arm_biquad_cascade_stereo_df2T_f32.c

Go to the documentation of this file.

/* ----------------------------------------------------------------------    
* Copyright (C) 2010-2014 ARM Limited. All rights reserved.    
*    
* $Date:        19. March 2015 
* $Revision:    V.1.4.5
*    
* Project:      CMSIS DSP Library    
* Title:        arm_biquad_cascade_stereo_df2T_f32.c    
*    
* Description:  Processing function for the floating-point transposed    
*               direct form II Biquad cascade filter. 2 channels  
*    
* Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
*  
* Redistribution and use in source and binary forms, with or without 
* modification, are permitted provided that the following conditions
* are met:
*   - Redistributions of source code must retain the above copyright
*     notice, this list of conditions and the following disclaimer.
*   - Redistributions in binary form must reproduce the above copyright
*     notice, this list of conditions and the following disclaimer in
*     the documentation and/or other materials provided with the 
*     distribution.
*   - Neither the name of ARM LIMITED nor the names of its contributors
*     may be used to endorse or promote products derived from this
*     software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
* FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE 
* COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
* BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
* LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
* CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
* ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.   
* -------------------------------------------------------------------- */

#include "arm_math.h"

LOW_OPTIMIZATION_ENTER
void arm_biquad_cascade_stereo_df2T_f32(
const arm_biquad_cascade_stereo_df2T_instance_f32 * S,
float32_t * pSrc,
float32_t * pDst,
uint32_t blockSize)
{

    float32_t *pIn = pSrc;                         /*  source pointer            */
    float32_t *pOut = pDst;                        /*  destination pointer       */
    float32_t *pState = S->pState;                 /*  State pointer             */
    float32_t *pCoeffs = S->pCoeffs;               /*  coefficient pointer       */
    float32_t acc1a, acc1b;                        /*  accumulator               */
    float32_t b0, b1, b2, a1, a2;                  /*  Filter coefficients       */
    float32_t Xn1a, Xn1b;                          /*  temporary input           */
    float32_t d1a, d2a, d1b, d2b;                  /*  state variables           */
    uint32_t sample, stage = S->numStages;         /*  loop counters             */

#if defined(ARM_MATH_CM7)
    
    float32_t Xn2a, Xn3a, Xn4a, Xn5a, Xn6a, Xn7a, Xn8a;         /*  Input State variables     */
    float32_t Xn2b, Xn3b, Xn4b, Xn5b, Xn6b, Xn7b, Xn8b;         /*  Input State variables     */
    float32_t acc2a, acc3a, acc4a, acc5a, acc6a, acc7a, acc8a;  /*  Simulates the accumulator */
    float32_t acc2b, acc3b, acc4b, acc5b, acc6b, acc7b, acc8b;  /*  Simulates the accumulator */

    do
    {
        /* Reading the coefficients */ 
        b0 = pCoeffs[0]; 
        b1 = pCoeffs[1]; 
        b2 = pCoeffs[2]; 
        a1 = pCoeffs[3]; 
        /* Apply loop unrolling and compute 8 output values simultaneously. */ 
        sample = blockSize >> 3u; 
        a2 = pCoeffs[4]; 

        /*Reading the state values */ 
        d1a = pState[0]; 
        d2a = pState[1]; 
        d1b = pState[2]; 
        d2b = pState[3]; 

        pCoeffs += 5u;

        /* First part of the processing with loop unrolling.  Compute 8 outputs at a time.       
        ** a second loop below computes the remaining 1 to 7 samples. */
        while(sample > 0u) {

            /* y[n] = b0 * x[n] + d1 */
            /* d1 = b1 * x[n] + a1 * y[n] + d2 */
            /* d2 = b2 * x[n] + a2 * y[n] */

            /* Read the first 2 inputs. 2 cycles */
            Xn1a  = pIn[0 ];
            Xn1b  = pIn[1 ];

            /* Sample 1. 5 cycles */
            Xn2a  = pIn[2 ];
            acc1a = b0 * Xn1a + d1a;

            Xn2b  = pIn[3 ];
            d1a = b1 * Xn1a + d2a;

            Xn3a  = pIn[4 ];
            d2a = b2 * Xn1a;

            Xn3b  = pIn[5 ];
            d1a += a1 * acc1a;

            Xn4a  = pIn[6 ];
            d2a += a2 * acc1a;

            /* Sample 2. 5 cycles */
            Xn4b  = pIn[7 ];
            acc1b = b0 * Xn1b + d1b;

            Xn5a  = pIn[8 ];
            d1b = b1 * Xn1b + d2b;

            Xn5b = pIn[9 ];
            d2b = b2 * Xn1b;

            Xn6a = pIn[10];
            d1b += a1 * acc1b;

            Xn6b = pIn[11];
            d2b += a2 * acc1b;

            /* Sample 3. 5 cycles */
            Xn7a = pIn[12];
            acc2a = b0 * Xn2a + d1a;

            Xn7b = pIn[13];
            d1a = b1 * Xn2a + d2a;

            Xn8a = pIn[14];
            d2a = b2 * Xn2a;

            Xn8b = pIn[15];
            d1a += a1 * acc2a;

            pIn += 16;
            d2a += a2 * acc2a;

            /* Sample 4. 5 cycles */
            acc2b = b0 * Xn2b + d1b;
            d1b = b1 * Xn2b + d2b;
            d2b = b2 * Xn2b;
            d1b += a1 * acc2b;
            d2b += a2 * acc2b;

            /* Sample 5. 5 cycles */
            acc3a = b0 * Xn3a + d1a;
            d1a = b1 * Xn3a + d2a;
            d2a = b2 * Xn3a;
            d1a += a1 * acc3a;
            d2a += a2 * acc3a;

            /* Sample 6. 5 cycles */
            acc3b = b0 * Xn3b + d1b;
            d1b = b1 * Xn3b + d2b;
            d2b = b2 * Xn3b;
            d1b += a1 * acc3b;
            d2b += a2 * acc3b;

            /* Sample 7. 5 cycles */
            acc4a = b0 * Xn4a + d1a;
            d1a = b1 * Xn4a + d2a;
            d2a = b2 * Xn4a;
            d1a += a1 * acc4a;
            d2a += a2 * acc4a;

            /* Sample 8. 5 cycles */
            acc4b = b0 * Xn4b + d1b;
            d1b = b1 * Xn4b + d2b;
            d2b = b2 * Xn4b;
            d1b += a1 * acc4b;
            d2b += a2 * acc4b;

            /* Sample 9. 5 cycles */
            acc5a = b0 * Xn5a + d1a;
            d1a = b1 * Xn5a + d2a;
            d2a = b2 * Xn5a;
            d1a += a1 * acc5a;
            d2a += a2 * acc5a;

            /* Sample 10. 5 cycles */
            acc5b = b0 * Xn5b + d1b;
            d1b = b1 * Xn5b + d2b;
            d2b = b2 * Xn5b;
            d1b += a1 * acc5b;
            d2b += a2 * acc5b;

            /* Sample 11. 5 cycles */
            acc6a = b0 * Xn6a + d1a;
            d1a = b1 * Xn6a + d2a;
            d2a = b2 * Xn6a;
            d1a += a1 * acc6a;
            d2a += a2 * acc6a;

            /* Sample 12. 5 cycles */
            acc6b = b0 * Xn6b + d1b;
            d1b = b1 * Xn6b + d2b;
            d2b = b2 * Xn6b;
            d1b += a1 * acc6b;
            d2b += a2 * acc6b;

            /* Sample 13. 5 cycles */
            acc7a = b0 * Xn7a + d1a;         
            d1a = b1 * Xn7a + d2a;   
            
            pOut[0 ] = acc1a ;      
            d2a = b2 * Xn7a;

            pOut[1 ] = acc1b ;  
            d1a += a1 * acc7a;

            pOut[2 ] = acc2a ;  
            d2a += a2 * acc7a;

            /* Sample 14. 5 cycles */
            pOut[3 ] = acc2b ;
            acc7b = b0 * Xn7b + d1b;

            pOut[4 ] = acc3a ; 
            d1b = b1 * Xn7b + d2b;

            pOut[5 ] = acc3b ;  
            d2b = b2 * Xn7b;

            pOut[6 ] = acc4a ;    
            d1b += a1 * acc7b;

            pOut[7 ] = acc4b ;
            d2b += a2 * acc7b;

            /* Sample 15. 5 cycles */
            pOut[8 ] = acc5a ;  
            acc8a = b0 * Xn8a + d1a;

            pOut[9 ] = acc5b;   
            d1a = b1 * Xn8a + d2a;

            pOut[10] = acc6a;   
            d2a = b2 * Xn8a;

            pOut[11] = acc6b;
            d1a += a1 * acc8a;

            pOut[12] = acc7a;
            d2a += a2 * acc8a;

            /* Sample 16. 5 cycles */
            pOut[13] = acc7b;   
            acc8b = b0 * Xn8b + d1b;

            pOut[14] = acc8a;   
            d1b = b1 * Xn8b + d2b;

            pOut[15] = acc8b;
            d2b = b2 * Xn8b;

            sample--;    
            d1b += a1 * acc8b;

            pOut += 16;
            d2b += a2 * acc8b;
        }

        sample = blockSize & 0x7u;
        while(sample > 0u) {
            /* Read the input */
            Xn1a = *pIn++; //Channel a
            Xn1b = *pIn++; //Channel b

            /* y[n] = b0 * x[n] + d1 */
            acc1a = (b0 * Xn1a) + d1a;
            acc1b = (b0 * Xn1b) + d1b;

            /* Store the result in the accumulator in the destination buffer. */
            *pOut++ = acc1a;
            *pOut++ = acc1b;

            /* Every time after the output is computed state should be updated. */
            /* d1 = b1 * x[n] + a1 * y[n] + d2 */
            d1a = ((b1 * Xn1a) + (a1 * acc1a)) + d2a;
            d1b = ((b1 * Xn1b) + (a1 * acc1b)) + d2b;

            /* d2 = b2 * x[n] + a2 * y[n] */
            d2a = (b2 * Xn1a) + (a2 * acc1a);
            d2b = (b2 * Xn1b) + (a2 * acc1b);

            sample--;   
        }

        /* Store the updated state variables back into the state array */ 
        pState[0] = d1a; 
        pState[1] = d2a;         

        pState[2] = d1b; 
        pState[3] = d2b; 
        
        /* The current stage input is given as the output to the next stage */ 
        pIn = pDst; 
        /* decrement the loop counter */ 
        stage--; 

        pState += 4u;
        /*Reset the output working pointer */ 
        pOut = pDst; 

    } while(stage > 0u);
    
#elif defined(ARM_MATH_CM0_FAMILY)

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

    do
    {
        /* Reading the coefficients */
        b0 = *pCoeffs++;
        b1 = *pCoeffs++;
        b2 = *pCoeffs++;
        a1 = *pCoeffs++;
        a2 = *pCoeffs++;

        /*Reading the state values */
        d1a = pState[0];
        d2a = pState[1];
        d1b = pState[2];
        d2b = pState[3];


        sample = blockSize;

        while(sample > 0u)
        {
            /* Read the input */
            Xn1a = *pIn++; //Channel a
            Xn1b = *pIn++; //Channel b

            /* y[n] = b0 * x[n] + d1 */
            acc1a = (b0 * Xn1a) + d1a;
            acc1b = (b0 * Xn1b) + d1b;

            /* Store the result in the accumulator in the destination buffer. */
            *pOut++ = acc1a;
            *pOut++ = acc1b;

            /* Every time after the output is computed state should be updated. */
            /* d1 = b1 * x[n] + a1 * y[n] + d2 */
            d1a = ((b1 * Xn1a) + (a1 * acc1a)) + d2a;
            d1b = ((b1 * Xn1b) + (a1 * acc1b)) + d2b;

            /* d2 = b2 * x[n] + a2 * y[n] */
            d2a = (b2 * Xn1a) + (a2 * acc1a);
            d2b = (b2 * Xn1b) + (a2 * acc1b);

            /* decrement the loop counter */
            sample--;
        }

        /* Store the updated state variables back into the state array */
        *pState++ = d1a;
        *pState++ = d2a;
        *pState++ = d1b;
        *pState++ = d2b;

        /* The current stage input is given as the output to the next stage */
        pIn = pDst;

        /*Reset the output working pointer */
        pOut = pDst;

        /* decrement the loop counter */
        stage--;

    } while(stage > 0u);
     
#else

    float32_t Xn2a, Xn3a, Xn4a;                          /*  Input State variables     */
    float32_t Xn2b, Xn3b, Xn4b;                          /*  Input State variables     */
    float32_t acc2a, acc3a, acc4a;                       /*  accumulator               */
    float32_t acc2b, acc3b, acc4b;                       /*  accumulator               */
    float32_t p0a, p1a, p2a, p3a, p4a, A1a;
    float32_t p0b, p1b, p2b, p3b, p4b, A1b;

    /* Run the below code for Cortex-M4 and Cortex-M3 */
    do
    {
        /* Reading the coefficients */     
        b0 = *pCoeffs++;
        b1 = *pCoeffs++;
        b2 = *pCoeffs++;
        a1 = *pCoeffs++;
        a2 = *pCoeffs++;      

        /*Reading the state values */
        d1a = pState[0];
        d2a = pState[1];
        d1b = pState[2];
        d2b = pState[3];

        /* Apply loop unrolling and compute 4 output values simultaneously. */
        sample = blockSize >> 2u;

        /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.       
        ** a second loop below computes the remaining 1 to 3 samples. */
        while(sample > 0u) {

            /* y[n] = b0 * x[n] + d1 */
            /* d1 = b1 * x[n] + a1 * y[n] + d2 */
            /* d2 = b2 * x[n] + a2 * y[n] */

            /* Read the four inputs */
            Xn1a = pIn[0];
            Xn1b = pIn[1];
            Xn2a = pIn[2];
            Xn2b = pIn[3];
            Xn3a = pIn[4];
            Xn3b = pIn[5];
            Xn4a = pIn[6];
            Xn4b = pIn[7];
            pIn += 8;     
            
            p0a = b0 * Xn1a; 
            p0b = b0 * Xn1b; 
            p1a = b1 * Xn1a;
            p1b = b1 * Xn1b;
            acc1a = p0a + d1a;
            acc1b = p0b + d1b;
            p0a = b0 * Xn2a; 
            p0b = b0 * Xn2b; 
            p3a = a1 * acc1a;
            p3b = a1 * acc1b;
            p2a = b2 * Xn1a;
            p2b = b2 * Xn1b;
            A1a = p1a + p3a;
            A1b = p1b + p3b;
            p4a = a2 * acc1a;
            p4b = a2 * acc1b;
            d1a = A1a + d2a;
            d1b = A1b + d2b;
            d2a = p2a + p4a;
            d2b = p2b + p4b;
            
            p1a = b1 * Xn2a;
            p1b = b1 * Xn2b;
            acc2a = p0a + d1a;
            acc2b = p0b + d1b;
            p0a = b0 * Xn3a; 
            p0b = b0 * Xn3b; 
            p3a = a1 * acc2a;
            p3b = a1 * acc2b;
            p2a = b2 * Xn2a;
            p2b = b2 * Xn2b;
            A1a = p1a + p3a;
            A1b = p1b + p3b;
            p4a = a2 * acc2a;
            p4b = a2 * acc2b;
            d1a = A1a + d2a;
            d1b = A1b + d2b;
            d2a = p2a + p4a;
            d2b = p2b + p4b;
            
            p1a = b1 * Xn3a;
            p1b = b1 * Xn3b;
            acc3a = p0a + d1a;
            acc3b = p0b + d1b;
            p0a = b0 * Xn4a; 
            p0b = b0 * Xn4b; 
            p3a = a1 * acc3a;
            p3b = a1 * acc3b;
            p2a = b2 * Xn3a;
            p2b = b2 * Xn3b;
            A1a = p1a + p3a;
            A1b = p1b + p3b;
            p4a = a2 * acc3a;
            p4b = a2 * acc3b;
            d1a = A1a + d2a;
            d1b = A1b + d2b;
            d2a = p2a + p4a;
            d2b = p2b + p4b;
            
            acc4a = p0a + d1a;
            acc4b = p0b + d1b;
            p1a = b1 * Xn4a;
            p1b = b1 * Xn4b;
            p3a = a1 * acc4a;
            p3b = a1 * acc4b;
            p2a = b2 * Xn4a;
            p2b = b2 * Xn4b;
            A1a = p1a + p3a;
            A1b = p1b + p3b;
            p4a = a2 * acc4a;
            p4b = a2 * acc4b;
            d1a = A1a + d2a;
            d1b = A1b + d2b;
            d2a = p2a + p4a;
            d2b = p2b + p4b;

            pOut[0] = acc1a;    
            pOut[1] = acc1b;    
            pOut[2] = acc2a;    
            pOut[3] = acc2b;
            pOut[4] = acc3a;    
            pOut[5] = acc3b;    
            pOut[6] = acc4a;    
            pOut[7] = acc4b;
            pOut += 8;
             
            sample--;          
        }

        sample = blockSize & 0x3u;
        while(sample > 0u) {
            Xn1a = *pIn++;
            Xn1b = *pIn++;

            p0a = b0 * Xn1a; 
            p0b = b0 * Xn1b; 
            p1a = b1 * Xn1a;
            p1b = b1 * Xn1b;
            acc1a = p0a + d1a;
            acc1b = p0b + d1b;
            p3a = a1 * acc1a;
            p3b = a1 * acc1b;
            p2a = b2 * Xn1a;
            p2b = b2 * Xn1b;
            A1a = p1a + p3a;
            A1b = p1b + p3b;
            p4a = a2 * acc1a;
            p4b = a2 * acc1b;
            d1a = A1a + d2a;
            d1b = A1b + d2b;
            d2a = p2a + p4a;
            d2b = p2b + p4b;

            *pOut++ = acc1a;
            *pOut++ = acc1b;

            sample--;          
        }

        /* Store the updated state variables back into the state array */
        *pState++ = d1a;
        *pState++ = d2a;
        *pState++ = d1b;
        *pState++ = d2b;

        /* The current stage input is given as the output to the next stage */
        pIn = pDst;

        /*Reset the output working pointer */
        pOut = pDst;

        /* decrement the loop counter */
        stage--;

    } while(stage > 0u);

#endif 

}
LOW_OPTIMIZATION_EXIT