arm_dot_prod_f32.c

00001 /* ----------------------------------------------------------------------
00002  * Project:      CMSIS DSP Library
00003  * Title:        arm_dot_prod_f32.c
00004  * Description:  Floating-point dot product
00005  *
00006  * $Date:        27. January 2017
00007  * $Revision:    V.1.5.1
00008  *
00009  * Target Processor: Cortex-M cores
00010  * -------------------------------------------------------------------- */
00011 /*
00012  * Copyright (C) 2010-2017 ARM Limited or its affiliates. All rights reserved.
00013  *
00014  * SPDX-License-Identifier: Apache-2.0
00015  *
00016  * Licensed under the Apache License, Version 2.0 (the License); you may
00017  * not use this file except in compliance with the License.
00018  * You may obtain a copy of the License at
00019  *
00020  * www.apache.org/licenses/LICENSE-2.0
00021  *
00022  * Unless required by applicable law or agreed to in writing, software
00023  * distributed under the License is distributed on an AS IS BASIS, WITHOUT
00024  * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
00025  * See the License for the specific language governing permissions and
00026  * limitations under the License.
00027  */
00028 
00029 #include "arm_math.h"
00030 
00031 /**
00032  * @ingroup groupMath
00033  */
00034 
00035 /**
00036  * @defgroup dot_prod Vector Dot Product
00037  *
00038  * Computes the dot product of two vectors.
00039  * The vectors are multiplied element-by-element and then summed.
00040  *
00041  * <pre>
00042  *     sum = pSrcA[0]*pSrcB[0] + pSrcA[1]*pSrcB[1] + ... + pSrcA[blockSize-1]*pSrcB[blockSize-1]
00043  * </pre>
00044  *
00045  * There are separate functions for floating-point, Q7, Q15, and Q31 data types.
00046  */
00047 
00048 /**
00049  * @addtogroup dot_prod
00050  * @{
00051  */
00052 
00053 /**
00054  * @brief Dot product of floating-point vectors.
00055  * @param[in]       *pSrcA points to the first input vector
00056  * @param[in]       *pSrcB points to the second input vector
00057  * @param[in]       blockSize number of samples in each vector
00058  * @param[out]      *result output result returned here
00059  * @return none.
00060  */
00061 
00062 
00063 void arm_dot_prod_f32(
00064   float32_t * pSrcA,
00065   float32_t * pSrcB,
00066   uint32_t blockSize,
00067   float32_t * result)
00068 {
00069   float32_t sum = 0.0f;                          /* Temporary result storage */
00070   uint32_t blkCnt;                               /* loop counter */
00071 
00072 
00073 #if defined (ARM_MATH_DSP)
00074 
00075 /* Run the below code for Cortex-M4 and Cortex-M3 */
00076   /*loop Unrolling */
00077   blkCnt = blockSize >> 2U;
00078 
00079   /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.
00080    ** a second loop below computes the remaining 1 to 3 samples. */
00081   while (blkCnt > 0U)
00082   {
00083     /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
00084     /* Calculate dot product and then store the result in a temporary buffer */
00085     sum += (*pSrcA++) * (*pSrcB++);
00086     sum += (*pSrcA++) * (*pSrcB++);
00087     sum += (*pSrcA++) * (*pSrcB++);
00088     sum += (*pSrcA++) * (*pSrcB++);
00089 
00090     /* Decrement the loop counter */
00091     blkCnt--;
00092   }
00093 
00094   /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
00095    ** No loop unrolling is used. */
00096   blkCnt = blockSize % 0x4U;
00097 
00098 #else
00099 
00100   /* Run the below code for Cortex-M0 */
00101 
00102   /* Initialize blkCnt with number of samples */
00103   blkCnt = blockSize;
00104 
00105 #endif /* #if defined (ARM_MATH_DSP) */
00106 
00107 
00108   while (blkCnt > 0U)
00109   {
00110     /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
00111     /* Calculate dot product and then store the result in a temporary buffer. */
00112     sum += (*pSrcA++) * (*pSrcB++);
00113 
00114     /* Decrement the loop counter */
00115     blkCnt--;
00116   }
00117   /* Store the result back in the destination buffer */
00118   *result = sum;
00119 }
00120 
00121 /**
00122  * @} end of dot_prod group
00123  */
00124