V4.0.1 of the ARM CMSIS DSP libraries. Note that arm_bitreversal2.s, arm_cfft_f32.c and arm_rfft_fast_f32.c had to be removed. arm_bitreversal2.s will not assemble with the online tools. So, the fast f32 FFT functions are not yet available. All the other FFT functions are available.
Dependents: MPU9150_Example fir_f32 fir_f32 MPU9150_nucleo_noni2cdev ... more
TransformFunctions/arm_dct4_f32.c@0:3d9c67d97d6f, 2014-07-28 (annotated)
- Committer:
- emh203
- Date:
- Mon Jul 28 15:03:15 2014 +0000
- Revision:
- 0:3d9c67d97d6f
1st working commit. Had to remove arm_bitreversal2.s arm_cfft_f32.c and arm_rfft_fast_f32.c. The .s will not assemble. For now I removed these functions so we could at least have a library for the other functions.
Who changed what in which revision?
User | Revision | Line number | New contents of line |
---|---|---|---|
emh203 | 0:3d9c67d97d6f | 1 | /* ---------------------------------------------------------------------- |
emh203 | 0:3d9c67d97d6f | 2 | * Copyright (C) 2010-2014 ARM Limited. All rights reserved. |
emh203 | 0:3d9c67d97d6f | 3 | * |
emh203 | 0:3d9c67d97d6f | 4 | * $Date: 12. March 2014 |
emh203 | 0:3d9c67d97d6f | 5 | * $Revision: V1.4.3 |
emh203 | 0:3d9c67d97d6f | 6 | * |
emh203 | 0:3d9c67d97d6f | 7 | * Project: CMSIS DSP Library |
emh203 | 0:3d9c67d97d6f | 8 | * Title: arm_dct4_f32.c |
emh203 | 0:3d9c67d97d6f | 9 | * |
emh203 | 0:3d9c67d97d6f | 10 | * Description: Processing function of DCT4 & IDCT4 F32. |
emh203 | 0:3d9c67d97d6f | 11 | * |
emh203 | 0:3d9c67d97d6f | 12 | * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0 |
emh203 | 0:3d9c67d97d6f | 13 | * |
emh203 | 0:3d9c67d97d6f | 14 | * Redistribution and use in source and binary forms, with or without |
emh203 | 0:3d9c67d97d6f | 15 | * modification, are permitted provided that the following conditions |
emh203 | 0:3d9c67d97d6f | 16 | * are met: |
emh203 | 0:3d9c67d97d6f | 17 | * - Redistributions of source code must retain the above copyright |
emh203 | 0:3d9c67d97d6f | 18 | * notice, this list of conditions and the following disclaimer. |
emh203 | 0:3d9c67d97d6f | 19 | * - Redistributions in binary form must reproduce the above copyright |
emh203 | 0:3d9c67d97d6f | 20 | * notice, this list of conditions and the following disclaimer in |
emh203 | 0:3d9c67d97d6f | 21 | * the documentation and/or other materials provided with the |
emh203 | 0:3d9c67d97d6f | 22 | * distribution. |
emh203 | 0:3d9c67d97d6f | 23 | * - Neither the name of ARM LIMITED nor the names of its contributors |
emh203 | 0:3d9c67d97d6f | 24 | * may be used to endorse or promote products derived from this |
emh203 | 0:3d9c67d97d6f | 25 | * software without specific prior written permission. |
emh203 | 0:3d9c67d97d6f | 26 | * |
emh203 | 0:3d9c67d97d6f | 27 | * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
emh203 | 0:3d9c67d97d6f | 28 | * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
emh203 | 0:3d9c67d97d6f | 29 | * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS |
emh203 | 0:3d9c67d97d6f | 30 | * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE |
emh203 | 0:3d9c67d97d6f | 31 | * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, |
emh203 | 0:3d9c67d97d6f | 32 | * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, |
emh203 | 0:3d9c67d97d6f | 33 | * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; |
emh203 | 0:3d9c67d97d6f | 34 | * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER |
emh203 | 0:3d9c67d97d6f | 35 | * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
emh203 | 0:3d9c67d97d6f | 36 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN |
emh203 | 0:3d9c67d97d6f | 37 | * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE |
emh203 | 0:3d9c67d97d6f | 38 | * POSSIBILITY OF SUCH DAMAGE. |
emh203 | 0:3d9c67d97d6f | 39 | * -------------------------------------------------------------------- */ |
emh203 | 0:3d9c67d97d6f | 40 | |
emh203 | 0:3d9c67d97d6f | 41 | #include "arm_math.h" |
emh203 | 0:3d9c67d97d6f | 42 | |
emh203 | 0:3d9c67d97d6f | 43 | /** |
emh203 | 0:3d9c67d97d6f | 44 | * @ingroup groupTransforms |
emh203 | 0:3d9c67d97d6f | 45 | */ |
emh203 | 0:3d9c67d97d6f | 46 | |
emh203 | 0:3d9c67d97d6f | 47 | /** |
emh203 | 0:3d9c67d97d6f | 48 | * @defgroup DCT4_IDCT4 DCT Type IV Functions |
emh203 | 0:3d9c67d97d6f | 49 | * Representation of signals by minimum number of values is important for storage and transmission. |
emh203 | 0:3d9c67d97d6f | 50 | * The possibility of large discontinuity between the beginning and end of a period of a signal |
emh203 | 0:3d9c67d97d6f | 51 | * in DFT can be avoided by extending the signal so that it is even-symmetric. |
emh203 | 0:3d9c67d97d6f | 52 | * Discrete Cosine Transform (DCT) is constructed such that its energy is heavily concentrated in the lower part of the |
emh203 | 0:3d9c67d97d6f | 53 | * spectrum and is very widely used in signal and image coding applications. |
emh203 | 0:3d9c67d97d6f | 54 | * The family of DCTs (DCT type- 1,2,3,4) is the outcome of different combinations of homogeneous boundary conditions. |
emh203 | 0:3d9c67d97d6f | 55 | * DCT has an excellent energy-packing capability, hence has many applications and in data compression in particular. |
emh203 | 0:3d9c67d97d6f | 56 | * |
emh203 | 0:3d9c67d97d6f | 57 | * DCT is essentially the Discrete Fourier Transform(DFT) of an even-extended real signal. |
emh203 | 0:3d9c67d97d6f | 58 | * Reordering of the input data makes the computation of DCT just a problem of |
emh203 | 0:3d9c67d97d6f | 59 | * computing the DFT of a real signal with a few additional operations. |
emh203 | 0:3d9c67d97d6f | 60 | * This approach provides regular, simple, and very efficient DCT algorithms for practical hardware and software implementations. |
emh203 | 0:3d9c67d97d6f | 61 | * |
emh203 | 0:3d9c67d97d6f | 62 | * DCT type-II can be implemented using Fast fourier transform (FFT) internally, as the transform is applied on real values, Real FFT can be used. |
emh203 | 0:3d9c67d97d6f | 63 | * DCT4 is implemented using DCT2 as their implementations are similar except with some added pre-processing and post-processing. |
emh203 | 0:3d9c67d97d6f | 64 | * DCT2 implementation can be described in the following steps: |
emh203 | 0:3d9c67d97d6f | 65 | * - Re-ordering input |
emh203 | 0:3d9c67d97d6f | 66 | * - Calculating Real FFT |
emh203 | 0:3d9c67d97d6f | 67 | * - Multiplication of weights and Real FFT output and getting real part from the product. |
emh203 | 0:3d9c67d97d6f | 68 | * |
emh203 | 0:3d9c67d97d6f | 69 | * This process is explained by the block diagram below: |
emh203 | 0:3d9c67d97d6f | 70 | * \image html DCT4.gif "Discrete Cosine Transform - type-IV" |
emh203 | 0:3d9c67d97d6f | 71 | * |
emh203 | 0:3d9c67d97d6f | 72 | * \par Algorithm: |
emh203 | 0:3d9c67d97d6f | 73 | * The N-point type-IV DCT is defined as a real, linear transformation by the formula: |
emh203 | 0:3d9c67d97d6f | 74 | * \image html DCT4Equation.gif |
emh203 | 0:3d9c67d97d6f | 75 | * where <code>k = 0,1,2,.....N-1</code> |
emh203 | 0:3d9c67d97d6f | 76 | *\par |
emh203 | 0:3d9c67d97d6f | 77 | * Its inverse is defined as follows: |
emh203 | 0:3d9c67d97d6f | 78 | * \image html IDCT4Equation.gif |
emh203 | 0:3d9c67d97d6f | 79 | * where <code>n = 0,1,2,.....N-1</code> |
emh203 | 0:3d9c67d97d6f | 80 | *\par |
emh203 | 0:3d9c67d97d6f | 81 | * The DCT4 matrices become involutory (i.e. they are self-inverse) by multiplying with an overall scale factor of sqrt(2/N). |
emh203 | 0:3d9c67d97d6f | 82 | * The symmetry of the transform matrix indicates that the fast algorithms for the forward |
emh203 | 0:3d9c67d97d6f | 83 | * and inverse transform computation are identical. |
emh203 | 0:3d9c67d97d6f | 84 | * Note that the implementation of Inverse DCT4 and DCT4 is same, hence same process function can be used for both. |
emh203 | 0:3d9c67d97d6f | 85 | * |
emh203 | 0:3d9c67d97d6f | 86 | * \par Lengths supported by the transform: |
emh203 | 0:3d9c67d97d6f | 87 | * As DCT4 internally uses Real FFT, it supports all the lengths supported by arm_rfft_f32(). |
emh203 | 0:3d9c67d97d6f | 88 | * The library provides separate functions for Q15, Q31, and floating-point data types. |
emh203 | 0:3d9c67d97d6f | 89 | * \par Instance Structure |
emh203 | 0:3d9c67d97d6f | 90 | * The instances for Real FFT and FFT, cosine values table and twiddle factor table are stored in an instance data structure. |
emh203 | 0:3d9c67d97d6f | 91 | * A separate instance structure must be defined for each transform. |
emh203 | 0:3d9c67d97d6f | 92 | * There are separate instance structure declarations for each of the 3 supported data types. |
emh203 | 0:3d9c67d97d6f | 93 | * |
emh203 | 0:3d9c67d97d6f | 94 | * \par Initialization Functions |
emh203 | 0:3d9c67d97d6f | 95 | * There is also an associated initialization function for each data type. |
emh203 | 0:3d9c67d97d6f | 96 | * The initialization function performs the following operations: |
emh203 | 0:3d9c67d97d6f | 97 | * - Sets the values of the internal structure fields. |
emh203 | 0:3d9c67d97d6f | 98 | * - Initializes Real FFT as its process function is used internally in DCT4, by calling arm_rfft_init_f32(). |
emh203 | 0:3d9c67d97d6f | 99 | * \par |
emh203 | 0:3d9c67d97d6f | 100 | * Use of the initialization function is optional. |
emh203 | 0:3d9c67d97d6f | 101 | * However, if the initialization function is used, then the instance structure cannot be placed into a const data section. |
emh203 | 0:3d9c67d97d6f | 102 | * To place an instance structure into a const data section, the instance structure must be manually initialized. |
emh203 | 0:3d9c67d97d6f | 103 | * Manually initialize the instance structure as follows: |
emh203 | 0:3d9c67d97d6f | 104 | * <pre> |
emh203 | 0:3d9c67d97d6f | 105 | *arm_dct4_instance_f32 S = {N, Nby2, normalize, pTwiddle, pCosFactor, pRfft, pCfft}; |
emh203 | 0:3d9c67d97d6f | 106 | *arm_dct4_instance_q31 S = {N, Nby2, normalize, pTwiddle, pCosFactor, pRfft, pCfft}; |
emh203 | 0:3d9c67d97d6f | 107 | *arm_dct4_instance_q15 S = {N, Nby2, normalize, pTwiddle, pCosFactor, pRfft, pCfft}; |
emh203 | 0:3d9c67d97d6f | 108 | * </pre> |
emh203 | 0:3d9c67d97d6f | 109 | * where \c N is the length of the DCT4; \c Nby2 is half of the length of the DCT4; |
emh203 | 0:3d9c67d97d6f | 110 | * \c normalize is normalizing factor used and is equal to <code>sqrt(2/N)</code>; |
emh203 | 0:3d9c67d97d6f | 111 | * \c pTwiddle points to the twiddle factor table; |
emh203 | 0:3d9c67d97d6f | 112 | * \c pCosFactor points to the cosFactor table; |
emh203 | 0:3d9c67d97d6f | 113 | * \c pRfft points to the real FFT instance; |
emh203 | 0:3d9c67d97d6f | 114 | * \c pCfft points to the complex FFT instance; |
emh203 | 0:3d9c67d97d6f | 115 | * The CFFT and RFFT structures also needs to be initialized, refer to arm_cfft_radix4_f32() |
emh203 | 0:3d9c67d97d6f | 116 | * and arm_rfft_f32() respectively for details regarding static initialization. |
emh203 | 0:3d9c67d97d6f | 117 | * |
emh203 | 0:3d9c67d97d6f | 118 | * \par Fixed-Point Behavior |
emh203 | 0:3d9c67d97d6f | 119 | * Care must be taken when using the fixed-point versions of the DCT4 transform functions. |
emh203 | 0:3d9c67d97d6f | 120 | * In particular, the overflow and saturation behavior of the accumulator used in each function must be considered. |
emh203 | 0:3d9c67d97d6f | 121 | * Refer to the function specific documentation below for usage guidelines. |
emh203 | 0:3d9c67d97d6f | 122 | */ |
emh203 | 0:3d9c67d97d6f | 123 | |
emh203 | 0:3d9c67d97d6f | 124 | /** |
emh203 | 0:3d9c67d97d6f | 125 | * @addtogroup DCT4_IDCT4 |
emh203 | 0:3d9c67d97d6f | 126 | * @{ |
emh203 | 0:3d9c67d97d6f | 127 | */ |
emh203 | 0:3d9c67d97d6f | 128 | |
emh203 | 0:3d9c67d97d6f | 129 | /** |
emh203 | 0:3d9c67d97d6f | 130 | * @brief Processing function for the floating-point DCT4/IDCT4. |
emh203 | 0:3d9c67d97d6f | 131 | * @param[in] *S points to an instance of the floating-point DCT4/IDCT4 structure. |
emh203 | 0:3d9c67d97d6f | 132 | * @param[in] *pState points to state buffer. |
emh203 | 0:3d9c67d97d6f | 133 | * @param[in,out] *pInlineBuffer points to the in-place input and output buffer. |
emh203 | 0:3d9c67d97d6f | 134 | * @return none. |
emh203 | 0:3d9c67d97d6f | 135 | */ |
emh203 | 0:3d9c67d97d6f | 136 | |
emh203 | 0:3d9c67d97d6f | 137 | void arm_dct4_f32( |
emh203 | 0:3d9c67d97d6f | 138 | const arm_dct4_instance_f32 * S, |
emh203 | 0:3d9c67d97d6f | 139 | float32_t * pState, |
emh203 | 0:3d9c67d97d6f | 140 | float32_t * pInlineBuffer) |
emh203 | 0:3d9c67d97d6f | 141 | { |
emh203 | 0:3d9c67d97d6f | 142 | uint32_t i; /* Loop counter */ |
emh203 | 0:3d9c67d97d6f | 143 | float32_t *weights = S->pTwiddle; /* Pointer to the Weights table */ |
emh203 | 0:3d9c67d97d6f | 144 | float32_t *cosFact = S->pCosFactor; /* Pointer to the cos factors table */ |
emh203 | 0:3d9c67d97d6f | 145 | float32_t *pS1, *pS2, *pbuff; /* Temporary pointers for input buffer and pState buffer */ |
emh203 | 0:3d9c67d97d6f | 146 | float32_t in; /* Temporary variable */ |
emh203 | 0:3d9c67d97d6f | 147 | |
emh203 | 0:3d9c67d97d6f | 148 | |
emh203 | 0:3d9c67d97d6f | 149 | /* DCT4 computation involves DCT2 (which is calculated using RFFT) |
emh203 | 0:3d9c67d97d6f | 150 | * along with some pre-processing and post-processing. |
emh203 | 0:3d9c67d97d6f | 151 | * Computational procedure is explained as follows: |
emh203 | 0:3d9c67d97d6f | 152 | * (a) Pre-processing involves multiplying input with cos factor, |
emh203 | 0:3d9c67d97d6f | 153 | * r(n) = 2 * u(n) * cos(pi*(2*n+1)/(4*n)) |
emh203 | 0:3d9c67d97d6f | 154 | * where, |
emh203 | 0:3d9c67d97d6f | 155 | * r(n) -- output of preprocessing |
emh203 | 0:3d9c67d97d6f | 156 | * u(n) -- input to preprocessing(actual Source buffer) |
emh203 | 0:3d9c67d97d6f | 157 | * (b) Calculation of DCT2 using FFT is divided into three steps: |
emh203 | 0:3d9c67d97d6f | 158 | * Step1: Re-ordering of even and odd elements of input. |
emh203 | 0:3d9c67d97d6f | 159 | * Step2: Calculating FFT of the re-ordered input. |
emh203 | 0:3d9c67d97d6f | 160 | * Step3: Taking the real part of the product of FFT output and weights. |
emh203 | 0:3d9c67d97d6f | 161 | * (c) Post-processing - DCT4 can be obtained from DCT2 output using the following equation: |
emh203 | 0:3d9c67d97d6f | 162 | * Y4(k) = Y2(k) - Y4(k-1) and Y4(-1) = Y4(0) |
emh203 | 0:3d9c67d97d6f | 163 | * where, |
emh203 | 0:3d9c67d97d6f | 164 | * Y4 -- DCT4 output, Y2 -- DCT2 output |
emh203 | 0:3d9c67d97d6f | 165 | * (d) Multiplying the output with the normalizing factor sqrt(2/N). |
emh203 | 0:3d9c67d97d6f | 166 | */ |
emh203 | 0:3d9c67d97d6f | 167 | |
emh203 | 0:3d9c67d97d6f | 168 | /*-------- Pre-processing ------------*/ |
emh203 | 0:3d9c67d97d6f | 169 | /* Multiplying input with cos factor i.e. r(n) = 2 * x(n) * cos(pi*(2*n+1)/(4*n)) */ |
emh203 | 0:3d9c67d97d6f | 170 | arm_scale_f32(pInlineBuffer, 2.0f, pInlineBuffer, S->N); |
emh203 | 0:3d9c67d97d6f | 171 | arm_mult_f32(pInlineBuffer, cosFact, pInlineBuffer, S->N); |
emh203 | 0:3d9c67d97d6f | 172 | |
emh203 | 0:3d9c67d97d6f | 173 | /* ---------------------------------------------------------------- |
emh203 | 0:3d9c67d97d6f | 174 | * Step1: Re-ordering of even and odd elements as, |
emh203 | 0:3d9c67d97d6f | 175 | * pState[i] = pInlineBuffer[2*i] and |
emh203 | 0:3d9c67d97d6f | 176 | * pState[N-i-1] = pInlineBuffer[2*i+1] where i = 0 to N/2 |
emh203 | 0:3d9c67d97d6f | 177 | ---------------------------------------------------------------------*/ |
emh203 | 0:3d9c67d97d6f | 178 | |
emh203 | 0:3d9c67d97d6f | 179 | /* pS1 initialized to pState */ |
emh203 | 0:3d9c67d97d6f | 180 | pS1 = pState; |
emh203 | 0:3d9c67d97d6f | 181 | |
emh203 | 0:3d9c67d97d6f | 182 | /* pS2 initialized to pState+N-1, so that it points to the end of the state buffer */ |
emh203 | 0:3d9c67d97d6f | 183 | pS2 = pState + (S->N - 1u); |
emh203 | 0:3d9c67d97d6f | 184 | |
emh203 | 0:3d9c67d97d6f | 185 | /* pbuff initialized to input buffer */ |
emh203 | 0:3d9c67d97d6f | 186 | pbuff = pInlineBuffer; |
emh203 | 0:3d9c67d97d6f | 187 | |
emh203 | 0:3d9c67d97d6f | 188 | #ifndef ARM_MATH_CM0_FAMILY |
emh203 | 0:3d9c67d97d6f | 189 | |
emh203 | 0:3d9c67d97d6f | 190 | /* Run the below code for Cortex-M4 and Cortex-M3 */ |
emh203 | 0:3d9c67d97d6f | 191 | |
emh203 | 0:3d9c67d97d6f | 192 | /* Initializing the loop counter to N/2 >> 2 for loop unrolling by 4 */ |
emh203 | 0:3d9c67d97d6f | 193 | i = (uint32_t) S->Nby2 >> 2u; |
emh203 | 0:3d9c67d97d6f | 194 | |
emh203 | 0:3d9c67d97d6f | 195 | /* First part of the processing with loop unrolling. Compute 4 outputs at a time. |
emh203 | 0:3d9c67d97d6f | 196 | ** a second loop below computes the remaining 1 to 3 samples. */ |
emh203 | 0:3d9c67d97d6f | 197 | do |
emh203 | 0:3d9c67d97d6f | 198 | { |
emh203 | 0:3d9c67d97d6f | 199 | /* Re-ordering of even and odd elements */ |
emh203 | 0:3d9c67d97d6f | 200 | /* pState[i] = pInlineBuffer[2*i] */ |
emh203 | 0:3d9c67d97d6f | 201 | *pS1++ = *pbuff++; |
emh203 | 0:3d9c67d97d6f | 202 | /* pState[N-i-1] = pInlineBuffer[2*i+1] */ |
emh203 | 0:3d9c67d97d6f | 203 | *pS2-- = *pbuff++; |
emh203 | 0:3d9c67d97d6f | 204 | |
emh203 | 0:3d9c67d97d6f | 205 | *pS1++ = *pbuff++; |
emh203 | 0:3d9c67d97d6f | 206 | *pS2-- = *pbuff++; |
emh203 | 0:3d9c67d97d6f | 207 | |
emh203 | 0:3d9c67d97d6f | 208 | *pS1++ = *pbuff++; |
emh203 | 0:3d9c67d97d6f | 209 | *pS2-- = *pbuff++; |
emh203 | 0:3d9c67d97d6f | 210 | |
emh203 | 0:3d9c67d97d6f | 211 | *pS1++ = *pbuff++; |
emh203 | 0:3d9c67d97d6f | 212 | *pS2-- = *pbuff++; |
emh203 | 0:3d9c67d97d6f | 213 | |
emh203 | 0:3d9c67d97d6f | 214 | /* Decrement the loop counter */ |
emh203 | 0:3d9c67d97d6f | 215 | i--; |
emh203 | 0:3d9c67d97d6f | 216 | } while(i > 0u); |
emh203 | 0:3d9c67d97d6f | 217 | |
emh203 | 0:3d9c67d97d6f | 218 | /* pbuff initialized to input buffer */ |
emh203 | 0:3d9c67d97d6f | 219 | pbuff = pInlineBuffer; |
emh203 | 0:3d9c67d97d6f | 220 | |
emh203 | 0:3d9c67d97d6f | 221 | /* pS1 initialized to pState */ |
emh203 | 0:3d9c67d97d6f | 222 | pS1 = pState; |
emh203 | 0:3d9c67d97d6f | 223 | |
emh203 | 0:3d9c67d97d6f | 224 | /* Initializing the loop counter to N/4 instead of N for loop unrolling */ |
emh203 | 0:3d9c67d97d6f | 225 | i = (uint32_t) S->N >> 2u; |
emh203 | 0:3d9c67d97d6f | 226 | |
emh203 | 0:3d9c67d97d6f | 227 | /* Processing with loop unrolling 4 times as N is always multiple of 4. |
emh203 | 0:3d9c67d97d6f | 228 | * Compute 4 outputs at a time */ |
emh203 | 0:3d9c67d97d6f | 229 | do |
emh203 | 0:3d9c67d97d6f | 230 | { |
emh203 | 0:3d9c67d97d6f | 231 | /* Writing the re-ordered output back to inplace input buffer */ |
emh203 | 0:3d9c67d97d6f | 232 | *pbuff++ = *pS1++; |
emh203 | 0:3d9c67d97d6f | 233 | *pbuff++ = *pS1++; |
emh203 | 0:3d9c67d97d6f | 234 | *pbuff++ = *pS1++; |
emh203 | 0:3d9c67d97d6f | 235 | *pbuff++ = *pS1++; |
emh203 | 0:3d9c67d97d6f | 236 | |
emh203 | 0:3d9c67d97d6f | 237 | /* Decrement the loop counter */ |
emh203 | 0:3d9c67d97d6f | 238 | i--; |
emh203 | 0:3d9c67d97d6f | 239 | } while(i > 0u); |
emh203 | 0:3d9c67d97d6f | 240 | |
emh203 | 0:3d9c67d97d6f | 241 | |
emh203 | 0:3d9c67d97d6f | 242 | /* --------------------------------------------------------- |
emh203 | 0:3d9c67d97d6f | 243 | * Step2: Calculate RFFT for N-point input |
emh203 | 0:3d9c67d97d6f | 244 | * ---------------------------------------------------------- */ |
emh203 | 0:3d9c67d97d6f | 245 | /* pInlineBuffer is real input of length N , pState is the complex output of length 2N */ |
emh203 | 0:3d9c67d97d6f | 246 | arm_rfft_f32(S->pRfft, pInlineBuffer, pState); |
emh203 | 0:3d9c67d97d6f | 247 | |
emh203 | 0:3d9c67d97d6f | 248 | /*---------------------------------------------------------------------- |
emh203 | 0:3d9c67d97d6f | 249 | * Step3: Multiply the FFT output with the weights. |
emh203 | 0:3d9c67d97d6f | 250 | *----------------------------------------------------------------------*/ |
emh203 | 0:3d9c67d97d6f | 251 | arm_cmplx_mult_cmplx_f32(pState, weights, pState, S->N); |
emh203 | 0:3d9c67d97d6f | 252 | |
emh203 | 0:3d9c67d97d6f | 253 | /* ----------- Post-processing ---------- */ |
emh203 | 0:3d9c67d97d6f | 254 | /* DCT-IV can be obtained from DCT-II by the equation, |
emh203 | 0:3d9c67d97d6f | 255 | * Y4(k) = Y2(k) - Y4(k-1) and Y4(-1) = Y4(0) |
emh203 | 0:3d9c67d97d6f | 256 | * Hence, Y4(0) = Y2(0)/2 */ |
emh203 | 0:3d9c67d97d6f | 257 | /* Getting only real part from the output and Converting to DCT-IV */ |
emh203 | 0:3d9c67d97d6f | 258 | |
emh203 | 0:3d9c67d97d6f | 259 | /* Initializing the loop counter to N >> 2 for loop unrolling by 4 */ |
emh203 | 0:3d9c67d97d6f | 260 | i = ((uint32_t) S->N - 1u) >> 2u; |
emh203 | 0:3d9c67d97d6f | 261 | |
emh203 | 0:3d9c67d97d6f | 262 | /* pbuff initialized to input buffer. */ |
emh203 | 0:3d9c67d97d6f | 263 | pbuff = pInlineBuffer; |
emh203 | 0:3d9c67d97d6f | 264 | |
emh203 | 0:3d9c67d97d6f | 265 | /* pS1 initialized to pState */ |
emh203 | 0:3d9c67d97d6f | 266 | pS1 = pState; |
emh203 | 0:3d9c67d97d6f | 267 | |
emh203 | 0:3d9c67d97d6f | 268 | /* Calculating Y4(0) from Y2(0) using Y4(0) = Y2(0)/2 */ |
emh203 | 0:3d9c67d97d6f | 269 | in = *pS1++ * (float32_t) 0.5; |
emh203 | 0:3d9c67d97d6f | 270 | /* input buffer acts as inplace, so output values are stored in the input itself. */ |
emh203 | 0:3d9c67d97d6f | 271 | *pbuff++ = in; |
emh203 | 0:3d9c67d97d6f | 272 | |
emh203 | 0:3d9c67d97d6f | 273 | /* pState pointer is incremented twice as the real values are located alternatively in the array */ |
emh203 | 0:3d9c67d97d6f | 274 | pS1++; |
emh203 | 0:3d9c67d97d6f | 275 | |
emh203 | 0:3d9c67d97d6f | 276 | /* First part of the processing with loop unrolling. Compute 4 outputs at a time. |
emh203 | 0:3d9c67d97d6f | 277 | ** a second loop below computes the remaining 1 to 3 samples. */ |
emh203 | 0:3d9c67d97d6f | 278 | do |
emh203 | 0:3d9c67d97d6f | 279 | { |
emh203 | 0:3d9c67d97d6f | 280 | /* Calculating Y4(1) to Y4(N-1) from Y2 using equation Y4(k) = Y2(k) - Y4(k-1) */ |
emh203 | 0:3d9c67d97d6f | 281 | /* pState pointer (pS1) is incremented twice as the real values are located alternatively in the array */ |
emh203 | 0:3d9c67d97d6f | 282 | in = *pS1++ - in; |
emh203 | 0:3d9c67d97d6f | 283 | *pbuff++ = in; |
emh203 | 0:3d9c67d97d6f | 284 | /* points to the next real value */ |
emh203 | 0:3d9c67d97d6f | 285 | pS1++; |
emh203 | 0:3d9c67d97d6f | 286 | |
emh203 | 0:3d9c67d97d6f | 287 | in = *pS1++ - in; |
emh203 | 0:3d9c67d97d6f | 288 | *pbuff++ = in; |
emh203 | 0:3d9c67d97d6f | 289 | pS1++; |
emh203 | 0:3d9c67d97d6f | 290 | |
emh203 | 0:3d9c67d97d6f | 291 | in = *pS1++ - in; |
emh203 | 0:3d9c67d97d6f | 292 | *pbuff++ = in; |
emh203 | 0:3d9c67d97d6f | 293 | pS1++; |
emh203 | 0:3d9c67d97d6f | 294 | |
emh203 | 0:3d9c67d97d6f | 295 | in = *pS1++ - in; |
emh203 | 0:3d9c67d97d6f | 296 | *pbuff++ = in; |
emh203 | 0:3d9c67d97d6f | 297 | pS1++; |
emh203 | 0:3d9c67d97d6f | 298 | |
emh203 | 0:3d9c67d97d6f | 299 | /* Decrement the loop counter */ |
emh203 | 0:3d9c67d97d6f | 300 | i--; |
emh203 | 0:3d9c67d97d6f | 301 | } while(i > 0u); |
emh203 | 0:3d9c67d97d6f | 302 | |
emh203 | 0:3d9c67d97d6f | 303 | /* If the blockSize is not a multiple of 4, compute any remaining output samples here. |
emh203 | 0:3d9c67d97d6f | 304 | ** No loop unrolling is used. */ |
emh203 | 0:3d9c67d97d6f | 305 | i = ((uint32_t) S->N - 1u) % 0x4u; |
emh203 | 0:3d9c67d97d6f | 306 | |
emh203 | 0:3d9c67d97d6f | 307 | while(i > 0u) |
emh203 | 0:3d9c67d97d6f | 308 | { |
emh203 | 0:3d9c67d97d6f | 309 | /* Calculating Y4(1) to Y4(N-1) from Y2 using equation Y4(k) = Y2(k) - Y4(k-1) */ |
emh203 | 0:3d9c67d97d6f | 310 | /* pState pointer (pS1) is incremented twice as the real values are located alternatively in the array */ |
emh203 | 0:3d9c67d97d6f | 311 | in = *pS1++ - in; |
emh203 | 0:3d9c67d97d6f | 312 | *pbuff++ = in; |
emh203 | 0:3d9c67d97d6f | 313 | /* points to the next real value */ |
emh203 | 0:3d9c67d97d6f | 314 | pS1++; |
emh203 | 0:3d9c67d97d6f | 315 | |
emh203 | 0:3d9c67d97d6f | 316 | /* Decrement the loop counter */ |
emh203 | 0:3d9c67d97d6f | 317 | i--; |
emh203 | 0:3d9c67d97d6f | 318 | } |
emh203 | 0:3d9c67d97d6f | 319 | |
emh203 | 0:3d9c67d97d6f | 320 | |
emh203 | 0:3d9c67d97d6f | 321 | /*------------ Normalizing the output by multiplying with the normalizing factor ----------*/ |
emh203 | 0:3d9c67d97d6f | 322 | |
emh203 | 0:3d9c67d97d6f | 323 | /* Initializing the loop counter to N/4 instead of N for loop unrolling */ |
emh203 | 0:3d9c67d97d6f | 324 | i = (uint32_t) S->N >> 2u; |
emh203 | 0:3d9c67d97d6f | 325 | |
emh203 | 0:3d9c67d97d6f | 326 | /* pbuff initialized to the pInlineBuffer(now contains the output values) */ |
emh203 | 0:3d9c67d97d6f | 327 | pbuff = pInlineBuffer; |
emh203 | 0:3d9c67d97d6f | 328 | |
emh203 | 0:3d9c67d97d6f | 329 | /* Processing with loop unrolling 4 times as N is always multiple of 4. Compute 4 outputs at a time */ |
emh203 | 0:3d9c67d97d6f | 330 | do |
emh203 | 0:3d9c67d97d6f | 331 | { |
emh203 | 0:3d9c67d97d6f | 332 | /* Multiplying pInlineBuffer with the normalizing factor sqrt(2/N) */ |
emh203 | 0:3d9c67d97d6f | 333 | in = *pbuff; |
emh203 | 0:3d9c67d97d6f | 334 | *pbuff++ = in * S->normalize; |
emh203 | 0:3d9c67d97d6f | 335 | |
emh203 | 0:3d9c67d97d6f | 336 | in = *pbuff; |
emh203 | 0:3d9c67d97d6f | 337 | *pbuff++ = in * S->normalize; |
emh203 | 0:3d9c67d97d6f | 338 | |
emh203 | 0:3d9c67d97d6f | 339 | in = *pbuff; |
emh203 | 0:3d9c67d97d6f | 340 | *pbuff++ = in * S->normalize; |
emh203 | 0:3d9c67d97d6f | 341 | |
emh203 | 0:3d9c67d97d6f | 342 | in = *pbuff; |
emh203 | 0:3d9c67d97d6f | 343 | *pbuff++ = in * S->normalize; |
emh203 | 0:3d9c67d97d6f | 344 | |
emh203 | 0:3d9c67d97d6f | 345 | /* Decrement the loop counter */ |
emh203 | 0:3d9c67d97d6f | 346 | i--; |
emh203 | 0:3d9c67d97d6f | 347 | } while(i > 0u); |
emh203 | 0:3d9c67d97d6f | 348 | |
emh203 | 0:3d9c67d97d6f | 349 | |
emh203 | 0:3d9c67d97d6f | 350 | #else |
emh203 | 0:3d9c67d97d6f | 351 | |
emh203 | 0:3d9c67d97d6f | 352 | /* Run the below code for Cortex-M0 */ |
emh203 | 0:3d9c67d97d6f | 353 | |
emh203 | 0:3d9c67d97d6f | 354 | /* Initializing the loop counter to N/2 */ |
emh203 | 0:3d9c67d97d6f | 355 | i = (uint32_t) S->Nby2; |
emh203 | 0:3d9c67d97d6f | 356 | |
emh203 | 0:3d9c67d97d6f | 357 | do |
emh203 | 0:3d9c67d97d6f | 358 | { |
emh203 | 0:3d9c67d97d6f | 359 | /* Re-ordering of even and odd elements */ |
emh203 | 0:3d9c67d97d6f | 360 | /* pState[i] = pInlineBuffer[2*i] */ |
emh203 | 0:3d9c67d97d6f | 361 | *pS1++ = *pbuff++; |
emh203 | 0:3d9c67d97d6f | 362 | /* pState[N-i-1] = pInlineBuffer[2*i+1] */ |
emh203 | 0:3d9c67d97d6f | 363 | *pS2-- = *pbuff++; |
emh203 | 0:3d9c67d97d6f | 364 | |
emh203 | 0:3d9c67d97d6f | 365 | /* Decrement the loop counter */ |
emh203 | 0:3d9c67d97d6f | 366 | i--; |
emh203 | 0:3d9c67d97d6f | 367 | } while(i > 0u); |
emh203 | 0:3d9c67d97d6f | 368 | |
emh203 | 0:3d9c67d97d6f | 369 | /* pbuff initialized to input buffer */ |
emh203 | 0:3d9c67d97d6f | 370 | pbuff = pInlineBuffer; |
emh203 | 0:3d9c67d97d6f | 371 | |
emh203 | 0:3d9c67d97d6f | 372 | /* pS1 initialized to pState */ |
emh203 | 0:3d9c67d97d6f | 373 | pS1 = pState; |
emh203 | 0:3d9c67d97d6f | 374 | |
emh203 | 0:3d9c67d97d6f | 375 | /* Initializing the loop counter */ |
emh203 | 0:3d9c67d97d6f | 376 | i = (uint32_t) S->N; |
emh203 | 0:3d9c67d97d6f | 377 | |
emh203 | 0:3d9c67d97d6f | 378 | do |
emh203 | 0:3d9c67d97d6f | 379 | { |
emh203 | 0:3d9c67d97d6f | 380 | /* Writing the re-ordered output back to inplace input buffer */ |
emh203 | 0:3d9c67d97d6f | 381 | *pbuff++ = *pS1++; |
emh203 | 0:3d9c67d97d6f | 382 | |
emh203 | 0:3d9c67d97d6f | 383 | /* Decrement the loop counter */ |
emh203 | 0:3d9c67d97d6f | 384 | i--; |
emh203 | 0:3d9c67d97d6f | 385 | } while(i > 0u); |
emh203 | 0:3d9c67d97d6f | 386 | |
emh203 | 0:3d9c67d97d6f | 387 | |
emh203 | 0:3d9c67d97d6f | 388 | /* --------------------------------------------------------- |
emh203 | 0:3d9c67d97d6f | 389 | * Step2: Calculate RFFT for N-point input |
emh203 | 0:3d9c67d97d6f | 390 | * ---------------------------------------------------------- */ |
emh203 | 0:3d9c67d97d6f | 391 | /* pInlineBuffer is real input of length N , pState is the complex output of length 2N */ |
emh203 | 0:3d9c67d97d6f | 392 | arm_rfft_f32(S->pRfft, pInlineBuffer, pState); |
emh203 | 0:3d9c67d97d6f | 393 | |
emh203 | 0:3d9c67d97d6f | 394 | /*---------------------------------------------------------------------- |
emh203 | 0:3d9c67d97d6f | 395 | * Step3: Multiply the FFT output with the weights. |
emh203 | 0:3d9c67d97d6f | 396 | *----------------------------------------------------------------------*/ |
emh203 | 0:3d9c67d97d6f | 397 | arm_cmplx_mult_cmplx_f32(pState, weights, pState, S->N); |
emh203 | 0:3d9c67d97d6f | 398 | |
emh203 | 0:3d9c67d97d6f | 399 | /* ----------- Post-processing ---------- */ |
emh203 | 0:3d9c67d97d6f | 400 | /* DCT-IV can be obtained from DCT-II by the equation, |
emh203 | 0:3d9c67d97d6f | 401 | * Y4(k) = Y2(k) - Y4(k-1) and Y4(-1) = Y4(0) |
emh203 | 0:3d9c67d97d6f | 402 | * Hence, Y4(0) = Y2(0)/2 */ |
emh203 | 0:3d9c67d97d6f | 403 | /* Getting only real part from the output and Converting to DCT-IV */ |
emh203 | 0:3d9c67d97d6f | 404 | |
emh203 | 0:3d9c67d97d6f | 405 | /* pbuff initialized to input buffer. */ |
emh203 | 0:3d9c67d97d6f | 406 | pbuff = pInlineBuffer; |
emh203 | 0:3d9c67d97d6f | 407 | |
emh203 | 0:3d9c67d97d6f | 408 | /* pS1 initialized to pState */ |
emh203 | 0:3d9c67d97d6f | 409 | pS1 = pState; |
emh203 | 0:3d9c67d97d6f | 410 | |
emh203 | 0:3d9c67d97d6f | 411 | /* Calculating Y4(0) from Y2(0) using Y4(0) = Y2(0)/2 */ |
emh203 | 0:3d9c67d97d6f | 412 | in = *pS1++ * (float32_t) 0.5; |
emh203 | 0:3d9c67d97d6f | 413 | /* input buffer acts as inplace, so output values are stored in the input itself. */ |
emh203 | 0:3d9c67d97d6f | 414 | *pbuff++ = in; |
emh203 | 0:3d9c67d97d6f | 415 | |
emh203 | 0:3d9c67d97d6f | 416 | /* pState pointer is incremented twice as the real values are located alternatively in the array */ |
emh203 | 0:3d9c67d97d6f | 417 | pS1++; |
emh203 | 0:3d9c67d97d6f | 418 | |
emh203 | 0:3d9c67d97d6f | 419 | /* Initializing the loop counter */ |
emh203 | 0:3d9c67d97d6f | 420 | i = ((uint32_t) S->N - 1u); |
emh203 | 0:3d9c67d97d6f | 421 | |
emh203 | 0:3d9c67d97d6f | 422 | do |
emh203 | 0:3d9c67d97d6f | 423 | { |
emh203 | 0:3d9c67d97d6f | 424 | /* Calculating Y4(1) to Y4(N-1) from Y2 using equation Y4(k) = Y2(k) - Y4(k-1) */ |
emh203 | 0:3d9c67d97d6f | 425 | /* pState pointer (pS1) is incremented twice as the real values are located alternatively in the array */ |
emh203 | 0:3d9c67d97d6f | 426 | in = *pS1++ - in; |
emh203 | 0:3d9c67d97d6f | 427 | *pbuff++ = in; |
emh203 | 0:3d9c67d97d6f | 428 | /* points to the next real value */ |
emh203 | 0:3d9c67d97d6f | 429 | pS1++; |
emh203 | 0:3d9c67d97d6f | 430 | |
emh203 | 0:3d9c67d97d6f | 431 | |
emh203 | 0:3d9c67d97d6f | 432 | /* Decrement the loop counter */ |
emh203 | 0:3d9c67d97d6f | 433 | i--; |
emh203 | 0:3d9c67d97d6f | 434 | } while(i > 0u); |
emh203 | 0:3d9c67d97d6f | 435 | |
emh203 | 0:3d9c67d97d6f | 436 | |
emh203 | 0:3d9c67d97d6f | 437 | /*------------ Normalizing the output by multiplying with the normalizing factor ----------*/ |
emh203 | 0:3d9c67d97d6f | 438 | |
emh203 | 0:3d9c67d97d6f | 439 | /* Initializing the loop counter */ |
emh203 | 0:3d9c67d97d6f | 440 | i = (uint32_t) S->N; |
emh203 | 0:3d9c67d97d6f | 441 | |
emh203 | 0:3d9c67d97d6f | 442 | /* pbuff initialized to the pInlineBuffer(now contains the output values) */ |
emh203 | 0:3d9c67d97d6f | 443 | pbuff = pInlineBuffer; |
emh203 | 0:3d9c67d97d6f | 444 | |
emh203 | 0:3d9c67d97d6f | 445 | do |
emh203 | 0:3d9c67d97d6f | 446 | { |
emh203 | 0:3d9c67d97d6f | 447 | /* Multiplying pInlineBuffer with the normalizing factor sqrt(2/N) */ |
emh203 | 0:3d9c67d97d6f | 448 | in = *pbuff; |
emh203 | 0:3d9c67d97d6f | 449 | *pbuff++ = in * S->normalize; |
emh203 | 0:3d9c67d97d6f | 450 | |
emh203 | 0:3d9c67d97d6f | 451 | /* Decrement the loop counter */ |
emh203 | 0:3d9c67d97d6f | 452 | i--; |
emh203 | 0:3d9c67d97d6f | 453 | } while(i > 0u); |
emh203 | 0:3d9c67d97d6f | 454 | |
emh203 | 0:3d9c67d97d6f | 455 | #endif /* #ifndef ARM_MATH_CM0_FAMILY */ |
emh203 | 0:3d9c67d97d6f | 456 | |
emh203 | 0:3d9c67d97d6f | 457 | } |
emh203 | 0:3d9c67d97d6f | 458 | |
emh203 | 0:3d9c67d97d6f | 459 | /** |
emh203 | 0:3d9c67d97d6f | 460 | * @} end of DCT4_IDCT4 group |
emh203 | 0:3d9c67d97d6f | 461 | */ |