Subversion Repositories DashDisplay

/* ----------------------------------------------------------------------

 * Project:      CMSIS DSP Library

 * Title:        arm_rfft_q15.c

 * Description:  RFFT & RIFFT Q15 process function

 *

 * $Date:        27. January 2017

 * $Revision:    V.1.5.1

 *

 * Target Processor: Cortex-M cores

 * -------------------------------------------------------------------- */

/*

 * Copyright (C) 2010-2017 ARM Limited or its affiliates. All rights reserved.

 *

 * SPDX-License-Identifier: Apache-2.0

 *

 * Licensed under the Apache License, Version 2.0 (the License); you may

 * not use this file except in compliance with the License.

 * You may obtain a copy of the License at

 *

 * www.apache.org/licenses/LICENSE-2.0

 *

 * Unless required by applicable law or agreed to in writing, software

 * distributed under the License is distributed on an AS IS BASIS, WITHOUT

 * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.

 * See the License for the specific language governing permissions and

 * limitations under the License.

 */

#include "arm_math.h"

/* ----------------------------------------------------------------------

 * Internal functions prototypes

 * -------------------------------------------------------------------- */

void arm_split_rfft_q15(

    q15_t * pSrc,

    uint32_t fftLen,

    q15_t * pATable,

    q15_t * pBTable,

    q15_t * pDst,

    uint32_t modifier);

void arm_split_rifft_q15(

    q15_t * pSrc,

    uint32_t fftLen,

    q15_t * pATable,

    q15_t * pBTable,

    q15_t * pDst,

    uint32_t modifier);

/**

* @addtogroup RealFFT

* @{

*/

/**

* @brief Processing function for the Q15 RFFT/RIFFT.

* @param[in]  *S    points to an instance of the Q15 RFFT/RIFFT structure.

* @param[in]  *pSrc points to the input buffer.

* @param[out] *pDst points to the output buffer.

* @return none.

*

* \par Input an output formats:

* \par

* Internally input is downscaled by 2 for every stage to avoid saturations inside CFFT/CIFFT process.

* Hence the output format is different for different RFFT sizes.

* The input and output formats for different RFFT sizes and number of bits to upscale are mentioned in the tables below for RFFT and RIFFT:

* \par

* \image html RFFTQ15.gif "Input and Output Formats for Q15 RFFT"

* \par

* \image html RIFFTQ15.gif "Input and Output Formats for Q15 RIFFT"

*/

void arm_rfft_q15(

    const arm_rfft_instance_q15 * S,

    q15_t * pSrc,

    q15_t * pDst)

{

    const arm_cfft_instance_q15 *S_CFFT = S->pCfft;

    uint32_t i;

    uint32_t L2 = S->fftLenReal >> 1;

    /* Calculation of RIFFT of input */

    if (S->ifftFlagR == 1U)

    {

        /*  Real IFFT core process */

        arm_split_rifft_q15(pSrc, L2, S->pTwiddleAReal,

                            S->pTwiddleBReal, pDst, S->twidCoefRModifier);

        /* Complex IFFT process */

        arm_cfft_q15(S_CFFT, pDst, S->ifftFlagR, S->bitReverseFlagR);

        for(i=0;i<S->fftLenReal;i++)

        {

            pDst[i] = pDst[i] << 1;

        }

    }

    else

    {

        /* Calculation of RFFT of input */

        /* Complex FFT process */

        arm_cfft_q15(S_CFFT, pSrc, S->ifftFlagR, S->bitReverseFlagR);

        /*  Real FFT core process */

        arm_split_rfft_q15(pSrc, L2, S->pTwiddleAReal,

                            S->pTwiddleBReal, pDst, S->twidCoefRModifier);

    }

}

/**

* @} end of RealFFT group

*/

/**

* @brief  Core Real FFT process

* @param  *pSrc                                 points to the input buffer.

* @param  fftLen                                length of FFT.

* @param  *pATable                      points to the A twiddle Coef buffer.

* @param  *pBTable                      points to the B twiddle Coef buffer.

* @param  *pDst                                 points to the output buffer.

* @param  modifier              twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.

* @return none.

* The function implements a Real FFT

*/

void arm_split_rfft_q15(

    q15_t * pSrc,

    uint32_t fftLen,

    q15_t * pATable,

    q15_t * pBTable,

    q15_t * pDst,

    uint32_t modifier)

{

    uint32_t i;                                    /* Loop Counter */

    q31_t outR, outI;                              /* Temporary variables for output */

    q15_t *pCoefA, *pCoefB;                        /* Temporary pointers for twiddle factors */

    q15_t *pSrc1, *pSrc2;

#if defined (ARM_MATH_DSP)

    q15_t *pD1, *pD2;

#endif

    //  pSrc[2U * fftLen] = pSrc[0];

    //  pSrc[(2U * fftLen) + 1U] = pSrc[1];

    pCoefA = &pATable[modifier * 2U];

    pCoefB = &pBTable[modifier * 2U];

    pSrc1 = &pSrc[2];

    pSrc2 = &pSrc[(2U * fftLen) - 2U];

#if defined (ARM_MATH_DSP)

    /* Run the below code for Cortex-M4 and Cortex-M3 */

    i = 1U;

    pD1 = pDst + 2;

    pD2 = pDst + (4U * fftLen) - 2;

    for(i = fftLen - 1; i > 0; i--)

    {

        /*

        outR = (pSrc[2 * i] * pATable[2 * i] - pSrc[2 * i + 1] * pATable[2 * i + 1]

        + pSrc[2 * n - 2 * i] * pBTable[2 * i] +

        pSrc[2 * n - 2 * i + 1] * pBTable[2 * i + 1]);

        */

        /* outI = (pIn[2 * i + 1] * pATable[2 * i] + pIn[2 * i] * pATable[2 * i + 1] +

        pIn[2 * n - 2 * i] * pBTable[2 * i + 1] -

        pIn[2 * n - 2 * i + 1] * pBTable[2 * i]); */

#ifndef ARM_MATH_BIG_ENDIAN

        /* pSrc[2 * i] * pATable[2 * i] - pSrc[2 * i + 1] * pATable[2 * i + 1] */

        outR = __SMUSD(*__SIMD32(pSrc1), *__SIMD32(pCoefA));

#else

        /* -(pSrc[2 * i + 1] * pATable[2 * i + 1] - pSrc[2 * i] * pATable[2 * i]) */

        outR = -(__SMUSD(*__SIMD32(pSrc1), *__SIMD32(pCoefA)));

#endif /*      #ifndef ARM_MATH_BIG_ENDIAN     */

        /* pSrc[2 * n - 2 * i] * pBTable[2 * i] +

        pSrc[2 * n - 2 * i + 1] * pBTable[2 * i + 1]) */

        outR = __SMLAD(*__SIMD32(pSrc2), *__SIMD32(pCoefB), outR) >> 16U;

        /* pIn[2 * n - 2 * i] * pBTable[2 * i + 1] -

        pIn[2 * n - 2 * i + 1] * pBTable[2 * i] */

#ifndef ARM_MATH_BIG_ENDIAN

        outI = __SMUSDX(*__SIMD32(pSrc2)--, *__SIMD32(pCoefB));

#else

        outI = __SMUSDX(*__SIMD32(pCoefB), *__SIMD32(pSrc2)--);

#endif /*      #ifndef ARM_MATH_BIG_ENDIAN     */

        /* (pIn[2 * i + 1] * pATable[2 * i] + pIn[2 * i] * pATable[2 * i + 1] */

        outI = __SMLADX(*__SIMD32(pSrc1)++, *__SIMD32(pCoefA), outI);

        /* write output */

        *pD1++ = (q15_t) outR;

        *pD1++ = outI >> 16U;

        /* write complex conjugate output */

        pD2[0] = (q15_t) outR;

        pD2[1] = -(outI >> 16U);

        pD2 -= 2;

        /* update coefficient pointer */

        pCoefB = pCoefB + (2U * modifier);

        pCoefA = pCoefA + (2U * modifier);

    }

    pDst[2U * fftLen] = (pSrc[0] - pSrc[1]) >> 1;

    pDst[(2U * fftLen) + 1U] = 0;

    pDst[0] = (pSrc[0] + pSrc[1]) >> 1;

    pDst[1] = 0;

#else

    /* Run the below code for Cortex-M0 */

    i = 1U;

    while (i < fftLen)

    {

        /*

        outR = (pSrc[2 * i] * pATable[2 * i] - pSrc[2 * i + 1] * pATable[2 * i + 1]

        + pSrc[2 * n - 2 * i] * pBTable[2 * i] +

        pSrc[2 * n - 2 * i + 1] * pBTable[2 * i + 1]);

        */

        outR = *pSrc1 * *pCoefA;

        outR = outR - (*(pSrc1 + 1) * *(pCoefA + 1));

        outR = outR + (*pSrc2 * *pCoefB);

        outR = (outR + (*(pSrc2 + 1) * *(pCoefB + 1))) >> 16;

        /* outI = (pIn[2 * i + 1] * pATable[2 * i] + pIn[2 * i] * pATable[2 * i + 1] +

        pIn[2 * n - 2 * i] * pBTable[2 * i + 1] -

        pIn[2 * n - 2 * i + 1] * pBTable[2 * i]);

        */

        outI = *pSrc2 * *(pCoefB + 1);

        outI = outI - (*(pSrc2 + 1) * *pCoefB);

        outI = outI + (*(pSrc1 + 1) * *pCoefA);

        outI = outI + (*pSrc1 * *(pCoefA + 1));

        /* update input pointers */

        pSrc1 += 2U;

        pSrc2 -= 2U;

        /* write output */

        pDst[2U * i] = (q15_t) outR;

        pDst[(2U * i) + 1U] = outI >> 16U;

        /* write complex conjugate output */

        pDst[(4U * fftLen) - (2U * i)] = (q15_t) outR;

        pDst[((4U * fftLen) - (2U * i)) + 1U] = -(outI >> 16U);

        /* update coefficient pointer */

        pCoefB = pCoefB + (2U * modifier);

        pCoefA = pCoefA + (2U * modifier);

        i++;

    }

    pDst[2U * fftLen] = (pSrc[0] - pSrc[1]) >> 1;

    pDst[(2U * fftLen) + 1U] = 0;

    pDst[0] = (pSrc[0] + pSrc[1]) >> 1;

    pDst[1] = 0;

#endif /* #if defined (ARM_MATH_DSP) */

}

/**

* @brief  Core Real IFFT process

* @param[in]   *pSrc                            points to the input buffer.

* @param[in]   fftLen               length of FFT.

* @param[in]   *pATable                         points to the twiddle Coef A buffer.

* @param[in]   *pBTable                         points to the twiddle Coef B buffer.

* @param[out]  *pDst                            points to the output buffer.

* @param[in]   modifier                 twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.

* @return none.

* The function implements a Real IFFT

*/

void arm_split_rifft_q15(

    q15_t * pSrc,

    uint32_t fftLen,

    q15_t * pATable,

    q15_t * pBTable,

    q15_t * pDst,

    uint32_t modifier)

{

    uint32_t i;                                    /* Loop Counter */

    q31_t outR, outI;                              /* Temporary variables for output */

    q15_t *pCoefA, *pCoefB;                        /* Temporary pointers for twiddle factors */

    q15_t *pSrc1, *pSrc2;

    q15_t *pDst1 = &pDst[0];

    pCoefA = &pATable[0];

    pCoefB = &pBTable[0];

    pSrc1 = &pSrc[0];

    pSrc2 = &pSrc[2U * fftLen];

#if defined (ARM_MATH_DSP)

    /* Run the below code for Cortex-M4 and Cortex-M3 */

    i = fftLen;

    while (i > 0U)

    {

        /*

        outR = (pIn[2 * i] * pATable[2 * i] + pIn[2 * i + 1] * pATable[2 * i + 1] +

        pIn[2 * n - 2 * i] * pBTable[2 * i] -

        pIn[2 * n - 2 * i + 1] * pBTable[2 * i + 1]);

        outI = (pIn[2 * i + 1] * pATable[2 * i] - pIn[2 * i] * pATable[2 * i + 1] -

        pIn[2 * n - 2 * i] * pBTable[2 * i + 1] -

        pIn[2 * n - 2 * i + 1] * pBTable[2 * i]);

        */

#ifndef ARM_MATH_BIG_ENDIAN

        /* pIn[2 * n - 2 * i] * pBTable[2 * i] -

        pIn[2 * n - 2 * i + 1] * pBTable[2 * i + 1]) */

        outR = __SMUSD(*__SIMD32(pSrc2), *__SIMD32(pCoefB));

#else

        /* -(-pIn[2 * n - 2 * i] * pBTable[2 * i] +

        pIn[2 * n - 2 * i + 1] * pBTable[2 * i + 1])) */

        outR = -(__SMUSD(*__SIMD32(pSrc2), *__SIMD32(pCoefB)));

#endif /*      #ifndef ARM_MATH_BIG_ENDIAN     */

        /* pIn[2 * i] * pATable[2 * i] + pIn[2 * i + 1] * pATable[2 * i + 1] +

        pIn[2 * n - 2 * i] * pBTable[2 * i] */

        outR = __SMLAD(*__SIMD32(pSrc1), *__SIMD32(pCoefA), outR) >> 16U;

        /*

        -pIn[2 * n - 2 * i] * pBTable[2 * i + 1] +

        pIn[2 * n - 2 * i + 1] * pBTable[2 * i] */

        outI = __SMUADX(*__SIMD32(pSrc2)--, *__SIMD32(pCoefB));

        /* pIn[2 * i + 1] * pATable[2 * i] - pIn[2 * i] * pATable[2 * i + 1] */

#ifndef ARM_MATH_BIG_ENDIAN

        outI = __SMLSDX(*__SIMD32(pCoefA), *__SIMD32(pSrc1)++, -outI);

#else

        outI = __SMLSDX(*__SIMD32(pSrc1)++, *__SIMD32(pCoefA), -outI);

#endif /*      #ifndef ARM_MATH_BIG_ENDIAN     */

        /* write output */

#ifndef ARM_MATH_BIG_ENDIAN

        *__SIMD32(pDst1)++ = __PKHBT(outR, (outI >> 16U), 16);

#else

        *__SIMD32(pDst1)++ = __PKHBT((outI >> 16U), outR, 16);

#endif /*      #ifndef ARM_MATH_BIG_ENDIAN     */

        /* update coefficient pointer */

        pCoefB = pCoefB + (2U * modifier);

        pCoefA = pCoefA + (2U * modifier);

        i--;

    }

#else

    /* Run the below code for Cortex-M0 */

    i = fftLen;

    while (i > 0U)

    {

        /*

        outR = (pIn[2 * i] * pATable[2 * i] + pIn[2 * i + 1] * pATable[2 * i + 1] +

        pIn[2 * n - 2 * i] * pBTable[2 * i] -

        pIn[2 * n - 2 * i + 1] * pBTable[2 * i + 1]);

        */

        outR = *pSrc2 * *pCoefB;

        outR = outR - (*(pSrc2 + 1) * *(pCoefB + 1));

        outR = outR + (*pSrc1 * *pCoefA);

        outR = (outR + (*(pSrc1 + 1) * *(pCoefA + 1))) >> 16;

        /*

        outI = (pIn[2 * i + 1] * pATable[2 * i] - pIn[2 * i] * pATable[2 * i + 1] -

        pIn[2 * n - 2 * i] * pBTable[2 * i + 1] -

        pIn[2 * n - 2 * i + 1] * pBTable[2 * i]);

        */

        outI = *(pSrc1 + 1) * *pCoefA;

        outI = outI - (*pSrc1 * *(pCoefA + 1));

        outI = outI - (*pSrc2 * *(pCoefB + 1));

        outI = outI - (*(pSrc2 + 1) * *(pCoefB));

        /* update input pointers */

        pSrc1 += 2U;

        pSrc2 -= 2U;

        /* write output */

        *pDst1++ = (q15_t) outR;

        *pDst1++ = (q15_t) (outI >> 16);

        /* update coefficient pointer */

        pCoefB = pCoefB + (2U * modifier);

        pCoefA = pCoefA + (2U * modifier);

        i--;

    }

#endif /* #if defined (ARM_MATH_DSP) */

}