arm_scale_f32.c

00001 /* ----------------------------------------------------------------------
00002  * Project:      CMSIS DSP Library
00003  * Title:        arm_scale_f32.c
00004  * Description:  Multiplies a floating-point vector by a scalar
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 scale Vector Scale
00037  *
00038  * Multiply a vector by a scalar value.  For floating-point data, the algorithm used is:
00039  *
00040  * <pre>
00041  *     pDst[n] = pSrc[n] * scale,   0 <= n < blockSize.
00042  * </pre>
00043  *
00044  * In the fixed-point Q7, Q15, and Q31 functions, <code>scale</code> is represented by
00045  * a fractional multiplication <code>scaleFract</code> and an arithmetic shift <code>shift</code>.
00046  * The shift allows the gain of the scaling operation to exceed 1.0.
00047  * The algorithm used with fixed-point data is:
00048  *
00049  * <pre>
00050  *     pDst[n] = (pSrc[n] * scaleFract) << shift,   0 <= n < blockSize.
00051  * </pre>
00052  *
00053  * The overall scale factor applied to the fixed-point data is
00054  * <pre>
00055  *     scale = scaleFract * 2^shift.
00056  * </pre>
00057  *
00058  * The functions support in-place computation allowing the source and destination
00059  * pointers to reference the same memory buffer.
00060  */
00061 
00062 /**
00063  * @addtogroup scale
00064  * @{
00065  */
00066 
00067 /**
00068  * @brief Multiplies a floating-point vector by a scalar.
00069  * @param[in]       *pSrc points to the input vector
00070  * @param[in]       scale scale factor to be applied
00071  * @param[out]      *pDst points to the output vector
00072  * @param[in]       blockSize number of samples in the vector
00073  * @return none.
00074  */
00075 
00076 
00077 void arm_scale_f32(
00078   float32_t * pSrc,
00079   float32_t scale,
00080   float32_t * pDst,
00081   uint32_t blockSize)
00082 {
00083   uint32_t blkCnt;                               /* loop counter */
00084 #if defined (ARM_MATH_DSP)
00085 
00086 /* Run the below code for Cortex-M4 and Cortex-M3 */
00087   float32_t in1, in2, in3, in4;                  /* temporary variabels */
00088 
00089   /*loop Unrolling */
00090   blkCnt = blockSize >> 2U;
00091 
00092   /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.
00093    ** a second loop below computes the remaining 1 to 3 samples. */
00094   while (blkCnt > 0U)
00095   {
00096     /* C = A * scale */
00097     /* Scale the input and then store the results in the destination buffer. */
00098     /* read input samples from source */
00099     in1 = *pSrc;
00100     in2 = *(pSrc + 1);
00101 
00102     /* multiply with scaling factor */
00103     in1 = in1 * scale;
00104 
00105     /* read input sample from source */
00106     in3 = *(pSrc + 2);
00107 
00108     /* multiply with scaling factor */
00109     in2 = in2 * scale;
00110 
00111     /* read input sample from source */
00112     in4 = *(pSrc + 3);
00113 
00114     /* multiply with scaling factor */
00115     in3 = in3 * scale;
00116     in4 = in4 * scale;
00117     /* store the result to destination */
00118     *pDst = in1;
00119     *(pDst + 1) = in2;
00120     *(pDst + 2) = in3;
00121     *(pDst + 3) = in4;
00122 
00123     /* update pointers to process next samples */
00124     pSrc += 4U;
00125     pDst += 4U;
00126 
00127     /* Decrement the loop counter */
00128     blkCnt--;
00129   }
00130 
00131   /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
00132    ** No loop unrolling is used. */
00133   blkCnt = blockSize % 0x4U;
00134 
00135 #else
00136 
00137   /* Run the below code for Cortex-M0 */
00138 
00139   /* Initialize blkCnt with number of samples */
00140   blkCnt = blockSize;
00141 
00142 #endif /* #if defined (ARM_MATH_DSP) */
00143 
00144   while (blkCnt > 0U)
00145   {
00146     /* C = A * scale */
00147     /* Scale the input and then store the result in the destination buffer. */
00148     *pDst++ = (*pSrc++) * scale;
00149 
00150     /* Decrement the loop counter */
00151     blkCnt--;
00152   }
00153 }
00154 
00155 /**
00156  * @} end of scale group
00157  */
00158