arm_cmplx_mult_real_q15.c

00001 /* ----------------------------------------------------------------------    
00002 * Copyright (C) 2010-2013 ARM Limited. All rights reserved.    
00003 *    
00004 * $Date:        17. January 2013
00005 * $Revision:    V1.4.1
00006 *    
00007 * Project:      CMSIS DSP Library    
00008 * Title:        arm_cmplx_mult_real_q15.c    
00009 *    
00010 * Description:  Q15 complex by real multiplication    
00011 *    
00012 * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
00013 *  
00014 * Redistribution and use in source and binary forms, with or without 
00015 * modification, are permitted provided that the following conditions
00016 * are met:
00017 *   - Redistributions of source code must retain the above copyright
00018 *     notice, this list of conditions and the following disclaimer.
00019 *   - Redistributions in binary form must reproduce the above copyright
00020 *     notice, this list of conditions and the following disclaimer in
00021 *     the documentation and/or other materials provided with the 
00022 *     distribution.
00023 *   - Neither the name of ARM LIMITED nor the names of its contributors
00024 *     may be used to endorse or promote products derived from this
00025 *     software without specific prior written permission.
00026 *
00027 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
00028 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
00029 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
00030 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE 
00031 * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
00032 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
00033 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
00034 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
00035 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
00036 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
00037 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
00038 * POSSIBILITY OF SUCH DAMAGE. 
00039 * -------------------------------------------------------------------- */
00040 
00041 #include "arm_math.h"
00042 
00043 /**    
00044  * @ingroup groupCmplxMath    
00045  */
00046 
00047 /**    
00048  * @addtogroup CmplxByRealMult    
00049  * @{    
00050  */
00051 
00052 
00053 /**    
00054  * @brief  Q15 complex-by-real multiplication    
00055  * @param[in]  *pSrcCmplx points to the complex input vector    
00056  * @param[in]  *pSrcReal points to the real input vector    
00057  * @param[out]  *pCmplxDst points to the complex output vector    
00058  * @param[in]  numSamples number of samples in each vector    
00059  * @return none.    
00060  *    
00061  * <b>Scaling and Overflow Behavior:</b>    
00062  * \par    
00063  * The function uses saturating arithmetic.    
00064  * Results outside of the allowable Q15 range [0x8000 0x7FFF] will be saturated.    
00065  */
00066 
00067 void arm_cmplx_mult_real_q15(
00068   q15_t * pSrcCmplx,
00069   q15_t * pSrcReal,
00070   q15_t * pCmplxDst,
00071   uint32_t numSamples)
00072 {
00073   q15_t in;                                      /* Temporary variable to store input value */
00074 
00075 #ifndef ARM_MATH_CM0_FAMILY
00076 
00077   /* Run the below code for Cortex-M4 and Cortex-M3 */
00078   uint32_t blkCnt;                               /* loop counters */
00079   q31_t inA1, inA2;                              /* Temporary variables to hold input data */
00080   q31_t inB1;                                    /* Temporary variables to hold input data */
00081   q15_t out1, out2, out3, out4;                  /* Temporary variables to hold output data */
00082   q31_t mul1, mul2, mul3, mul4;                  /* Temporary variables to hold intermediate data */
00083 
00084   /* loop Unrolling */
00085   blkCnt = numSamples >> 2u;
00086 
00087   /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.    
00088    ** a second loop below computes the remaining 1 to 3 samples. */
00089   while(blkCnt > 0u)
00090   {
00091     /* C[2 * i] = A[2 * i] * B[i].            */
00092     /* C[2 * i + 1] = A[2 * i + 1] * B[i].        */
00093     /* read complex number both real and imaginary from complex input buffer */
00094     inA1 = *__SIMD32(pSrcCmplx)++;
00095     /* read two real values at a time from real input buffer */
00096     inB1 = *__SIMD32(pSrcReal)++;
00097     /* read complex number both real and imaginary from complex input buffer */
00098     inA2 = *__SIMD32(pSrcCmplx)++;
00099 
00100     /* multiply complex number with real numbers */
00101 #ifndef ARM_MATH_BIG_ENDIAN
00102 
00103     mul1 = (q31_t) ((q15_t) (inA1) * (q15_t) (inB1));
00104     mul2 = (q31_t) ((q15_t) (inA1 >> 16) * (q15_t) (inB1));
00105     mul3 = (q31_t) ((q15_t) (inA2) * (q15_t) (inB1 >> 16));
00106     mul4 = (q31_t) ((q15_t) (inA2 >> 16) * (q15_t) (inB1 >> 16));
00107 
00108 #else
00109 
00110     mul2 = (q31_t) ((q15_t) (inA1 >> 16) * (q15_t) (inB1 >> 16));
00111     mul1 = (q31_t) ((q15_t) inA1 * (q15_t) (inB1 >> 16));
00112     mul4 = (q31_t) ((q15_t) (inA2 >> 16) * (q15_t) inB1);
00113     mul3 = (q31_t) ((q15_t) inA2 * (q15_t) inB1);
00114 
00115 #endif //      #ifndef ARM_MATH_BIG_ENDIAN
00116 
00117     /* saturate the result */
00118     out1 = (q15_t) __SSAT(mul1 >> 15u, 16);
00119     out2 = (q15_t) __SSAT(mul2 >> 15u, 16);
00120     out3 = (q15_t) __SSAT(mul3 >> 15u, 16);
00121     out4 = (q15_t) __SSAT(mul4 >> 15u, 16);
00122 
00123     /* pack real and imaginary outputs and store them to destination */
00124     *__SIMD32(pCmplxDst)++ = __PKHBT(out1, out2, 16);
00125     *__SIMD32(pCmplxDst)++ = __PKHBT(out3, out4, 16);
00126 
00127     inA1 = *__SIMD32(pSrcCmplx)++;
00128     inB1 = *__SIMD32(pSrcReal)++;
00129     inA2 = *__SIMD32(pSrcCmplx)++;
00130 
00131 #ifndef ARM_MATH_BIG_ENDIAN
00132 
00133     mul1 = (q31_t) ((q15_t) (inA1) * (q15_t) (inB1));
00134     mul2 = (q31_t) ((q15_t) (inA1 >> 16) * (q15_t) (inB1));
00135     mul3 = (q31_t) ((q15_t) (inA2) * (q15_t) (inB1 >> 16));
00136     mul4 = (q31_t) ((q15_t) (inA2 >> 16) * (q15_t) (inB1 >> 16));
00137 
00138 #else
00139 
00140     mul2 = (q31_t) ((q15_t) (inA1 >> 16) * (q15_t) (inB1 >> 16));
00141     mul1 = (q31_t) ((q15_t) inA1 * (q15_t) (inB1 >> 16));
00142     mul4 = (q31_t) ((q15_t) (inA2 >> 16) * (q15_t) inB1);
00143     mul3 = (q31_t) ((q15_t) inA2 * (q15_t) inB1);
00144 
00145 #endif //      #ifndef ARM_MATH_BIG_ENDIAN
00146 
00147     out1 = (q15_t) __SSAT(mul1 >> 15u, 16);
00148     out2 = (q15_t) __SSAT(mul2 >> 15u, 16);
00149     out3 = (q15_t) __SSAT(mul3 >> 15u, 16);
00150     out4 = (q15_t) __SSAT(mul4 >> 15u, 16);
00151 
00152     *__SIMD32(pCmplxDst)++ = __PKHBT(out1, out2, 16);
00153     *__SIMD32(pCmplxDst)++ = __PKHBT(out3, out4, 16);
00154 
00155     /* Decrement the numSamples loop counter */
00156     blkCnt--;
00157   }
00158 
00159   /* If the numSamples is not a multiple of 4, compute any remaining output samples here.    
00160    ** No loop unrolling is used. */
00161   blkCnt = numSamples % 0x4u;
00162 
00163   while(blkCnt > 0u)
00164   {
00165     /* C[2 * i] = A[2 * i] * B[i].            */
00166     /* C[2 * i + 1] = A[2 * i + 1] * B[i].        */
00167     in = *pSrcReal++;
00168     /* store the result in the destination buffer. */
00169     *pCmplxDst++ =
00170       (q15_t) __SSAT((((q31_t) (*pSrcCmplx++) * (in)) >> 15), 16);
00171     *pCmplxDst++ =
00172       (q15_t) __SSAT((((q31_t) (*pSrcCmplx++) * (in)) >> 15), 16);
00173 
00174     /* Decrement the numSamples loop counter */
00175     blkCnt--;
00176   }
00177 
00178 #else
00179 
00180   /* Run the below code for Cortex-M0 */
00181 
00182   while(numSamples > 0u)
00183   {
00184     /* realOut = realA * realB.            */
00185     /* imagOut = imagA * realB.                */
00186     in = *pSrcReal++;
00187     /* store the result in the destination buffer. */
00188     *pCmplxDst++ =
00189       (q15_t) __SSAT((((q31_t) (*pSrcCmplx++) * (in)) >> 15), 16);
00190     *pCmplxDst++ =
00191       (q15_t) __SSAT((((q31_t) (*pSrcCmplx++) * (in)) >> 15), 16);
00192 
00193     /* Decrement the numSamples loop counter */
00194     numSamples--;
00195   }
00196 
00197 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
00198 
00199 }
00200 
00201 /**    
00202  * @} end of CmplxByRealMult group    
00203  */