 | T1 = __SSAT(T1 - U1, 16); |
| emilmont | 1:fdd22bb7aa52 | 743 | |
| emilmont | 1:fdd22bb7aa52 | 744 | /* R1 = (ya-yc) + (xb- xd), R0 = (xa-xc) - (yb-yd)) */ |
| mbed_official | 5:3762170b6d4d | 745 | R0 = (q15_t) __SSAT((q31_t) (S0 - T1), 16); |
| mbed_official | 5:3762170b6d4d | 746 | R1 = (q15_t) __SSAT((q31_t) (S1 + T0), 16); |
| emilmont | 1:fdd22bb7aa52 | 747 | |
| emilmont | 1:fdd22bb7aa52 | 748 | /* S1 = (ya-yc) - (xb- xd), S0 = (xa-xc) + (yb-yd)) */ |
| mbed_official | 5:3762170b6d4d | 749 | S0 = (q15_t) __SSAT(((q31_t) S0 + T1), 16u); |
| mbed_official | 5:3762170b6d4d | 750 | S1 = (q15_t) __SSAT(((q31_t) S1 - T0), 16u); |
| emilmont | 1:fdd22bb7aa52 | 751 | |
| emilmont | 1:fdd22bb7aa52 | 752 | /* co1 & si1 are read from Coefficient pointer */ |
| emilmont | 1:fdd22bb7aa52 | 753 | Co1 = pCoef16[ic * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 754 | Si1 = pCoef16[(ic * 2u) + 1]; |
| emilmont | 1:fdd22bb7aa52 | 755 | /* Butterfly process for the i0+fftLen/2 sample */ |
| emilmont | 1:fdd22bb7aa52 | 756 | /* xb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) */ |
| mbed_official | 5:3762170b6d4d | 757 | out1 = (q15_t) ((Si1 * S1 + Co1 * S0) >> 16); |
| emilmont | 1:fdd22bb7aa52 | 758 | /* yb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) */ |
| mbed_official | 5:3762170b6d4d | 759 | out2 = (q15_t) ((-Si1 * S0 + Co1 * S1) >> 16); |
| emilmont | 1:fdd22bb7aa52 | 760 | |
| emilmont | 1:fdd22bb7aa52 | 761 | /* writing output(xb', yb') in little endian format */ |
| emilmont | 1:fdd22bb7aa52 | 762 | pSrc16[i2 * 2u] = out1; |
| emilmont | 1:fdd22bb7aa52 | 763 | pSrc16[(i2 * 2u) + 1] = out2; |
| emilmont | 1:fdd22bb7aa52 | 764 | |
| emilmont | 1:fdd22bb7aa52 | 765 | /* Co3 & si3 are read from Coefficient pointer */ |
| emilmont | 1:fdd22bb7aa52 | 766 | Co3 = pCoef16[3u * (ic * 2u)]; |
| emilmont | 1:fdd22bb7aa52 | 767 | Si3 = pCoef16[(3u * (ic * 2u)) + 1]; |
| emilmont | 1:fdd22bb7aa52 | 768 | /* Butterfly process for the i0+3fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 769 | /* xd' = (xa-yb-xc+yd)* Co3 + (ya+xb-yc-xd)* (si3) */ |
| mbed_official | 5:3762170b6d4d | 770 | out1 = (q15_t) ((Si3 * R1 + Co3 * R0) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 771 | /* yd' = (ya+xb-yc-xd)* Co3 - (xa-yb-xc+yd)* (si3) */ |
| mbed_official | 5:3762170b6d4d | 772 | out2 = (q15_t) ((-Si3 * R0 + Co3 * R1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 773 | /* writing output(xd', yd') in little endian format */ |
| emilmont | 1:fdd22bb7aa52 | 774 | pSrc16[i3 * 2u] = out1; |
| emilmont | 1:fdd22bb7aa52 | 775 | pSrc16[(i3 * 2u) + 1] = out2; |
| emilmont | 1:fdd22bb7aa52 | 776 | |
| emilmont | 1:fdd22bb7aa52 | 777 | /* Twiddle coefficients index modifier */ |
| emilmont | 1:fdd22bb7aa52 | 778 | ic = ic + twidCoefModifier; |
| emilmont | 1:fdd22bb7aa52 | 779 | |
| emilmont | 1:fdd22bb7aa52 | 780 | /* Updating input index */ |
| emilmont | 1:fdd22bb7aa52 | 781 | i0 = i0 + 1u; |
| emilmont | 1:fdd22bb7aa52 | 782 | |
| emilmont | 1:fdd22bb7aa52 | 783 | } while(--j); |
| emilmont | 1:fdd22bb7aa52 | 784 | /* data is in 4.11(q11) format */ |
| emilmont | 1:fdd22bb7aa52 | 785 | |
| emilmont | 1:fdd22bb7aa52 | 786 | /* end of first stage process */ |
| emilmont | 1:fdd22bb7aa52 | 787 | |
| emilmont | 1:fdd22bb7aa52 | 788 | |
| emilmont | 1:fdd22bb7aa52 | 789 | /* start of middle stage process */ |
| emilmont | 1:fdd22bb7aa52 | 790 | |
| emilmont | 1:fdd22bb7aa52 | 791 | /* Twiddle coefficients index modifier */ |
| emilmont | 1:fdd22bb7aa52 | 792 | twidCoefModifier <<= 2u; |
| emilmont | 1:fdd22bb7aa52 | 793 | |
| emilmont | 1:fdd22bb7aa52 | 794 | /* Calculation of Middle stage */ |
| emilmont | 1:fdd22bb7aa52 | 795 | for (k = fftLen / 4u; k > 4u; k >>= 2u) |
| emilmont | 1:fdd22bb7aa52 | 796 | { |
| emilmont | 1:fdd22bb7aa52 | 797 | /* Initializations for the middle stage */ |
| emilmont | 1:fdd22bb7aa52 | 798 | n1 = n2; |
| emilmont | 1:fdd22bb7aa52 | 799 | n2 >>= 2u; |
| emilmont | 1:fdd22bb7aa52 | 800 | ic = 0u; |
| emilmont | 1:fdd22bb7aa52 | 801 | |
| emilmont | 1:fdd22bb7aa52 | 802 | for (j = 0u; j <= (n2 - 1u); j++) |
| emilmont | 1:fdd22bb7aa52 | 803 | { |
| emilmont | 1:fdd22bb7aa52 | 804 | /* index calculation for the coefficients */ |
| emilmont | 1:fdd22bb7aa52 | 805 | Co1 = pCoef16[ic * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 806 | Si1 = pCoef16[(ic * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 807 | Co2 = pCoef16[2u * (ic * 2u)]; |
| emilmont | 1:fdd22bb7aa52 | 808 | Si2 = pCoef16[(2u * (ic * 2u)) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 809 | Co3 = pCoef16[3u * (ic * 2u)]; |
| emilmont | 1:fdd22bb7aa52 | 810 | Si3 = pCoef16[(3u * (ic * 2u)) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 811 | |
| emilmont | 1:fdd22bb7aa52 | 812 | /* Twiddle coefficients index modifier */ |
| emilmont | 1:fdd22bb7aa52 | 813 | ic = ic + twidCoefModifier; |
| emilmont | 1:fdd22bb7aa52 | 814 | |
| emilmont | 1:fdd22bb7aa52 | 815 | /* Butterfly implementation */ |
| emilmont | 1:fdd22bb7aa52 | 816 | for (i0 = j; i0 < fftLen; i0 += n1) |
| emilmont | 1:fdd22bb7aa52 | 817 | { |
| emilmont | 1:fdd22bb7aa52 | 818 | /* index calculation for the input as, */ |
| emilmont | 1:fdd22bb7aa52 | 819 | /* pSrc16[i0 + 0], pSrc16[i0 + fftLen/4], pSrc16[i0 + fftLen/2], pSrc16[i0 + 3fftLen/4] */ |
| emilmont | 1:fdd22bb7aa52 | 820 | i1 = i0 + n2; |
| emilmont | 1:fdd22bb7aa52 | 821 | i2 = i1 + n2; |
| emilmont | 1:fdd22bb7aa52 | 822 | i3 = i2 + n2; |
| emilmont | 1:fdd22bb7aa52 | 823 | |
| emilmont | 1:fdd22bb7aa52 | 824 | /* Reading i0, i0+fftLen/2 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 825 | /* Read ya (real), xa(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 826 | T0 = pSrc16[i0 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 827 | T1 = pSrc16[(i0 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 828 | |
| emilmont | 1:fdd22bb7aa52 | 829 | /* Read yc (real), xc(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 830 | S0 = pSrc16[i2 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 831 | S1 = pSrc16[(i2 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 832 | |
| emilmont | 1:fdd22bb7aa52 | 833 | /* R0 = (ya + yc), R1 = (xa + xc) */ |
| emilmont | 1:fdd22bb7aa52 | 834 | R0 = __SSAT(T0 + S0, 16); |
| emilmont | 1:fdd22bb7aa52 | 835 | R1 = __SSAT(T1 + S1, 16); |
| emilmont | 1:fdd22bb7aa52 | 836 | |
| emilmont | 1:fdd22bb7aa52 | 837 | /* S0 = (ya - yc), S1 =(xa - xc) */ |
| emilmont | 1:fdd22bb7aa52 | 838 | S0 = __SSAT(T0 - S0, 16); |
| emilmont | 1:fdd22bb7aa52 | 839 | S1 = __SSAT(T1 - S1, 16); |
| emilmont | 1:fdd22bb7aa52 | 840 | |
| emilmont | 1:fdd22bb7aa52 | 841 | /* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 842 | /* Read yb (real), xb(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 843 | T0 = pSrc16[i1 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 844 | T1 = pSrc16[(i1 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 845 | |
| emilmont | 1:fdd22bb7aa52 | 846 | /* Read yd (real), xd(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 847 | U0 = pSrc16[i3 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 848 | U1 = pSrc16[(i3 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 849 | |
| emilmont | 1:fdd22bb7aa52 | 850 | |
| emilmont | 1:fdd22bb7aa52 | 851 | /* T0 = (yb + yd), T1 = (xb + xd) */ |
| emilmont | 1:fdd22bb7aa52 | 852 | T0 = __SSAT(T0 + U0, 16); |
| emilmont | 1:fdd22bb7aa52 | 853 | T1 = __SSAT(T1 + U1, 16); |
| emilmont | 1:fdd22bb7aa52 | 854 | |
| emilmont | 1:fdd22bb7aa52 | 855 | /* writing the butterfly processed i0 sample */ |
| emilmont | 1:fdd22bb7aa52 | 856 | |
| emilmont | 1:fdd22bb7aa52 | 857 | /* xa' = xa + xb + xc + xd */ |
| emilmont | 1:fdd22bb7aa52 | 858 | /* ya' = ya + yb + yc + yd */ |
| emilmont | 1:fdd22bb7aa52 | 859 | out1 = ((R0 >> 1u) + (T0 >> 1u)) >> 1u; |
| emilmont | 1:fdd22bb7aa52 | 860 | out2 = ((R1 >> 1u) + (T1 >> 1u)) >> 1u; |
| emilmont | 1:fdd22bb7aa52 | 861 | |
| emilmont | 1:fdd22bb7aa52 | 862 | pSrc16[i0 * 2u] = out1; |
| emilmont | 1:fdd22bb7aa52 | 863 | pSrc16[(2u * i0) + 1u] = out2; |
| emilmont | 1:fdd22bb7aa52 | 864 | |
| emilmont | 1:fdd22bb7aa52 | 865 | /* R0 = (ya + yc) - (yb + yd), R1 = (xa + xc) - (xb + xd) */ |
| emilmont | 1:fdd22bb7aa52 | 866 | R0 = (R0 >> 1u) - (T0 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 867 | R1 = (R1 >> 1u) - (T1 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 868 | |
| emilmont | 1:fdd22bb7aa52 | 869 | /* (ya-yb+yc-yd)* (si2) + (xa-xb+xc-xd)* co2 */ |
| mbed_official | 5:3762170b6d4d | 870 | out1 = (q15_t) ((Co2 * R0 + Si2 * R1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 871 | |
| emilmont | 1:fdd22bb7aa52 | 872 | /* (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */ |
| mbed_official | 5:3762170b6d4d | 873 | out2 = (q15_t) ((-Si2 * R0 + Co2 * R1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 874 | |
| emilmont | 1:fdd22bb7aa52 | 875 | /* Reading i0+3fftLen/4 */ |
| emilmont | 1:fdd22bb7aa52 | 876 | /* Read yb (real), xb(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 877 | T0 = pSrc16[i1 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 878 | T1 = pSrc16[(i1 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 879 | |
| emilmont | 1:fdd22bb7aa52 | 880 | /* writing the butterfly processed i0 + fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 881 | /* xc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 882 | /* yc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 883 | pSrc16[i1 * 2u] = out1; |
| emilmont | 1:fdd22bb7aa52 | 884 | pSrc16[(i1 * 2u) + 1u] = out2; |
| emilmont | 1:fdd22bb7aa52 | 885 | |
| emilmont | 1:fdd22bb7aa52 | 886 | /* Butterfly calculations */ |
| emilmont | 1:fdd22bb7aa52 | 887 | |
| emilmont | 1:fdd22bb7aa52 | 888 | /* Read yd (real), xd(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 889 | U0 = pSrc16[i3 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 890 | U1 = pSrc16[(i3 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 891 | |
| emilmont | 1:fdd22bb7aa52 | 892 | /* T0 = yb-yd, T1 = xb-xd */ |
| emilmont | 1:fdd22bb7aa52 | 893 | T0 = __SSAT(T0 - U0, 16); |
| emilmont | 1:fdd22bb7aa52 | 894 | T1 = __SSAT(T1 - U1, 16); |
| emilmont | 1:fdd22bb7aa52 | 895 | |
| emilmont | 1:fdd22bb7aa52 | 896 | /* R0 = (ya-yc) + (xb- xd), R1 = (xa-xc) - (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 897 | R0 = (S0 >> 1u) - (T1 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 898 | R1 = (S1 >> 1u) + (T0 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 899 | |
| emilmont | 1:fdd22bb7aa52 | 900 | /* S0 = (ya-yc) - (xb- xd), S1 = (xa-xc) + (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 901 | S0 = (S0 >> 1u) + (T1 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 902 | S1 = (S1 >> 1u) - (T0 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 903 | |
| emilmont | 1:fdd22bb7aa52 | 904 | /* Butterfly process for the i0+fftLen/2 sample */ |
| mbed_official | 5:3762170b6d4d | 905 | out1 = (q15_t) ((Co1 * S0 + Si1 * S1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 906 | |
| mbed_official | 5:3762170b6d4d | 907 | out2 = (q15_t) ((-Si1 * S0 + Co1 * S1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 908 | |
| emilmont | 1:fdd22bb7aa52 | 909 | /* xb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) */ |
| emilmont | 1:fdd22bb7aa52 | 910 | /* yb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) */ |
| emilmont | 1:fdd22bb7aa52 | 911 | pSrc16[i2 * 2u] = out1; |
| emilmont | 1:fdd22bb7aa52 | 912 | pSrc16[(i2 * 2u) + 1u] = out2; |
| emilmont | 1:fdd22bb7aa52 | 913 | |
| emilmont | 1:fdd22bb7aa52 | 914 | /* Butterfly process for the i0+3fftLen/4 sample */ |
| mbed_official | 5:3762170b6d4d | 915 | out1 = (q15_t) ((Si3 * R1 + Co3 * R0) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 916 | |
| mbed_official | 5:3762170b6d4d | 917 | out2 = (q15_t) ((-Si3 * R0 + Co3 * R1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 918 | /* xd' = (xa-yb-xc+yd)* Co3 + (ya+xb-yc-xd)* (si3) */ |
| emilmont | 1:fdd22bb7aa52 | 919 | /* yd' = (ya+xb-yc-xd)* Co3 - (xa-yb-xc+yd)* (si3) */ |
| emilmont | 1:fdd22bb7aa52 | 920 | pSrc16[i3 * 2u] = out1; |
| emilmont | 1:fdd22bb7aa52 | 921 | pSrc16[(i3 * 2u) + 1u] = out2; |
| emilmont | 1:fdd22bb7aa52 | 922 | } |
| emilmont | 1:fdd22bb7aa52 | 923 | } |
| emilmont | 1:fdd22bb7aa52 | 924 | /* Twiddle coefficients index modifier */ |
| emilmont | 1:fdd22bb7aa52 | 925 | twidCoefModifier <<= 2u; |
| emilmont | 1:fdd22bb7aa52 | 926 | } |
| emilmont | 1:fdd22bb7aa52 | 927 | /* end of middle stage process */ |
| emilmont | 1:fdd22bb7aa52 | 928 | |
| emilmont | 1:fdd22bb7aa52 | 929 | |
| emilmont | 1:fdd22bb7aa52 | 930 | /* data is in 10.6(q6) format for the 1024 point */ |
| emilmont | 1:fdd22bb7aa52 | 931 | /* data is in 8.8(q8) format for the 256 point */ |
| emilmont | 1:fdd22bb7aa52 | 932 | /* data is in 6.10(q10) format for the 64 point */ |
| emilmont | 1:fdd22bb7aa52 | 933 | /* data is in 4.12(q12) format for the 16 point */ |
| emilmont | 1:fdd22bb7aa52 | 934 | |
| emilmont | 1:fdd22bb7aa52 | 935 | /* Initializations for the last stage */ |
| emilmont | 1:fdd22bb7aa52 | 936 | n1 = n2; |
| emilmont | 1:fdd22bb7aa52 | 937 | n2 >>= 2u; |
| emilmont | 1:fdd22bb7aa52 | 938 | |
| emilmont | 1:fdd22bb7aa52 | 939 | /* start of last stage process */ |
| emilmont | 1:fdd22bb7aa52 | 940 | |
| emilmont | 1:fdd22bb7aa52 | 941 | /* Butterfly implementation */ |
| emilmont | 1:fdd22bb7aa52 | 942 | for (i0 = 0u; i0 <= (fftLen - n1); i0 += n1) |
| emilmont | 1:fdd22bb7aa52 | 943 | { |
| emilmont | 1:fdd22bb7aa52 | 944 | /* index calculation for the input as, */ |
| emilmont | 1:fdd22bb7aa52 | 945 | /* pSrc16[i0 + 0], pSrc16[i0 + fftLen/4], pSrc16[i0 + fftLen/2], pSrc16[i0 + 3fftLen/4] */ |
| emilmont | 1:fdd22bb7aa52 | 946 | i1 = i0 + n2; |
| emilmont | 1:fdd22bb7aa52 | 947 | i2 = i1 + n2; |
| emilmont | 1:fdd22bb7aa52 | 948 | i3 = i2 + n2; |
| emilmont | 1:fdd22bb7aa52 | 949 | |
| emilmont | 1:fdd22bb7aa52 | 950 | /* Reading i0, i0+fftLen/2 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 951 | /* Read ya (real), xa(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 952 | T0 = pSrc16[i0 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 953 | T1 = pSrc16[(i0 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 954 | |
| emilmont | 1:fdd22bb7aa52 | 955 | /* Read yc (real), xc(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 956 | S0 = pSrc16[i2 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 957 | S1 = pSrc16[(i2 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 958 | |
| emilmont | 1:fdd22bb7aa52 | 959 | /* R0 = (ya + yc), R1 = (xa + xc) */ |
| emilmont | 1:fdd22bb7aa52 | 960 | R0 = __SSAT(T0 + S0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 961 | R1 = __SSAT(T1 + S1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 962 | |
| emilmont | 1:fdd22bb7aa52 | 963 | /* S0 = (ya - yc), S1 = (xa - xc) */ |
| emilmont | 1:fdd22bb7aa52 | 964 | S0 = __SSAT(T0 - S0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 965 | S1 = __SSAT(T1 - S1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 966 | |
| emilmont | 1:fdd22bb7aa52 | 967 | /* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 968 | /* Read yb (real), xb(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 969 | T0 = pSrc16[i1 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 970 | T1 = pSrc16[(i1 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 971 | /* Read yd (real), xd(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 972 | U0 = pSrc16[i3 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 973 | U1 = pSrc16[(i3 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 974 | |
| emilmont | 1:fdd22bb7aa52 | 975 | /* T0 = (yb + yd), T1 = (xb + xd)) */ |
| emilmont | 1:fdd22bb7aa52 | 976 | T0 = __SSAT(T0 + U0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 977 | T1 = __SSAT(T1 + U1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 978 | |
| emilmont | 1:fdd22bb7aa52 | 979 | /* writing the butterfly processed i0 sample */ |
| emilmont | 1:fdd22bb7aa52 | 980 | /* xa' = xa + xb + xc + xd */ |
| emilmont | 1:fdd22bb7aa52 | 981 | /* ya' = ya + yb + yc + yd */ |
| emilmont | 1:fdd22bb7aa52 | 982 | pSrc16[i0 * 2u] = (R0 >> 1u) + (T0 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 983 | pSrc16[(i0 * 2u) + 1u] = (R1 >> 1u) + (T1 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 984 | |
| emilmont | 1:fdd22bb7aa52 | 985 | /* R0 = (ya + yc) - (yb + yd), R1 = (xa + xc) - (xb + xd) */ |
| emilmont | 1:fdd22bb7aa52 | 986 | R0 = (R0 >> 1u) - (T0 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 987 | R1 = (R1 >> 1u) - (T1 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 988 | /* Read yb (real), xb(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 989 | T0 = pSrc16[i1 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 990 | T1 = pSrc16[(i1 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 991 | |
| emilmont | 1:fdd22bb7aa52 | 992 | /* writing the butterfly processed i0 + fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 993 | /* xc' = (xa-xb+xc-xd) */ |
| emilmont | 1:fdd22bb7aa52 | 994 | /* yc' = (ya-yb+yc-yd) */ |
| emilmont | 1:fdd22bb7aa52 | 995 | pSrc16[i1 * 2u] = R0; |
| emilmont | 1:fdd22bb7aa52 | 996 | pSrc16[(i1 * 2u) + 1u] = R1; |
| emilmont | 1:fdd22bb7aa52 | 997 | |
| emilmont | 1:fdd22bb7aa52 | 998 | /* Read yd (real), xd(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 999 | U0 = pSrc16[i3 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1000 | U1 = pSrc16[(i3 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1001 | /* T0 = (yb - yd), T1 = (xb - xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1002 | T0 = __SSAT(T0 - U0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1003 | T1 = __SSAT(T1 - U1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1004 | |
| emilmont | 1:fdd22bb7aa52 | 1005 | /* writing the butterfly processed i0 + fftLen/2 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1006 | /* xb' = (xa+yb-xc-yd) */ |
| emilmont | 1:fdd22bb7aa52 | 1007 | /* yb' = (ya-xb-yc+xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1008 | pSrc16[i2 * 2u] = (S0 >> 1u) + (T1 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1009 | pSrc16[(i2 * 2u) + 1u] = (S1 >> 1u) - (T0 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1010 | |
| emilmont | 1:fdd22bb7aa52 | 1011 | /* writing the butterfly processed i0 + 3fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1012 | /* xd' = (xa-yb-xc+yd) */ |
| emilmont | 1:fdd22bb7aa52 | 1013 | /* yd' = (ya+xb-yc-xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1014 | pSrc16[i3 * 2u] = (S0 >> 1u) - (T1 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1015 | pSrc16[(i3 * 2u) + 1u] = (S1 >> 1u) + (T0 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1016 | |
| emilmont | 1:fdd22bb7aa52 | 1017 | } |
| emilmont | 1:fdd22bb7aa52 | 1018 | |
| emilmont | 1:fdd22bb7aa52 | 1019 | /* end of last stage process */ |
| emilmont | 1:fdd22bb7aa52 | 1020 | |
| emilmont | 1:fdd22bb7aa52 | 1021 | /* output is in 11.5(q5) format for the 1024 point */ |
| emilmont | 1:fdd22bb7aa52 | 1022 | /* output is in 9.7(q7) format for the 256 point */ |
| emilmont | 1:fdd22bb7aa52 | 1023 | /* output is in 7.9(q9) format for the 64 point */ |
| emilmont | 1:fdd22bb7aa52 | 1024 | /* output is in 5.11(q11) format for the 16 point */ |
| emilmont | 1:fdd22bb7aa52 | 1025 | |
| mbed_official | 3:7a284390b0ce | 1026 | #endif /* #ifndef ARM_MATH_CM0_FAMILY */ |
| emilmont | 1:fdd22bb7aa52 | 1027 | |
| emilmont | 1:fdd22bb7aa52 | 1028 | } |
| emilmont | 1:fdd22bb7aa52 | 1029 | |
| emilmont | 1:fdd22bb7aa52 | 1030 | |
| emilmont | 1:fdd22bb7aa52 | 1031 | /** |
| emilmont | 1:fdd22bb7aa52 | 1032 | * @brief Core function for the Q15 CIFFT butterfly process. |
| emilmont | 1:fdd22bb7aa52 | 1033 | * @param[in, out] *pSrc16 points to the in-place buffer of Q15 data type. |
| emilmont | 1:fdd22bb7aa52 | 1034 | * @param[in] fftLen length of the FFT. |
| emilmont | 1:fdd22bb7aa52 | 1035 | * @param[in] *pCoef16 points to twiddle coefficient buffer. |
| emilmont | 1:fdd22bb7aa52 | 1036 | * @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table. |
| emilmont | 1:fdd22bb7aa52 | 1037 | * @return none. |
| emilmont | 1:fdd22bb7aa52 | 1038 | */ |
| emilmont | 1:fdd22bb7aa52 | 1039 | |
| emilmont | 1:fdd22bb7aa52 | 1040 | /* |
| emilmont | 1:fdd22bb7aa52 | 1041 | * Radix-4 IFFT algorithm used is : |
| emilmont | 1:fdd22bb7aa52 | 1042 | * |
| emilmont | 1:fdd22bb7aa52 | 1043 | * CIFFT uses same twiddle coefficients as CFFT function |
| emilmont | 1:fdd22bb7aa52 | 1044 | * x[k] = x[n] + (j)k * x[n + fftLen/4] + (-1)k * x[n+fftLen/2] + (-j)k * x[n+3*fftLen/4] |
| emilmont | 1:fdd22bb7aa52 | 1045 | * |
| emilmont | 1:fdd22bb7aa52 | 1046 | * |
| emilmont | 1:fdd22bb7aa52 | 1047 | * IFFT is implemented with following changes in equations from FFT |
| emilmont | 1:fdd22bb7aa52 | 1048 | * |
| emilmont | 1:fdd22bb7aa52 | 1049 | * Input real and imaginary data: |
| emilmont | 1:fdd22bb7aa52 | 1050 | * x(n) = xa + j * ya |
| emilmont | 1:fdd22bb7aa52 | 1051 | * x(n+N/4 ) = xb + j * yb |
| emilmont | 1:fdd22bb7aa52 | 1052 | * x(n+N/2 ) = xc + j * yc |
| emilmont | 1:fdd22bb7aa52 | 1053 | * x(n+3N 4) = xd + j * yd |
| emilmont | 1:fdd22bb7aa52 | 1054 | * |
| emilmont | 1:fdd22bb7aa52 | 1055 | * |
| emilmont | 1:fdd22bb7aa52 | 1056 | * Output real and imaginary data: |
| emilmont | 1:fdd22bb7aa52 | 1057 | * x(4r) = xa'+ j * ya' |
| emilmont | 1:fdd22bb7aa52 | 1058 | * x(4r+1) = xb'+ j * yb' |
| emilmont | 1:fdd22bb7aa52 | 1059 | * x(4r+2) = xc'+ j * yc' |
| emilmont | 1:fdd22bb7aa52 | 1060 | * x(4r+3) = xd'+ j * yd' |
| emilmont | 1:fdd22bb7aa52 | 1061 | * |
| emilmont | 1:fdd22bb7aa52 | 1062 | * |
| emilmont | 1:fdd22bb7aa52 | 1063 | * Twiddle factors for radix-4 IFFT: |
| emilmont | 1:fdd22bb7aa52 | 1064 | * Wn = co1 + j * (si1) |
| emilmont | 1:fdd22bb7aa52 | 1065 | * W2n = co2 + j * (si2) |
| emilmont | 1:fdd22bb7aa52 | 1066 | * W3n = co3 + j * (si3) |
| emilmont | 1:fdd22bb7aa52 | 1067 | |
| emilmont | 1:fdd22bb7aa52 | 1068 | * The real and imaginary output values for the radix-4 butterfly are |
| emilmont | 1:fdd22bb7aa52 | 1069 | * xa' = xa + xb + xc + xd |
| emilmont | 1:fdd22bb7aa52 | 1070 | * ya' = ya + yb + yc + yd |
| emilmont | 1:fdd22bb7aa52 | 1071 | * xb' = (xa-yb-xc+yd)* co1 - (ya+xb-yc-xd)* (si1) |
| emilmont | 1:fdd22bb7aa52 | 1072 | * yb' = (ya+xb-yc-xd)* co1 + (xa-yb-xc+yd)* (si1) |
| emilmont | 1:fdd22bb7aa52 | 1073 | * xc' = (xa-xb+xc-xd)* co2 - (ya-yb+yc-yd)* (si2) |
| emilmont | 1:fdd22bb7aa52 | 1074 | * yc' = (ya-yb+yc-yd)* co2 + (xa-xb+xc-xd)* (si2) |
| emilmont | 1:fdd22bb7aa52 | 1075 | * xd' = (xa+yb-xc-yd)* co3 - (ya-xb-yc+xd)* (si3) |
| emilmont | 1:fdd22bb7aa52 | 1076 | * yd' = (ya-xb-yc+xd)* co3 + (xa+yb-xc-yd)* (si3) |
| emilmont | 1:fdd22bb7aa52 | 1077 | * |
| emilmont | 1:fdd22bb7aa52 | 1078 | */ |
| emilmont | 1:fdd22bb7aa52 | 1079 | |
| emilmont | 1:fdd22bb7aa52 | 1080 | void arm_radix4_butterfly_inverse_q15( |
| emilmont | 1:fdd22bb7aa52 | 1081 | q15_t * pSrc16, |
| emilmont | 1:fdd22bb7aa52 | 1082 | uint32_t fftLen, |
| emilmont | 1:fdd22bb7aa52 | 1083 | q15_t * pCoef16, |
| emilmont | 1:fdd22bb7aa52 | 1084 | uint32_t twidCoefModifier) |
| emilmont | 1:fdd22bb7aa52 | 1085 | { |
| emilmont | 1:fdd22bb7aa52 | 1086 | |
| mbed_official | 3:7a284390b0ce | 1087 | #ifndef ARM_MATH_CM0_FAMILY |
| emilmont | 1:fdd22bb7aa52 | 1088 | |
| emilmont | 1:fdd22bb7aa52 | 1089 | /* Run the below code for Cortex-M4 and Cortex-M3 */ |
| emilmont | 1:fdd22bb7aa52 | 1090 | |
| emilmont | 1:fdd22bb7aa52 | 1091 | q31_t R, S, T, U; |
| emilmont | 1:fdd22bb7aa52 | 1092 | q31_t C1, C2, C3, out1, out2; |
| mbed_official | 5:3762170b6d4d | 1093 | uint32_t n1, n2, ic, i0, j, k; |
| emilmont | 1:fdd22bb7aa52 | 1094 | |
| emilmont | 1:fdd22bb7aa52 | 1095 | q15_t *ptr1; |
| mbed_official | 5:3762170b6d4d | 1096 | q15_t *pSi0; |
| mbed_official | 5:3762170b6d4d | 1097 | q15_t *pSi1; |
| mbed_official | 5:3762170b6d4d | 1098 | q15_t *pSi2; |
| mbed_official | 5:3762170b6d4d | 1099 | q15_t *pSi3; |
| emilmont | 1:fdd22bb7aa52 | 1100 | |
| emilmont | 1:fdd22bb7aa52 | 1101 | q31_t xaya, xbyb, xcyc, xdyd; |
| emilmont | 1:fdd22bb7aa52 | 1102 | |
| emilmont | 1:fdd22bb7aa52 | 1103 | /* Total process is divided into three stages */ |
| emilmont | 1:fdd22bb7aa52 | 1104 | |
| emilmont | 1:fdd22bb7aa52 | 1105 | /* process first stage, middle stages, & last stage */ |
| emilmont | 1:fdd22bb7aa52 | 1106 | |
| emilmont | 1:fdd22bb7aa52 | 1107 | /* Initializations for the first stage */ |
| emilmont | 1:fdd22bb7aa52 | 1108 | n2 = fftLen; |
| emilmont | 1:fdd22bb7aa52 | 1109 | n1 = n2; |
| emilmont | 1:fdd22bb7aa52 | 1110 | |
| emilmont | 1:fdd22bb7aa52 | 1111 | /* n2 = fftLen/4 */ |
| emilmont | 1:fdd22bb7aa52 | 1112 | n2 >>= 2u; |
| emilmont | 1:fdd22bb7aa52 | 1113 | |
| emilmont | 1:fdd22bb7aa52 | 1114 | /* Index for twiddle coefficient */ |
| emilmont | 1:fdd22bb7aa52 | 1115 | ic = 0u; |
| emilmont | 1:fdd22bb7aa52 | 1116 | |
| emilmont | 1:fdd22bb7aa52 | 1117 | /* Index for input read and output write */ |
| emilmont | 1:fdd22bb7aa52 | 1118 | j = n2; |
| mbed_official | 5:3762170b6d4d | 1119 | |
| mbed_official | 5:3762170b6d4d | 1120 | pSi0 = pSrc16; |
| mbed_official | 5:3762170b6d4d | 1121 | pSi1 = pSi0 + 2 * n2; |
| mbed_official | 5:3762170b6d4d | 1122 | pSi2 = pSi1 + 2 * n2; |
| mbed_official | 5:3762170b6d4d | 1123 | pSi3 = pSi2 + 2 * n2; |
| emilmont | 1:fdd22bb7aa52 | 1124 | |
| emilmont | 1:fdd22bb7aa52 | 1125 | /* Input is in 1.15(q15) format */ |
| emilmont | 1:fdd22bb7aa52 | 1126 | |
| emilmont | 1:fdd22bb7aa52 | 1127 | /* start of first stage process */ |
| emilmont | 1:fdd22bb7aa52 | 1128 | do |
| emilmont | 1:fdd22bb7aa52 | 1129 | { |
| emilmont | 1:fdd22bb7aa52 | 1130 | /* Butterfly implementation */ |
| emilmont | 1:fdd22bb7aa52 | 1131 | |
| emilmont | 1:fdd22bb7aa52 | 1132 | /* Reading i0, i0+fftLen/2 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 1133 | /* Read ya (real), xa(imag) input */ |
| mbed_official | 5:3762170b6d4d | 1134 | T = _SIMD32_OFFSET(pSi0); |
| mbed_official | 5:3762170b6d4d | 1135 | T = __SHADD16(T, 0); |
| mbed_official | 5:3762170b6d4d | 1136 | T = __SHADD16(T, 0); |
| emilmont | 1:fdd22bb7aa52 | 1137 | |
| emilmont | 1:fdd22bb7aa52 | 1138 | /* Read yc (real), xc(imag) input */ |
| mbed_official | 5:3762170b6d4d | 1139 | S = _SIMD32_OFFSET(pSi2); |
| mbed_official | 5:3762170b6d4d | 1140 | S = __SHADD16(S, 0); |
| mbed_official | 5:3762170b6d4d | 1141 | S = __SHADD16(S, 0); |
| emilmont | 1:fdd22bb7aa52 | 1142 | |
| emilmont | 1:fdd22bb7aa52 | 1143 | /* R = packed((ya + yc), (xa + xc) ) */ |
| emilmont | 1:fdd22bb7aa52 | 1144 | R = __QADD16(T, S); |
| emilmont | 1:fdd22bb7aa52 | 1145 | |
| emilmont | 1:fdd22bb7aa52 | 1146 | /* S = packed((ya - yc), (xa - xc) ) */ |
| emilmont | 1:fdd22bb7aa52 | 1147 | S = __QSUB16(T, S); |
| emilmont | 1:fdd22bb7aa52 | 1148 | |
| emilmont | 1:fdd22bb7aa52 | 1149 | /* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 1150 | /* Read yb (real), xb(imag) input */ |
| mbed_official | 5:3762170b6d4d | 1151 | T = _SIMD32_OFFSET(pSi1); |
| mbed_official | 5:3762170b6d4d | 1152 | T = __SHADD16(T, 0); |
| mbed_official | 5:3762170b6d4d | 1153 | T = __SHADD16(T, 0); |
| emilmont | 1:fdd22bb7aa52 | 1154 | |
| emilmont | 1:fdd22bb7aa52 | 1155 | /* Read yd (real), xd(imag) input */ |
| mbed_official | 5:3762170b6d4d | 1156 | U = _SIMD32_OFFSET(pSi3); |
| mbed_official | 5:3762170b6d4d | 1157 | U = __SHADD16(U, 0); |
| mbed_official | 5:3762170b6d4d | 1158 | U = __SHADD16(U, 0); |
| emilmont | 1:fdd22bb7aa52 | 1159 | |
| emilmont | 1:fdd22bb7aa52 | 1160 | /* T = packed((yb + yd), (xb + xd) ) */ |
| emilmont | 1:fdd22bb7aa52 | 1161 | T = __QADD16(T, U); |
| emilmont | 1:fdd22bb7aa52 | 1162 | |
| emilmont | 1:fdd22bb7aa52 | 1163 | /* writing the butterfly processed i0 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1164 | /* xa' = xa + xb + xc + xd */ |
| emilmont | 1:fdd22bb7aa52 | 1165 | /* ya' = ya + yb + yc + yd */ |
| mbed_official | 5:3762170b6d4d | 1166 | _SIMD32_OFFSET(pSi0) = __SHADD16(R, T); |
| mbed_official | 5:3762170b6d4d | 1167 | pSi0 += 2; |
| emilmont | 1:fdd22bb7aa52 | 1168 | |
| emilmont | 1:fdd22bb7aa52 | 1169 | /* R = packed((ya + yc) - (yb + yd), (xa + xc)- (xb + xd)) */ |
| emilmont | 1:fdd22bb7aa52 | 1170 | R = __QSUB16(R, T); |
| emilmont | 1:fdd22bb7aa52 | 1171 | |
| emilmont | 1:fdd22bb7aa52 | 1172 | /* co2 & si2 are read from SIMD Coefficient pointer */ |
| emilmont | 1:fdd22bb7aa52 | 1173 | C2 = _SIMD32_OFFSET(pCoef16 + (4u * ic)); |
| emilmont | 1:fdd22bb7aa52 | 1174 | |
| emilmont | 1:fdd22bb7aa52 | 1175 | #ifndef ARM_MATH_BIG_ENDIAN |
| emilmont | 1:fdd22bb7aa52 | 1176 | |
| emilmont | 1:fdd22bb7aa52 | 1177 | /* xc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 1178 | out1 = __SMUSD(C2, R) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 1179 | /* yc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 1180 | out2 = __SMUADX(C2, R); |
| emilmont | 1:fdd22bb7aa52 | 1181 | |
| emilmont | 1:fdd22bb7aa52 | 1182 | #else |
| emilmont | 1:fdd22bb7aa52 | 1183 | |
| emilmont | 1:fdd22bb7aa52 | 1184 | /* xc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 1185 | out1 = __SMUADX(C2, R) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 1186 | /* yc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 1187 | out2 = __SMUSD(__QSUB16(0, C2), R); |
| emilmont | 1:fdd22bb7aa52 | 1188 | |
| emilmont | 1:fdd22bb7aa52 | 1189 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
| emilmont | 1:fdd22bb7aa52 | 1190 | |
| emilmont | 1:fdd22bb7aa52 | 1191 | /* Reading i0+fftLen/4 */ |
| emilmont | 1:fdd22bb7aa52 | 1192 | /* T = packed(yb, xb) */ |
| mbed_official | 5:3762170b6d4d | 1193 | T = _SIMD32_OFFSET(pSi1); |
| mbed_official | 5:3762170b6d4d | 1194 | T = __SHADD16(T, 0); |
| mbed_official | 5:3762170b6d4d | 1195 | T = __SHADD16(T, 0); |
| emilmont | 1:fdd22bb7aa52 | 1196 | |
| emilmont | 1:fdd22bb7aa52 | 1197 | /* writing the butterfly processed i0 + fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1198 | /* writing output(xc', yc') in little endian format */ |
| mbed_official | 5:3762170b6d4d | 1199 | _SIMD32_OFFSET(pSi1) = |
| emilmont | 1:fdd22bb7aa52 | 1200 | (q31_t) ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| mbed_official | 5:3762170b6d4d | 1201 | pSi1 += 2; |
| emilmont | 1:fdd22bb7aa52 | 1202 | |
| emilmont | 1:fdd22bb7aa52 | 1203 | /* Butterfly calculations */ |
| emilmont | 1:fdd22bb7aa52 | 1204 | /* U = packed(yd, xd) */ |
| mbed_official | 5:3762170b6d4d | 1205 | U = _SIMD32_OFFSET(pSi3); |
| mbed_official | 5:3762170b6d4d | 1206 | U = __SHADD16(U, 0); |
| mbed_official | 5:3762170b6d4d | 1207 | U = __SHADD16(U, 0); |
| emilmont | 1:fdd22bb7aa52 | 1208 | |
| emilmont | 1:fdd22bb7aa52 | 1209 | /* T = packed(yb-yd, xb-xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1210 | T = __QSUB16(T, U); |
| emilmont | 1:fdd22bb7aa52 | 1211 | |
| emilmont | 1:fdd22bb7aa52 | 1212 | #ifndef ARM_MATH_BIG_ENDIAN |
| emilmont | 1:fdd22bb7aa52 | 1213 | |
| emilmont | 1:fdd22bb7aa52 | 1214 | /* R = packed((ya-yc) + (xb- xd) , (xa-xc) - (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 1215 | R = __QSAX(S, T); |
| emilmont | 1:fdd22bb7aa52 | 1216 | /* S = packed((ya-yc) + (xb- xd), (xa-xc) - (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 1217 | S = __QASX(S, T); |
| emilmont | 1:fdd22bb7aa52 | 1218 | |
| emilmont | 1:fdd22bb7aa52 | 1219 | #else |
| emilmont | 1:fdd22bb7aa52 | 1220 | |
| emilmont | 1:fdd22bb7aa52 | 1221 | /* R = packed((ya-yc) + (xb- xd) , (xa-xc) - (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 1222 | R = __QASX(S, T); |
| emilmont | 1:fdd22bb7aa52 | 1223 | /* S = packed((ya-yc) - (xb- xd), (xa-xc) + (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 1224 | S = __QSAX(S, T); |
| emilmont | 1:fdd22bb7aa52 | 1225 | |
| emilmont | 1:fdd22bb7aa52 | 1226 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
| emilmont | 1:fdd22bb7aa52 | 1227 | |
| emilmont | 1:fdd22bb7aa52 | 1228 | /* co1 & si1 are read from SIMD Coefficient pointer */ |
| emilmont | 1:fdd22bb7aa52 | 1229 | C1 = _SIMD32_OFFSET(pCoef16 + (2u * ic)); |
| emilmont | 1:fdd22bb7aa52 | 1230 | /* Butterfly process for the i0+fftLen/2 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1231 | |
| emilmont | 1:fdd22bb7aa52 | 1232 | #ifndef ARM_MATH_BIG_ENDIAN |
| emilmont | 1:fdd22bb7aa52 | 1233 | |
| emilmont | 1:fdd22bb7aa52 | 1234 | /* xb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) */ |
| emilmont | 1:fdd22bb7aa52 | 1235 | out1 = __SMUSD(C1, S) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 1236 | /* yb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) */ |
| emilmont | 1:fdd22bb7aa52 | 1237 | out2 = __SMUADX(C1, S); |
| emilmont | 1:fdd22bb7aa52 | 1238 | |
| emilmont | 1:fdd22bb7aa52 | 1239 | #else |
| emilmont | 1:fdd22bb7aa52 | 1240 | |
| emilmont | 1:fdd22bb7aa52 | 1241 | /* xb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) */ |
| emilmont | 1:fdd22bb7aa52 | 1242 | out1 = __SMUADX(C1, S) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 1243 | /* yb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) */ |
| emilmont | 1:fdd22bb7aa52 | 1244 | out2 = __SMUSD(__QSUB16(0, C1), S); |
| emilmont | 1:fdd22bb7aa52 | 1245 | |
| emilmont | 1:fdd22bb7aa52 | 1246 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
| emilmont | 1:fdd22bb7aa52 | 1247 | |
| emilmont | 1:fdd22bb7aa52 | 1248 | /* writing output(xb', yb') in little endian format */ |
| mbed_official | 5:3762170b6d4d | 1249 | _SIMD32_OFFSET(pSi2) = |
| emilmont | 1:fdd22bb7aa52 | 1250 | ((out2) & 0xFFFF0000) | ((out1) & 0x0000FFFF); |
| mbed_official | 5:3762170b6d4d | 1251 | pSi2 += 2; |
| emilmont | 1:fdd22bb7aa52 | 1252 | |
| emilmont | 1:fdd22bb7aa52 | 1253 | |
| emilmont | 1:fdd22bb7aa52 | 1254 | /* co3 & si3 are read from SIMD Coefficient pointer */ |
| emilmont | 1:fdd22bb7aa52 | 1255 | C3 = _SIMD32_OFFSET(pCoef16 + (6u * ic)); |
| emilmont | 1:fdd22bb7aa52 | 1256 | /* Butterfly process for the i0+3fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1257 | |
| emilmont | 1:fdd22bb7aa52 | 1258 | #ifndef ARM_MATH_BIG_ENDIAN |
| emilmont | 1:fdd22bb7aa52 | 1259 | |
| emilmont | 1:fdd22bb7aa52 | 1260 | /* xd' = (xa-yb-xc+yd)* co3 + (ya+xb-yc-xd)* (si3) */ |
| emilmont | 1:fdd22bb7aa52 | 1261 | out1 = __SMUSD(C3, R) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 1262 | /* yd' = (ya+xb-yc-xd)* co3 - (xa-yb-xc+yd)* (si3) */ |
| emilmont | 1:fdd22bb7aa52 | 1263 | out2 = __SMUADX(C3, R); |
| emilmont | 1:fdd22bb7aa52 | 1264 | |
| emilmont | 1:fdd22bb7aa52 | 1265 | #else |
| emilmont | 1:fdd22bb7aa52 | 1266 | |
| emilmont | 1:fdd22bb7aa52 | 1267 | /* xd' = (ya+xb-yc-xd)* co3 - (xa-yb-xc+yd)* (si3) */ |
| emilmont | 1:fdd22bb7aa52 | 1268 | out1 = __SMUADX(C3, R) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 1269 | /* yd' = (xa-yb-xc+yd)* co3 + (ya+xb-yc-xd)* (si3) */ |
| emilmont | 1:fdd22bb7aa52 | 1270 | out2 = __SMUSD(__QSUB16(0, C3), R); |
| emilmont | 1:fdd22bb7aa52 | 1271 | |
| emilmont | 1:fdd22bb7aa52 | 1272 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
| emilmont | 1:fdd22bb7aa52 | 1273 | |
| emilmont | 1:fdd22bb7aa52 | 1274 | /* writing output(xd', yd') in little endian format */ |
| mbed_official | 5:3762170b6d4d | 1275 | _SIMD32_OFFSET(pSi3) = |
| emilmont | 1:fdd22bb7aa52 | 1276 | ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| mbed_official | 5:3762170b6d4d | 1277 | pSi3 += 2; |
| emilmont | 1:fdd22bb7aa52 | 1278 | |
| emilmont | 1:fdd22bb7aa52 | 1279 | /* Twiddle coefficients index modifier */ |
| emilmont | 1:fdd22bb7aa52 | 1280 | ic = ic + twidCoefModifier; |
| emilmont | 1:fdd22bb7aa52 | 1281 | |
| emilmont | 1:fdd22bb7aa52 | 1282 | } while(--j); |
| emilmont | 1:fdd22bb7aa52 | 1283 | /* data is in 4.11(q11) format */ |
| emilmont | 1:fdd22bb7aa52 | 1284 | |
| emilmont | 1:fdd22bb7aa52 | 1285 | /* end of first stage process */ |
| emilmont | 1:fdd22bb7aa52 | 1286 | |
| emilmont | 1:fdd22bb7aa52 | 1287 | |
| emilmont | 1:fdd22bb7aa52 | 1288 | /* start of middle stage process */ |
| emilmont | 1:fdd22bb7aa52 | 1289 | |
| emilmont | 1:fdd22bb7aa52 | 1290 | /* Twiddle coefficients index modifier */ |
| emilmont | 1:fdd22bb7aa52 | 1291 | twidCoefModifier <<= 2u; |
| emilmont | 1:fdd22bb7aa52 | 1292 | |
| emilmont | 1:fdd22bb7aa52 | 1293 | /* Calculation of Middle stage */ |
| emilmont | 1:fdd22bb7aa52 | 1294 | for (k = fftLen / 4u; k > 4u; k >>= 2u) |
| emilmont | 1:fdd22bb7aa52 | 1295 | { |
| emilmont | 1:fdd22bb7aa52 | 1296 | /* Initializations for the middle stage */ |
| emilmont | 1:fdd22bb7aa52 | 1297 | n1 = n2; |
| emilmont | 1:fdd22bb7aa52 | 1298 | n2 >>= 2u; |
| emilmont | 1:fdd22bb7aa52 | 1299 | ic = 0u; |
| emilmont | 1:fdd22bb7aa52 | 1300 | |
| emilmont | 1:fdd22bb7aa52 | 1301 | for (j = 0u; j <= (n2 - 1u); j++) |
| emilmont | 1:fdd22bb7aa52 | 1302 | { |
| emilmont | 1:fdd22bb7aa52 | 1303 | /* index calculation for the coefficients */ |
| emilmont | 1:fdd22bb7aa52 | 1304 | C1 = _SIMD32_OFFSET(pCoef16 + (2u * ic)); |
| emilmont | 1:fdd22bb7aa52 | 1305 | C2 = _SIMD32_OFFSET(pCoef16 + (4u * ic)); |
| emilmont | 1:fdd22bb7aa52 | 1306 | C3 = _SIMD32_OFFSET(pCoef16 + (6u * ic)); |
| emilmont | 1:fdd22bb7aa52 | 1307 | |
| emilmont | 1:fdd22bb7aa52 | 1308 | /* Twiddle coefficients index modifier */ |
| emilmont | 1:fdd22bb7aa52 | 1309 | ic = ic + twidCoefModifier; |
| mbed_official | 5:3762170b6d4d | 1310 | |
| mbed_official | 5:3762170b6d4d | 1311 | pSi0 = pSrc16 + 2 * j; |
| mbed_official | 5:3762170b6d4d | 1312 | pSi1 = pSi0 + 2 * n2; |
| mbed_official | 5:3762170b6d4d | 1313 | pSi2 = pSi1 + 2 * n2; |
| mbed_official | 5:3762170b6d4d | 1314 | pSi3 = pSi2 + 2 * n2; |
| emilmont | 1:fdd22bb7aa52 | 1315 | |
| emilmont | 1:fdd22bb7aa52 | 1316 | /* Butterfly implementation */ |
| emilmont | 1:fdd22bb7aa52 | 1317 | for (i0 = j; i0 < fftLen; i0 += n1) |
| emilmont | 1:fdd22bb7aa52 | 1318 | { |
| emilmont | 1:fdd22bb7aa52 | 1319 | /* Reading i0, i0+fftLen/2 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 1320 | /* Read ya (real), xa(imag) input */ |
| mbed_official | 5:3762170b6d4d | 1321 | T = _SIMD32_OFFSET(pSi0); |
| emilmont | 1:fdd22bb7aa52 | 1322 | |
| emilmont | 1:fdd22bb7aa52 | 1323 | /* Read yc (real), xc(imag) input */ |
| mbed_official | 5:3762170b6d4d | 1324 | S = _SIMD32_OFFSET(pSi2); |
| emilmont | 1:fdd22bb7aa52 | 1325 | |
| emilmont | 1:fdd22bb7aa52 | 1326 | /* R = packed( (ya + yc), (xa + xc)) */ |
| emilmont | 1:fdd22bb7aa52 | 1327 | R = __QADD16(T, S); |
| emilmont | 1:fdd22bb7aa52 | 1328 | |
| emilmont | 1:fdd22bb7aa52 | 1329 | /* S = packed((ya - yc), (xa - xc)) */ |
| emilmont | 1:fdd22bb7aa52 | 1330 | S = __QSUB16(T, S); |
| emilmont | 1:fdd22bb7aa52 | 1331 | |
| emilmont | 1:fdd22bb7aa52 | 1332 | /* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 1333 | /* Read yb (real), xb(imag) input */ |
| mbed_official | 5:3762170b6d4d | 1334 | T = _SIMD32_OFFSET(pSi1); |
| emilmont | 1:fdd22bb7aa52 | 1335 | |
| emilmont | 1:fdd22bb7aa52 | 1336 | /* Read yd (real), xd(imag) input */ |
| mbed_official | 5:3762170b6d4d | 1337 | U = _SIMD32_OFFSET(pSi3); |
| emilmont | 1:fdd22bb7aa52 | 1338 | |
| emilmont | 1:fdd22bb7aa52 | 1339 | /* T = packed( (yb + yd), (xb + xd)) */ |
| emilmont | 1:fdd22bb7aa52 | 1340 | T = __QADD16(T, U); |
| emilmont | 1:fdd22bb7aa52 | 1341 | |
| emilmont | 1:fdd22bb7aa52 | 1342 | /* writing the butterfly processed i0 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1343 | |
| emilmont | 1:fdd22bb7aa52 | 1344 | /* xa' = xa + xb + xc + xd */ |
| emilmont | 1:fdd22bb7aa52 | 1345 | /* ya' = ya + yb + yc + yd */ |
| emilmont | 1:fdd22bb7aa52 | 1346 | out1 = __SHADD16(R, T); |
| mbed_official | 5:3762170b6d4d | 1347 | out1 = __SHADD16(out1, 0); |
| mbed_official | 5:3762170b6d4d | 1348 | _SIMD32_OFFSET(pSi0) = out1; |
| mbed_official | 5:3762170b6d4d | 1349 | pSi0 += 2 * n1; |
| emilmont | 1:fdd22bb7aa52 | 1350 | |
| emilmont | 1:fdd22bb7aa52 | 1351 | /* R = packed( (ya + yc) - (yb + yd), (xa + xc) - (xb + xd)) */ |
| emilmont | 1:fdd22bb7aa52 | 1352 | R = __SHSUB16(R, T); |
| emilmont | 1:fdd22bb7aa52 | 1353 | |
| emilmont | 1:fdd22bb7aa52 | 1354 | #ifndef ARM_MATH_BIG_ENDIAN |
| emilmont | 1:fdd22bb7aa52 | 1355 | |
| emilmont | 1:fdd22bb7aa52 | 1356 | /* (ya-yb+yc-yd)* (si2) + (xa-xb+xc-xd)* co2 */ |
| emilmont | 1:fdd22bb7aa52 | 1357 | out1 = __SMUSD(C2, R) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 1358 | |
| emilmont | 1:fdd22bb7aa52 | 1359 | /* (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 1360 | out2 = __SMUADX(C2, R); |
| emilmont | 1:fdd22bb7aa52 | 1361 | |
| emilmont | 1:fdd22bb7aa52 | 1362 | #else |
| emilmont | 1:fdd22bb7aa52 | 1363 | |
| emilmont | 1:fdd22bb7aa52 | 1364 | /* (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 1365 | out1 = __SMUADX(R, C2) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 1366 | |
| emilmont | 1:fdd22bb7aa52 | 1367 | /* (ya-yb+yc-yd)* (si2) + (xa-xb+xc-xd)* co2 */ |
| emilmont | 1:fdd22bb7aa52 | 1368 | out2 = __SMUSD(__QSUB16(0, C2), R); |
| emilmont | 1:fdd22bb7aa52 | 1369 | |
| emilmont | 1:fdd22bb7aa52 | 1370 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
| emilmont | 1:fdd22bb7aa52 | 1371 | |
| emilmont | 1:fdd22bb7aa52 | 1372 | /* Reading i0+3fftLen/4 */ |
| emilmont | 1:fdd22bb7aa52 | 1373 | /* Read yb (real), xb(imag) input */ |
| mbed_official | 5:3762170b6d4d | 1374 | T = _SIMD32_OFFSET(pSi1); |
| emilmont | 1:fdd22bb7aa52 | 1375 | |
| emilmont | 1:fdd22bb7aa52 | 1376 | /* writing the butterfly processed i0 + fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1377 | /* xc' = (xa-xb+xc-xd)* co2 + (ya-yb+yc-yd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 1378 | /* yc' = (ya-yb+yc-yd)* co2 - (xa-xb+xc-xd)* (si2) */ |
| mbed_official | 5:3762170b6d4d | 1379 | _SIMD32_OFFSET(pSi1) = |
| emilmont | 1:fdd22bb7aa52 | 1380 | ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| mbed_official | 5:3762170b6d4d | 1381 | pSi1 += 2 * n1; |
| emilmont | 1:fdd22bb7aa52 | 1382 | |
| emilmont | 1:fdd22bb7aa52 | 1383 | /* Butterfly calculations */ |
| emilmont | 1:fdd22bb7aa52 | 1384 | |
| emilmont | 1:fdd22bb7aa52 | 1385 | /* Read yd (real), xd(imag) input */ |
| mbed_official | 5:3762170b6d4d | 1386 | U = _SIMD32_OFFSET(pSi3); |
| emilmont | 1:fdd22bb7aa52 | 1387 | |
| emilmont | 1:fdd22bb7aa52 | 1388 | /* T = packed(yb-yd, xb-xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1389 | T = __QSUB16(T, U); |
| emilmont | 1:fdd22bb7aa52 | 1390 | |
| emilmont | 1:fdd22bb7aa52 | 1391 | #ifndef ARM_MATH_BIG_ENDIAN |
| emilmont | 1:fdd22bb7aa52 | 1392 | |
| emilmont | 1:fdd22bb7aa52 | 1393 | /* R = packed((ya-yc) + (xb- xd) , (xa-xc) - (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 1394 | R = __SHSAX(S, T); |
| emilmont | 1:fdd22bb7aa52 | 1395 | |
| emilmont | 1:fdd22bb7aa52 | 1396 | /* S = packed((ya-yc) - (xb- xd), (xa-xc) + (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 1397 | S = __SHASX(S, T); |
| emilmont | 1:fdd22bb7aa52 | 1398 | |
| emilmont | 1:fdd22bb7aa52 | 1399 | |
| emilmont | 1:fdd22bb7aa52 | 1400 | /* Butterfly process for the i0+fftLen/2 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1401 | out1 = __SMUSD(C1, S) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 1402 | out2 = __SMUADX(C1, S); |
| emilmont | 1:fdd22bb7aa52 | 1403 | |
| emilmont | 1:fdd22bb7aa52 | 1404 | #else |
| emilmont | 1:fdd22bb7aa52 | 1405 | |
| emilmont | 1:fdd22bb7aa52 | 1406 | /* R = packed((ya-yc) + (xb- xd) , (xa-xc) - (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 1407 | R = __SHASX(S, T); |
| emilmont | 1:fdd22bb7aa52 | 1408 | |
| emilmont | 1:fdd22bb7aa52 | 1409 | /* S = packed((ya-yc) - (xb- xd), (xa-xc) + (yb-yd)) */ |
| emilmont | 1:fdd22bb7aa52 | 1410 | S = __SHSAX(S, T); |
| emilmont | 1:fdd22bb7aa52 | 1411 | |
| emilmont | 1:fdd22bb7aa52 | 1412 | |
| emilmont | 1:fdd22bb7aa52 | 1413 | /* Butterfly process for the i0+fftLen/2 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1414 | out1 = __SMUADX(S, C1) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 1415 | out2 = __SMUSD(__QSUB16(0, C1), S); |
| emilmont | 1:fdd22bb7aa52 | 1416 | |
| emilmont | 1:fdd22bb7aa52 | 1417 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
| emilmont | 1:fdd22bb7aa52 | 1418 | |
| emilmont | 1:fdd22bb7aa52 | 1419 | /* xb' = (xa+yb-xc-yd)* co1 + (ya-xb-yc+xd)* (si1) */ |
| emilmont | 1:fdd22bb7aa52 | 1420 | /* yb' = (ya-xb-yc+xd)* co1 - (xa+yb-xc-yd)* (si1) */ |
| mbed_official | 5:3762170b6d4d | 1421 | _SIMD32_OFFSET(pSi2) = |
| emilmont | 1:fdd22bb7aa52 | 1422 | ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| mbed_official | 5:3762170b6d4d | 1423 | pSi2 += 2 * n1; |
| emilmont | 1:fdd22bb7aa52 | 1424 | |
| emilmont | 1:fdd22bb7aa52 | 1425 | /* Butterfly process for the i0+3fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1426 | |
| emilmont | 1:fdd22bb7aa52 | 1427 | #ifndef ARM_MATH_BIG_ENDIAN |
| emilmont | 1:fdd22bb7aa52 | 1428 | |
| emilmont | 1:fdd22bb7aa52 | 1429 | out1 = __SMUSD(C3, R) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 1430 | out2 = __SMUADX(C3, R); |
| emilmont | 1:fdd22bb7aa52 | 1431 | |
| emilmont | 1:fdd22bb7aa52 | 1432 | #else |
| emilmont | 1:fdd22bb7aa52 | 1433 | |
| emilmont | 1:fdd22bb7aa52 | 1434 | out1 = __SMUADX(C3, R) >> 16u; |
| emilmont | 1:fdd22bb7aa52 | 1435 | out2 = __SMUSD(__QSUB16(0, C3), R); |
| emilmont | 1:fdd22bb7aa52 | 1436 | |
| emilmont | 1:fdd22bb7aa52 | 1437 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
| emilmont | 1:fdd22bb7aa52 | 1438 | |
| emilmont | 1:fdd22bb7aa52 | 1439 | /* xd' = (xa-yb-xc+yd)* co3 + (ya+xb-yc-xd)* (si3) */ |
| emilmont | 1:fdd22bb7aa52 | 1440 | /* yd' = (ya+xb-yc-xd)* co3 - (xa-yb-xc+yd)* (si3) */ |
| mbed_official | 5:3762170b6d4d | 1441 | _SIMD32_OFFSET(pSi3) = |
| emilmont | 1:fdd22bb7aa52 | 1442 | ((out2) & 0xFFFF0000) | (out1 & 0x0000FFFF); |
| mbed_official | 5:3762170b6d4d | 1443 | pSi3 += 2 * n1; |
| emilmont | 1:fdd22bb7aa52 | 1444 | } |
| emilmont | 1:fdd22bb7aa52 | 1445 | } |
| emilmont | 1:fdd22bb7aa52 | 1446 | /* Twiddle coefficients index modifier */ |
| emilmont | 1:fdd22bb7aa52 | 1447 | twidCoefModifier <<= 2u; |
| emilmont | 1:fdd22bb7aa52 | 1448 | } |
| emilmont | 1:fdd22bb7aa52 | 1449 | /* end of middle stage process */ |
| emilmont | 1:fdd22bb7aa52 | 1450 | |
| emilmont | 1:fdd22bb7aa52 | 1451 | /* data is in 10.6(q6) format for the 1024 point */ |
| emilmont | 1:fdd22bb7aa52 | 1452 | /* data is in 8.8(q8) format for the 256 point */ |
| emilmont | 1:fdd22bb7aa52 | 1453 | /* data is in 6.10(q10) format for the 64 point */ |
| emilmont | 1:fdd22bb7aa52 | 1454 | /* data is in 4.12(q12) format for the 16 point */ |
| emilmont | 1:fdd22bb7aa52 | 1455 | |
| emilmont | 1:fdd22bb7aa52 | 1456 | /* Initializations for the last stage */ |
| emilmont | 1:fdd22bb7aa52 | 1457 | j = fftLen >> 2; |
| emilmont | 1:fdd22bb7aa52 | 1458 | |
| emilmont | 1:fdd22bb7aa52 | 1459 | ptr1 = &pSrc16[0]; |
| emilmont | 1:fdd22bb7aa52 | 1460 | |
| emilmont | 1:fdd22bb7aa52 | 1461 | /* start of last stage process */ |
| emilmont | 1:fdd22bb7aa52 | 1462 | |
| emilmont | 1:fdd22bb7aa52 | 1463 | /* Butterfly implementation */ |
| emilmont | 1:fdd22bb7aa52 | 1464 | do |
| emilmont | 1:fdd22bb7aa52 | 1465 | { |
| emilmont | 1:fdd22bb7aa52 | 1466 | /* Read xa (real), ya(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1467 | xaya = *__SIMD32(ptr1)++; |
| emilmont | 1:fdd22bb7aa52 | 1468 | |
| emilmont | 1:fdd22bb7aa52 | 1469 | /* Read xb (real), yb(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1470 | xbyb = *__SIMD32(ptr1)++; |
| emilmont | 1:fdd22bb7aa52 | 1471 | |
| emilmont | 1:fdd22bb7aa52 | 1472 | /* Read xc (real), yc(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1473 | xcyc = *__SIMD32(ptr1)++; |
| emilmont | 1:fdd22bb7aa52 | 1474 | |
| emilmont | 1:fdd22bb7aa52 | 1475 | /* Read xd (real), yd(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1476 | xdyd = *__SIMD32(ptr1)++; |
| emilmont | 1:fdd22bb7aa52 | 1477 | |
| emilmont | 1:fdd22bb7aa52 | 1478 | /* R = packed((ya + yc), (xa + xc)) */ |
| emilmont | 1:fdd22bb7aa52 | 1479 | R = __QADD16(xaya, xcyc); |
| emilmont | 1:fdd22bb7aa52 | 1480 | |
| emilmont | 1:fdd22bb7aa52 | 1481 | /* T = packed((yb + yd), (xb + xd)) */ |
| emilmont | 1:fdd22bb7aa52 | 1482 | T = __QADD16(xbyb, xdyd); |
| emilmont | 1:fdd22bb7aa52 | 1483 | |
| emilmont | 1:fdd22bb7aa52 | 1484 | /* pointer updation for writing */ |
| emilmont | 1:fdd22bb7aa52 | 1485 | ptr1 = ptr1 - 8u; |
| emilmont | 1:fdd22bb7aa52 | 1486 | |
| emilmont | 1:fdd22bb7aa52 | 1487 | |
| emilmont | 1:fdd22bb7aa52 | 1488 | /* xa' = xa + xb + xc + xd */ |
| emilmont | 1:fdd22bb7aa52 | 1489 | /* ya' = ya + yb + yc + yd */ |
| emilmont | 1:fdd22bb7aa52 | 1490 | *__SIMD32(ptr1)++ = __SHADD16(R, T); |
| emilmont | 1:fdd22bb7aa52 | 1491 | |
| emilmont | 1:fdd22bb7aa52 | 1492 | /* T = packed((yb + yd), (xb + xd)) */ |
| emilmont | 1:fdd22bb7aa52 | 1493 | T = __QADD16(xbyb, xdyd); |
| emilmont | 1:fdd22bb7aa52 | 1494 | |
| emilmont | 1:fdd22bb7aa52 | 1495 | /* xc' = (xa-xb+xc-xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1496 | /* yc' = (ya-yb+yc-yd) */ |
| emilmont | 1:fdd22bb7aa52 | 1497 | *__SIMD32(ptr1)++ = __SHSUB16(R, T); |
| emilmont | 1:fdd22bb7aa52 | 1498 | |
| emilmont | 1:fdd22bb7aa52 | 1499 | /* S = packed((ya - yc), (xa - xc)) */ |
| emilmont | 1:fdd22bb7aa52 | 1500 | S = __QSUB16(xaya, xcyc); |
| emilmont | 1:fdd22bb7aa52 | 1501 | |
| emilmont | 1:fdd22bb7aa52 | 1502 | /* Read yd (real), xd(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1503 | /* T = packed( (yb - yd), (xb - xd)) */ |
| emilmont | 1:fdd22bb7aa52 | 1504 | U = __QSUB16(xbyb, xdyd); |
| emilmont | 1:fdd22bb7aa52 | 1505 | |
| emilmont | 1:fdd22bb7aa52 | 1506 | #ifndef ARM_MATH_BIG_ENDIAN |
| emilmont | 1:fdd22bb7aa52 | 1507 | |
| emilmont | 1:fdd22bb7aa52 | 1508 | /* xb' = (xa+yb-xc-yd) */ |
| emilmont | 1:fdd22bb7aa52 | 1509 | /* yb' = (ya-xb-yc+xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1510 | *__SIMD32(ptr1)++ = __SHASX(S, U); |
| emilmont | 1:fdd22bb7aa52 | 1511 | |
| emilmont | 1:fdd22bb7aa52 | 1512 | |
| emilmont | 1:fdd22bb7aa52 | 1513 | /* xd' = (xa-yb-xc+yd) */ |
| emilmont | 1:fdd22bb7aa52 | 1514 | /* yd' = (ya+xb-yc-xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1515 | *__SIMD32(ptr1)++ = __SHSAX(S, U); |
| emilmont | 1:fdd22bb7aa52 | 1516 | |
| emilmont | 1:fdd22bb7aa52 | 1517 | #else |
| emilmont | 1:fdd22bb7aa52 | 1518 | |
| emilmont | 1:fdd22bb7aa52 | 1519 | /* xb' = (xa+yb-xc-yd) */ |
| emilmont | 1:fdd22bb7aa52 | 1520 | /* yb' = (ya-xb-yc+xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1521 | *__SIMD32(ptr1)++ = __SHSAX(S, U); |
| emilmont | 1:fdd22bb7aa52 | 1522 | |
| emilmont | 1:fdd22bb7aa52 | 1523 | |
| emilmont | 1:fdd22bb7aa52 | 1524 | /* xd' = (xa-yb-xc+yd) */ |
| emilmont | 1:fdd22bb7aa52 | 1525 | /* yd' = (ya+xb-yc-xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1526 | *__SIMD32(ptr1)++ = __SHASX(S, U); |
| emilmont | 1:fdd22bb7aa52 | 1527 | |
| emilmont | 1:fdd22bb7aa52 | 1528 | |
| emilmont | 1:fdd22bb7aa52 | 1529 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
| emilmont | 1:fdd22bb7aa52 | 1530 | |
| emilmont | 1:fdd22bb7aa52 | 1531 | } while(--j); |
| emilmont | 1:fdd22bb7aa52 | 1532 | |
| emilmont | 1:fdd22bb7aa52 | 1533 | /* end of last stage process */ |
| emilmont | 1:fdd22bb7aa52 | 1534 | |
| emilmont | 1:fdd22bb7aa52 | 1535 | /* output is in 11.5(q5) format for the 1024 point */ |
| emilmont | 1:fdd22bb7aa52 | 1536 | /* output is in 9.7(q7) format for the 256 point */ |
| emilmont | 1:fdd22bb7aa52 | 1537 | /* output is in 7.9(q9) format for the 64 point */ |
| emilmont | 1:fdd22bb7aa52 | 1538 | /* output is in 5.11(q11) format for the 16 point */ |
| emilmont | 1:fdd22bb7aa52 | 1539 | |
| emilmont | 1:fdd22bb7aa52 | 1540 | |
| emilmont | 1:fdd22bb7aa52 | 1541 | #else |
| emilmont | 1:fdd22bb7aa52 | 1542 | |
| emilmont | 1:fdd22bb7aa52 | 1543 | /* Run the below code for Cortex-M0 */ |
| emilmont | 1:fdd22bb7aa52 | 1544 | |
| emilmont | 1:fdd22bb7aa52 | 1545 | q15_t R0, R1, S0, S1, T0, T1, U0, U1; |
| emilmont | 1:fdd22bb7aa52 | 1546 | q15_t Co1, Si1, Co2, Si2, Co3, Si3, out1, out2; |
| emilmont | 1:fdd22bb7aa52 | 1547 | uint32_t n1, n2, ic, i0, i1, i2, i3, j, k; |
| emilmont | 1:fdd22bb7aa52 | 1548 | |
| emilmont | 1:fdd22bb7aa52 | 1549 | /* Total process is divided into three stages */ |
| emilmont | 1:fdd22bb7aa52 | 1550 | |
| emilmont | 1:fdd22bb7aa52 | 1551 | /* process first stage, middle stages, & last stage */ |
| emilmont | 1:fdd22bb7aa52 | 1552 | |
| emilmont | 1:fdd22bb7aa52 | 1553 | /* Initializations for the first stage */ |
| emilmont | 1:fdd22bb7aa52 | 1554 | n2 = fftLen; |
| emilmont | 1:fdd22bb7aa52 | 1555 | n1 = n2; |
| emilmont | 1:fdd22bb7aa52 | 1556 | |
| emilmont | 1:fdd22bb7aa52 | 1557 | /* n2 = fftLen/4 */ |
| emilmont | 1:fdd22bb7aa52 | 1558 | n2 >>= 2u; |
| emilmont | 1:fdd22bb7aa52 | 1559 | |
| emilmont | 1:fdd22bb7aa52 | 1560 | /* Index for twiddle coefficient */ |
| emilmont | 1:fdd22bb7aa52 | 1561 | ic = 0u; |
| emilmont | 1:fdd22bb7aa52 | 1562 | |
| emilmont | 1:fdd22bb7aa52 | 1563 | /* Index for input read and output write */ |
| emilmont | 1:fdd22bb7aa52 | 1564 | i0 = 0u; |
| emilmont | 1:fdd22bb7aa52 | 1565 | |
| emilmont | 1:fdd22bb7aa52 | 1566 | j = n2; |
| emilmont | 1:fdd22bb7aa52 | 1567 | |
| emilmont | 1:fdd22bb7aa52 | 1568 | /* Input is in 1.15(q15) format */ |
| emilmont | 1:fdd22bb7aa52 | 1569 | |
| emilmont | 1:fdd22bb7aa52 | 1570 | /* Start of first stage process */ |
| emilmont | 1:fdd22bb7aa52 | 1571 | do |
| emilmont | 1:fdd22bb7aa52 | 1572 | { |
| emilmont | 1:fdd22bb7aa52 | 1573 | /* Butterfly implementation */ |
| emilmont | 1:fdd22bb7aa52 | 1574 | |
| emilmont | 1:fdd22bb7aa52 | 1575 | /* index calculation for the input as, */ |
| emilmont | 1:fdd22bb7aa52 | 1576 | /* pSrc16[i0 + 0], pSrc16[i0 + fftLen/4], pSrc16[i0 + fftLen/2], pSrc16[i0 + 3fftLen/4] */ |
| emilmont | 1:fdd22bb7aa52 | 1577 | i1 = i0 + n2; |
| emilmont | 1:fdd22bb7aa52 | 1578 | i2 = i1 + n2; |
| emilmont | 1:fdd22bb7aa52 | 1579 | i3 = i2 + n2; |
| emilmont | 1:fdd22bb7aa52 | 1580 | |
| emilmont | 1:fdd22bb7aa52 | 1581 | /* Reading i0, i0+fftLen/2 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 1582 | /* input is down scale by 4 to avoid overflow */ |
| emilmont | 1:fdd22bb7aa52 | 1583 | /* Read ya (real), xa(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1584 | T0 = pSrc16[i0 * 2u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 1585 | T1 = pSrc16[(i0 * 2u) + 1u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 1586 | /* input is down scale by 4 to avoid overflow */ |
| emilmont | 1:fdd22bb7aa52 | 1587 | /* Read yc (real), xc(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1588 | S0 = pSrc16[i2 * 2u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 1589 | S1 = pSrc16[(i2 * 2u) + 1u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 1590 | |
| emilmont | 1:fdd22bb7aa52 | 1591 | /* R0 = (ya + yc), R1 = (xa + xc) */ |
| emilmont | 1:fdd22bb7aa52 | 1592 | R0 = __SSAT(T0 + S0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1593 | R1 = __SSAT(T1 + S1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1594 | /* S0 = (ya - yc), S1 = (xa - xc) */ |
| emilmont | 1:fdd22bb7aa52 | 1595 | S0 = __SSAT(T0 - S0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1596 | S1 = __SSAT(T1 - S1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1597 | |
| emilmont | 1:fdd22bb7aa52 | 1598 | /* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 1599 | /* input is down scale by 4 to avoid overflow */ |
| emilmont | 1:fdd22bb7aa52 | 1600 | /* Read yb (real), xb(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1601 | T0 = pSrc16[i1 * 2u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 1602 | T1 = pSrc16[(i1 * 2u) + 1u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 1603 | /* Read yd (real), xd(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1604 | /* input is down scale by 4 to avoid overflow */ |
| emilmont | 1:fdd22bb7aa52 | 1605 | U0 = pSrc16[i3 * 2u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 1606 | U1 = pSrc16[(i3 * 2u) + 1u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 1607 | |
| emilmont | 1:fdd22bb7aa52 | 1608 | /* T0 = (yb + yd), T1 = (xb + xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1609 | T0 = __SSAT(T0 + U0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1610 | T1 = __SSAT(T1 + U1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1611 | |
| emilmont | 1:fdd22bb7aa52 | 1612 | /* writing the butterfly processed i0 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1613 | /* xa' = xa + xb + xc + xd */ |
| emilmont | 1:fdd22bb7aa52 | 1614 | /* ya' = ya + yb + yc + yd */ |
| emilmont | 1:fdd22bb7aa52 | 1615 | pSrc16[i0 * 2u] = (R0 >> 1u) + (T0 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1616 | pSrc16[(i0 * 2u) + 1u] = (R1 >> 1u) + (T1 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1617 | |
| emilmont | 1:fdd22bb7aa52 | 1618 | /* R0 = (ya + yc) - (yb + yd), R1 = (xa + xc)- (xb + xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1619 | R0 = __SSAT(R0 - T0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1620 | R1 = __SSAT(R1 - T1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1621 | /* co2 & si2 are read from Coefficient pointer */ |
| emilmont | 1:fdd22bb7aa52 | 1622 | Co2 = pCoef16[2u * ic * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1623 | Si2 = pCoef16[(2u * ic * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1624 | /* xc' = (xa-xb+xc-xd)* co2 - (ya-yb+yc-yd)* (si2) */ |
| mbed_official | 5:3762170b6d4d | 1625 | out1 = (q15_t) ((Co2 * R0 - Si2 * R1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 1626 | /* yc' = (ya-yb+yc-yd)* co2 + (xa-xb+xc-xd)* (si2) */ |
| mbed_official | 5:3762170b6d4d | 1627 | out2 = (q15_t) ((Si2 * R0 + Co2 * R1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 1628 | |
| emilmont | 1:fdd22bb7aa52 | 1629 | /* Reading i0+fftLen/4 */ |
| emilmont | 1:fdd22bb7aa52 | 1630 | /* input is down scale by 4 to avoid overflow */ |
| emilmont | 1:fdd22bb7aa52 | 1631 | /* T0 = yb, T1 = xb */ |
| emilmont | 1:fdd22bb7aa52 | 1632 | T0 = pSrc16[i1 * 2u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 1633 | T1 = pSrc16[(i1 * 2u) + 1u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 1634 | |
| emilmont | 1:fdd22bb7aa52 | 1635 | /* writing the butterfly processed i0 + fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1636 | /* writing output(xc', yc') in little endian format */ |
| emilmont | 1:fdd22bb7aa52 | 1637 | pSrc16[i1 * 2u] = out1; |
| emilmont | 1:fdd22bb7aa52 | 1638 | pSrc16[(i1 * 2u) + 1u] = out2; |
| emilmont | 1:fdd22bb7aa52 | 1639 | |
| emilmont | 1:fdd22bb7aa52 | 1640 | /* Butterfly calculations */ |
| emilmont | 1:fdd22bb7aa52 | 1641 | /* input is down scale by 4 to avoid overflow */ |
| emilmont | 1:fdd22bb7aa52 | 1642 | /* U0 = yd, U1 = xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1643 | U0 = pSrc16[i3 * 2u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 1644 | U1 = pSrc16[(i3 * 2u) + 1u] >> 2u; |
| emilmont | 1:fdd22bb7aa52 | 1645 | |
| emilmont | 1:fdd22bb7aa52 | 1646 | /* T0 = yb-yd, T1 = xb-xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1647 | T0 = __SSAT(T0 - U0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1648 | T1 = __SSAT(T1 - U1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1649 | /* R0 = (ya-yc) - (xb- xd) , R1 = (xa-xc) + (yb-yd) */ |
| mbed_official | 5:3762170b6d4d | 1650 | R0 = (q15_t) __SSAT((q31_t) (S0 + T1), 16); |
| mbed_official | 5:3762170b6d4d | 1651 | R1 = (q15_t) __SSAT((q31_t) (S1 - T0), 16); |
| emilmont | 1:fdd22bb7aa52 | 1652 | /* S = (ya-yc) + (xb- xd), S1 = (xa-xc) - (yb-yd) */ |
| mbed_official | 5:3762170b6d4d | 1653 | S0 = (q15_t) __SSAT((q31_t) (S0 - T1), 16); |
| mbed_official | 5:3762170b6d4d | 1654 | S1 = (q15_t) __SSAT((q31_t) (S1 + T0), 16); |
| emilmont | 1:fdd22bb7aa52 | 1655 | |
| emilmont | 1:fdd22bb7aa52 | 1656 | /* co1 & si1 are read from Coefficient pointer */ |
| emilmont | 1:fdd22bb7aa52 | 1657 | Co1 = pCoef16[ic * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1658 | Si1 = pCoef16[(ic * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1659 | /* Butterfly process for the i0+fftLen/2 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1660 | /* xb' = (xa-yb-xc+yd)* co1 - (ya+xb-yc-xd)* (si1) */ |
| mbed_official | 5:3762170b6d4d | 1661 | out1 = (q15_t) ((Co1 * S0 - Si1 * S1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 1662 | /* yb' = (ya+xb-yc-xd)* co1 + (xa-yb-xc+yd)* (si1) */ |
| mbed_official | 5:3762170b6d4d | 1663 | out2 = (q15_t) ((Si1 * S0 + Co1 * S1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 1664 | /* writing output(xb', yb') in little endian format */ |
| emilmont | 1:fdd22bb7aa52 | 1665 | pSrc16[i2 * 2u] = out1; |
| emilmont | 1:fdd22bb7aa52 | 1666 | pSrc16[(i2 * 2u) + 1u] = out2; |
| emilmont | 1:fdd22bb7aa52 | 1667 | |
| emilmont | 1:fdd22bb7aa52 | 1668 | /* Co3 & si3 are read from Coefficient pointer */ |
| emilmont | 1:fdd22bb7aa52 | 1669 | Co3 = pCoef16[3u * ic * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1670 | Si3 = pCoef16[(3u * ic * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1671 | /* Butterfly process for the i0+3fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1672 | /* xd' = (xa+yb-xc-yd)* Co3 - (ya-xb-yc+xd)* (si3) */ |
| mbed_official | 5:3762170b6d4d | 1673 | out1 = (q15_t) ((Co3 * R0 - Si3 * R1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 1674 | /* yd' = (ya-xb-yc+xd)* Co3 + (xa+yb-xc-yd)* (si3) */ |
| mbed_official | 5:3762170b6d4d | 1675 | out2 = (q15_t) ((Si3 * R0 + Co3 * R1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 1676 | /* writing output(xd', yd') in little endian format */ |
| emilmont | 1:fdd22bb7aa52 | 1677 | pSrc16[i3 * 2u] = out1; |
| emilmont | 1:fdd22bb7aa52 | 1678 | pSrc16[(i3 * 2u) + 1u] = out2; |
| emilmont | 1:fdd22bb7aa52 | 1679 | |
| emilmont | 1:fdd22bb7aa52 | 1680 | /* Twiddle coefficients index modifier */ |
| emilmont | 1:fdd22bb7aa52 | 1681 | ic = ic + twidCoefModifier; |
| emilmont | 1:fdd22bb7aa52 | 1682 | |
| emilmont | 1:fdd22bb7aa52 | 1683 | /* Updating input index */ |
| emilmont | 1:fdd22bb7aa52 | 1684 | i0 = i0 + 1u; |
| emilmont | 1:fdd22bb7aa52 | 1685 | |
| emilmont | 1:fdd22bb7aa52 | 1686 | } while(--j); |
| emilmont | 1:fdd22bb7aa52 | 1687 | |
| emilmont | 1:fdd22bb7aa52 | 1688 | /* End of first stage process */ |
| emilmont | 1:fdd22bb7aa52 | 1689 | |
| emilmont | 1:fdd22bb7aa52 | 1690 | /* data is in 4.11(q11) format */ |
| emilmont | 1:fdd22bb7aa52 | 1691 | |
| emilmont | 1:fdd22bb7aa52 | 1692 | |
| emilmont | 1:fdd22bb7aa52 | 1693 | /* Start of Middle stage process */ |
| emilmont | 1:fdd22bb7aa52 | 1694 | |
| emilmont | 1:fdd22bb7aa52 | 1695 | /* Twiddle coefficients index modifier */ |
| emilmont | 1:fdd22bb7aa52 | 1696 | twidCoefModifier <<= 2u; |
| emilmont | 1:fdd22bb7aa52 | 1697 | |
| emilmont | 1:fdd22bb7aa52 | 1698 | /* Calculation of Middle stage */ |
| emilmont | 1:fdd22bb7aa52 | 1699 | for (k = fftLen / 4u; k > 4u; k >>= 2u) |
| emilmont | 1:fdd22bb7aa52 | 1700 | { |
| emilmont | 1:fdd22bb7aa52 | 1701 | /* Initializations for the middle stage */ |
| emilmont | 1:fdd22bb7aa52 | 1702 | n1 = n2; |
| emilmont | 1:fdd22bb7aa52 | 1703 | n2 >>= 2u; |
| emilmont | 1:fdd22bb7aa52 | 1704 | ic = 0u; |
| emilmont | 1:fdd22bb7aa52 | 1705 | |
| emilmont | 1:fdd22bb7aa52 | 1706 | for (j = 0u; j <= (n2 - 1u); j++) |
| emilmont | 1:fdd22bb7aa52 | 1707 | { |
| emilmont | 1:fdd22bb7aa52 | 1708 | /* index calculation for the coefficients */ |
| emilmont | 1:fdd22bb7aa52 | 1709 | Co1 = pCoef16[ic * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1710 | Si1 = pCoef16[(ic * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1711 | Co2 = pCoef16[2u * ic * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1712 | Si2 = pCoef16[2u * ic * 2u + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1713 | Co3 = pCoef16[3u * ic * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1714 | Si3 = pCoef16[(3u * ic * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1715 | |
| emilmont | 1:fdd22bb7aa52 | 1716 | /* Twiddle coefficients index modifier */ |
| emilmont | 1:fdd22bb7aa52 | 1717 | ic = ic + twidCoefModifier; |
| emilmont | 1:fdd22bb7aa52 | 1718 | |
| emilmont | 1:fdd22bb7aa52 | 1719 | /* Butterfly implementation */ |
| emilmont | 1:fdd22bb7aa52 | 1720 | for (i0 = j; i0 < fftLen; i0 += n1) |
| emilmont | 1:fdd22bb7aa52 | 1721 | { |
| emilmont | 1:fdd22bb7aa52 | 1722 | /* index calculation for the input as, */ |
| emilmont | 1:fdd22bb7aa52 | 1723 | /* pSrc16[i0 + 0], pSrc16[i0 + fftLen/4], pSrc16[i0 + fftLen/2], pSrc16[i0 + 3fftLen/4] */ |
| emilmont | 1:fdd22bb7aa52 | 1724 | i1 = i0 + n2; |
| emilmont | 1:fdd22bb7aa52 | 1725 | i2 = i1 + n2; |
| emilmont | 1:fdd22bb7aa52 | 1726 | i3 = i2 + n2; |
| emilmont | 1:fdd22bb7aa52 | 1727 | |
| emilmont | 1:fdd22bb7aa52 | 1728 | /* Reading i0, i0+fftLen/2 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 1729 | /* Read ya (real), xa(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1730 | T0 = pSrc16[i0 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1731 | T1 = pSrc16[(i0 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1732 | |
| emilmont | 1:fdd22bb7aa52 | 1733 | /* Read yc (real), xc(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1734 | S0 = pSrc16[i2 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1735 | S1 = pSrc16[(i2 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1736 | |
| emilmont | 1:fdd22bb7aa52 | 1737 | |
| emilmont | 1:fdd22bb7aa52 | 1738 | /* R0 = (ya + yc), R1 = (xa + xc) */ |
| emilmont | 1:fdd22bb7aa52 | 1739 | R0 = __SSAT(T0 + S0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1740 | R1 = __SSAT(T1 + S1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1741 | /* S0 = (ya - yc), S1 = (xa - xc) */ |
| emilmont | 1:fdd22bb7aa52 | 1742 | S0 = __SSAT(T0 - S0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1743 | S1 = __SSAT(T1 - S1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1744 | |
| emilmont | 1:fdd22bb7aa52 | 1745 | /* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 1746 | /* Read yb (real), xb(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1747 | T0 = pSrc16[i1 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1748 | T1 = pSrc16[(i1 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1749 | |
| emilmont | 1:fdd22bb7aa52 | 1750 | /* Read yd (real), xd(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1751 | U0 = pSrc16[i3 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1752 | U1 = pSrc16[(i3 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1753 | |
| emilmont | 1:fdd22bb7aa52 | 1754 | /* T0 = (yb + yd), T1 = (xb + xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1755 | T0 = __SSAT(T0 + U0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1756 | T1 = __SSAT(T1 + U1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1757 | |
| emilmont | 1:fdd22bb7aa52 | 1758 | /* writing the butterfly processed i0 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1759 | /* xa' = xa + xb + xc + xd */ |
| emilmont | 1:fdd22bb7aa52 | 1760 | /* ya' = ya + yb + yc + yd */ |
| emilmont | 1:fdd22bb7aa52 | 1761 | pSrc16[i0 * 2u] = ((R0 >> 1u) + (T0 >> 1u)) >> 1u; |
| emilmont | 1:fdd22bb7aa52 | 1762 | pSrc16[(i0 * 2u) + 1u] = ((R1 >> 1u) + (T1 >> 1u)) >> 1u; |
| emilmont | 1:fdd22bb7aa52 | 1763 | |
| emilmont | 1:fdd22bb7aa52 | 1764 | /* R0 = (ya + yc) - (yb + yd), R1 = (xa + xc) - (xb + xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1765 | R0 = (R0 >> 1u) - (T0 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1766 | R1 = (R1 >> 1u) - (T1 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1767 | |
| emilmont | 1:fdd22bb7aa52 | 1768 | /* (ya-yb+yc-yd)* (si2) - (xa-xb+xc-xd)* co2 */ |
| mbed_official | 5:3762170b6d4d | 1769 | out1 = (q15_t) ((Co2 * R0 - Si2 * R1) >> 16); |
| emilmont | 1:fdd22bb7aa52 | 1770 | /* (ya-yb+yc-yd)* co2 + (xa-xb+xc-xd)* (si2) */ |
| mbed_official | 5:3762170b6d4d | 1771 | out2 = (q15_t) ((Si2 * R0 + Co2 * R1) >> 16); |
| emilmont | 1:fdd22bb7aa52 | 1772 | |
| emilmont | 1:fdd22bb7aa52 | 1773 | /* Reading i0+3fftLen/4 */ |
| emilmont | 1:fdd22bb7aa52 | 1774 | /* Read yb (real), xb(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1775 | T0 = pSrc16[i1 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1776 | T1 = pSrc16[(i1 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1777 | |
| emilmont | 1:fdd22bb7aa52 | 1778 | /* writing the butterfly processed i0 + fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1779 | /* xc' = (xa-xb+xc-xd)* co2 - (ya-yb+yc-yd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 1780 | /* yc' = (ya-yb+yc-yd)* co2 + (xa-xb+xc-xd)* (si2) */ |
| emilmont | 1:fdd22bb7aa52 | 1781 | pSrc16[i1 * 2u] = out1; |
| emilmont | 1:fdd22bb7aa52 | 1782 | pSrc16[(i1 * 2u) + 1u] = out2; |
| emilmont | 1:fdd22bb7aa52 | 1783 | |
| emilmont | 1:fdd22bb7aa52 | 1784 | /* Butterfly calculations */ |
| emilmont | 1:fdd22bb7aa52 | 1785 | /* Read yd (real), xd(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1786 | U0 = pSrc16[i3 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1787 | U1 = pSrc16[(i3 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1788 | |
| emilmont | 1:fdd22bb7aa52 | 1789 | /* T0 = yb-yd, T1 = xb-xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1790 | T0 = __SSAT(T0 - U0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1791 | T1 = __SSAT(T1 - U1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1792 | |
| emilmont | 1:fdd22bb7aa52 | 1793 | /* R0 = (ya-yc) - (xb- xd) , R1 = (xa-xc) + (yb-yd) */ |
| emilmont | 1:fdd22bb7aa52 | 1794 | R0 = (S0 >> 1u) + (T1 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1795 | R1 = (S1 >> 1u) - (T0 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1796 | |
| emilmont | 1:fdd22bb7aa52 | 1797 | /* S1 = (ya-yc) + (xb- xd), S1 = (xa-xc) - (yb-yd) */ |
| emilmont | 1:fdd22bb7aa52 | 1798 | S0 = (S0 >> 1u) - (T1 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1799 | S1 = (S1 >> 1u) + (T0 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1800 | |
| emilmont | 1:fdd22bb7aa52 | 1801 | /* Butterfly process for the i0+fftLen/2 sample */ |
| mbed_official | 5:3762170b6d4d | 1802 | out1 = (q15_t) ((Co1 * S0 - Si1 * S1) >> 16u); |
| mbed_official | 5:3762170b6d4d | 1803 | out2 = (q15_t) ((Si1 * S0 + Co1 * S1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 1804 | /* xb' = (xa-yb-xc+yd)* co1 - (ya+xb-yc-xd)* (si1) */ |
| emilmont | 1:fdd22bb7aa52 | 1805 | /* yb' = (ya+xb-yc-xd)* co1 + (xa-yb-xc+yd)* (si1) */ |
| emilmont | 1:fdd22bb7aa52 | 1806 | pSrc16[i2 * 2u] = out1; |
| emilmont | 1:fdd22bb7aa52 | 1807 | pSrc16[(i2 * 2u) + 1u] = out2; |
| emilmont | 1:fdd22bb7aa52 | 1808 | |
| emilmont | 1:fdd22bb7aa52 | 1809 | /* Butterfly process for the i0+3fftLen/4 sample */ |
| mbed_official | 5:3762170b6d4d | 1810 | out1 = (q15_t) ((Co3 * R0 - Si3 * R1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 1811 | |
| mbed_official | 5:3762170b6d4d | 1812 | out2 = (q15_t) ((Si3 * R0 + Co3 * R1) >> 16u); |
| emilmont | 1:fdd22bb7aa52 | 1813 | /* xd' = (xa+yb-xc-yd)* Co3 - (ya-xb-yc+xd)* (si3) */ |
| emilmont | 1:fdd22bb7aa52 | 1814 | /* yd' = (ya-xb-yc+xd)* Co3 + (xa+yb-xc-yd)* (si3) */ |
| emilmont | 1:fdd22bb7aa52 | 1815 | pSrc16[i3 * 2u] = out1; |
| emilmont | 1:fdd22bb7aa52 | 1816 | pSrc16[(i3 * 2u) + 1u] = out2; |
| emilmont | 1:fdd22bb7aa52 | 1817 | |
| emilmont | 1:fdd22bb7aa52 | 1818 | |
| emilmont | 1:fdd22bb7aa52 | 1819 | } |
| emilmont | 1:fdd22bb7aa52 | 1820 | } |
| emilmont | 1:fdd22bb7aa52 | 1821 | /* Twiddle coefficients index modifier */ |
| emilmont | 1:fdd22bb7aa52 | 1822 | twidCoefModifier <<= 2u; |
| emilmont | 1:fdd22bb7aa52 | 1823 | } |
| emilmont | 1:fdd22bb7aa52 | 1824 | /* End of Middle stages process */ |
| emilmont | 1:fdd22bb7aa52 | 1825 | |
| emilmont | 1:fdd22bb7aa52 | 1826 | |
| emilmont | 1:fdd22bb7aa52 | 1827 | /* data is in 10.6(q6) format for the 1024 point */ |
| emilmont | 1:fdd22bb7aa52 | 1828 | /* data is in 8.8(q8) format for the 256 point */ |
| emilmont | 1:fdd22bb7aa52 | 1829 | /* data is in 6.10(q10) format for the 64 point */ |
| emilmont | 1:fdd22bb7aa52 | 1830 | /* data is in 4.12(q12) format for the 16 point */ |
| emilmont | 1:fdd22bb7aa52 | 1831 | |
| emilmont | 1:fdd22bb7aa52 | 1832 | /* start of last stage process */ |
| emilmont | 1:fdd22bb7aa52 | 1833 | |
| emilmont | 1:fdd22bb7aa52 | 1834 | |
| emilmont | 1:fdd22bb7aa52 | 1835 | /* Initializations for the last stage */ |
| emilmont | 1:fdd22bb7aa52 | 1836 | n1 = n2; |
| emilmont | 1:fdd22bb7aa52 | 1837 | n2 >>= 2u; |
| emilmont | 1:fdd22bb7aa52 | 1838 | |
| emilmont | 1:fdd22bb7aa52 | 1839 | /* Butterfly implementation */ |
| emilmont | 1:fdd22bb7aa52 | 1840 | for (i0 = 0u; i0 <= (fftLen - n1); i0 += n1) |
| emilmont | 1:fdd22bb7aa52 | 1841 | { |
| emilmont | 1:fdd22bb7aa52 | 1842 | /* index calculation for the input as, */ |
| emilmont | 1:fdd22bb7aa52 | 1843 | /* pSrc16[i0 + 0], pSrc16[i0 + fftLen/4], pSrc16[i0 + fftLen/2], pSrc16[i0 + 3fftLen/4] */ |
| emilmont | 1:fdd22bb7aa52 | 1844 | i1 = i0 + n2; |
| emilmont | 1:fdd22bb7aa52 | 1845 | i2 = i1 + n2; |
| emilmont | 1:fdd22bb7aa52 | 1846 | i3 = i2 + n2; |
| emilmont | 1:fdd22bb7aa52 | 1847 | |
| emilmont | 1:fdd22bb7aa52 | 1848 | /* Reading i0, i0+fftLen/2 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 1849 | /* Read ya (real), xa(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1850 | T0 = pSrc16[i0 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1851 | T1 = pSrc16[(i0 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1852 | /* Read yc (real), xc(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1853 | S0 = pSrc16[i2 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1854 | S1 = pSrc16[(i2 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1855 | |
| emilmont | 1:fdd22bb7aa52 | 1856 | /* R0 = (ya + yc), R1 = (xa + xc) */ |
| emilmont | 1:fdd22bb7aa52 | 1857 | R0 = __SSAT(T0 + S0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1858 | R1 = __SSAT(T1 + S1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1859 | /* S0 = (ya - yc), S1 = (xa - xc) */ |
| emilmont | 1:fdd22bb7aa52 | 1860 | S0 = __SSAT(T0 - S0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1861 | S1 = __SSAT(T1 - S1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1862 | |
| emilmont | 1:fdd22bb7aa52 | 1863 | /* Reading i0+fftLen/4 , i0+3fftLen/4 inputs */ |
| emilmont | 1:fdd22bb7aa52 | 1864 | /* Read yb (real), xb(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1865 | T0 = pSrc16[i1 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1866 | T1 = pSrc16[(i1 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1867 | /* Read yd (real), xd(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1868 | U0 = pSrc16[i3 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1869 | U1 = pSrc16[(i3 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1870 | |
| emilmont | 1:fdd22bb7aa52 | 1871 | /* T0 = (yb + yd), T1 = (xb + xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1872 | T0 = __SSAT(T0 + U0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1873 | T1 = __SSAT(T1 + U1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1874 | |
| emilmont | 1:fdd22bb7aa52 | 1875 | /* writing the butterfly processed i0 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1876 | /* xa' = xa + xb + xc + xd */ |
| emilmont | 1:fdd22bb7aa52 | 1877 | /* ya' = ya + yb + yc + yd */ |
| emilmont | 1:fdd22bb7aa52 | 1878 | pSrc16[i0 * 2u] = (R0 >> 1u) + (T0 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1879 | pSrc16[(i0 * 2u) + 1u] = (R1 >> 1u) + (T1 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1880 | |
| emilmont | 1:fdd22bb7aa52 | 1881 | /* R0 = (ya + yc) - (yb + yd), R1 = (xa + xc) - (xb + xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1882 | R0 = (R0 >> 1u) - (T0 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1883 | R1 = (R1 >> 1u) - (T1 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1884 | |
| emilmont | 1:fdd22bb7aa52 | 1885 | /* Read yb (real), xb(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1886 | T0 = pSrc16[i1 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1887 | T1 = pSrc16[(i1 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1888 | |
| emilmont | 1:fdd22bb7aa52 | 1889 | /* writing the butterfly processed i0 + fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1890 | /* xc' = (xa-xb+xc-xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1891 | /* yc' = (ya-yb+yc-yd) */ |
| emilmont | 1:fdd22bb7aa52 | 1892 | pSrc16[i1 * 2u] = R0; |
| emilmont | 1:fdd22bb7aa52 | 1893 | pSrc16[(i1 * 2u) + 1u] = R1; |
| emilmont | 1:fdd22bb7aa52 | 1894 | |
| emilmont | 1:fdd22bb7aa52 | 1895 | /* Read yd (real), xd(imag) input */ |
| emilmont | 1:fdd22bb7aa52 | 1896 | U0 = pSrc16[i3 * 2u]; |
| emilmont | 1:fdd22bb7aa52 | 1897 | U1 = pSrc16[(i3 * 2u) + 1u]; |
| emilmont | 1:fdd22bb7aa52 | 1898 | /* T0 = (yb - yd), T1 = (xb - xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1899 | T0 = __SSAT(T0 - U0, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1900 | T1 = __SSAT(T1 - U1, 16u); |
| emilmont | 1:fdd22bb7aa52 | 1901 | |
| emilmont | 1:fdd22bb7aa52 | 1902 | /* writing the butterfly processed i0 + fftLen/2 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1903 | /* xb' = (xa-yb-xc+yd) */ |
| emilmont | 1:fdd22bb7aa52 | 1904 | /* yb' = (ya+xb-yc-xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1905 | pSrc16[i2 * 2u] = (S0 >> 1u) - (T1 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1906 | pSrc16[(i2 * 2u) + 1u] = (S1 >> 1u) + (T0 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1907 | |
| emilmont | 1:fdd22bb7aa52 | 1908 | |
| emilmont | 1:fdd22bb7aa52 | 1909 | /* writing the butterfly processed i0 + 3fftLen/4 sample */ |
| emilmont | 1:fdd22bb7aa52 | 1910 | /* xd' = (xa+yb-xc-yd) */ |
| emilmont | 1:fdd22bb7aa52 | 1911 | /* yd' = (ya-xb-yc+xd) */ |
| emilmont | 1:fdd22bb7aa52 | 1912 | pSrc16[i3 * 2u] = (S0 >> 1u) + (T1 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1913 | pSrc16[(i3 * 2u) + 1u] = (S1 >> 1u) - (T0 >> 1u); |
| emilmont | 1:fdd22bb7aa52 | 1914 | } |
| emilmont | 1:fdd22bb7aa52 | 1915 | /* end of last stage process */ |
| emilmont | 1:fdd22bb7aa52 | 1916 | |
| emilmont | 1:fdd22bb7aa52 | 1917 | /* output is in 11.5(q5) format for the 1024 point */ |
| emilmont | 1:fdd22bb7aa52 | 1918 | /* output is in 9.7(q7) format for the 256 point */ |
| emilmont | 1:fdd22bb7aa52 | 1919 | /* output is in 7.9(q9) format for the 64 point */ |
| emilmont | 1:fdd22bb7aa52 | 1920 | /* output is in 5.11(q11) format for the 16 point */ |
| emilmont | 1:fdd22bb7aa52 | 1921 | |
| mbed_official | 3:7a284390b0ce | 1922 | #endif /* #ifndef ARM_MATH_CM0_FAMILY */ |
| emilmont | 1:fdd22bb7aa52 | 1923 | |
| emilmont | 1:fdd22bb7aa52 | 1924 | } |