arm_scale_q15.c

00001 /* ----------------------------------------------------------------------    
00002 * Copyright (C) 2010-2014 ARM Limited. All rights reserved.    
00003 *    
00004 * $Date:        12. March 2014
00005 * $Revision:    V1.4.3
00006 *    
00007 * Project:      CMSIS DSP Library    
00008 * Title:        arm_scale_q15.c    
00009 *    
00010 * Description:  Multiplies a Q15 vector by a scalar.    
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 groupMath    
00045  */
00046 
00047 /**    
00048  * @addtogroup scale    
00049  * @{    
00050  */
00051 
00052 /**    
00053  * @brief Multiplies a Q15 vector by a scalar.    
00054  * @param[in]       *pSrc points to the input vector    
00055  * @param[in]       scaleFract fractional portion of the scale value    
00056  * @param[in]       shift number of bits to shift the result by    
00057  * @param[out]      *pDst points to the output vector    
00058  * @param[in]       blockSize number of samples in the vector    
00059  * @return none.    
00060  *    
00061  * <b>Scaling and Overflow Behavior:</b>    
00062  * \par    
00063  * The input data <code>*pSrc</code> and <code>scaleFract</code> are in 1.15 format.    
00064  * These are multiplied to yield a 2.30 intermediate result and this is shifted with saturation to 1.15 format.    
00065  */
00066 
00067 
00068 void arm_scale_q15(
00069   q15_t * pSrc,
00070   q15_t scaleFract,
00071   int8_t shift,
00072   q15_t * pDst,
00073   uint32_t blockSize)
00074 {
00075   int8_t kShift = 15 - shift;                    /* shift to apply after scaling */
00076   uint32_t blkCnt;                               /* loop counter */
00077 
00078 #ifndef ARM_MATH_CM0_FAMILY
00079 
00080 /* Run the below code for Cortex-M4 and Cortex-M3 */
00081   q15_t in1, in2, in3, in4;
00082   q31_t inA1, inA2;                              /* Temporary variables */
00083   q31_t out1, out2, out3, out4;
00084 
00085 
00086   /*loop Unrolling */
00087   blkCnt = blockSize >> 2u;
00088 
00089   /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.        
00090    ** a second loop below computes the remaining 1 to 3 samples. */
00091   while(blkCnt > 0u)
00092   {
00093     /* Reading 2 inputs from memory */
00094     inA1 = *__SIMD32(pSrc)++;
00095     inA2 = *__SIMD32(pSrc)++;
00096 
00097     /* C = A * scale */
00098     /* Scale the inputs and then store the 2 results in the destination buffer        
00099      * in single cycle by packing the outputs */
00100     out1 = (q31_t) ((q15_t) (inA1 >> 16) * scaleFract);
00101     out2 = (q31_t) ((q15_t) inA1 * scaleFract);
00102     out3 = (q31_t) ((q15_t) (inA2 >> 16) * scaleFract);
00103     out4 = (q31_t) ((q15_t) inA2 * scaleFract);
00104 
00105     /* apply shifting */
00106     out1 = out1 >> kShift;
00107     out2 = out2 >> kShift;
00108     out3 = out3 >> kShift;
00109     out4 = out4 >> kShift;
00110 
00111     /* saturate the output */
00112     in1 = (q15_t) (__SSAT(out1, 16));
00113     in2 = (q15_t) (__SSAT(out2, 16));
00114     in3 = (q15_t) (__SSAT(out3, 16));
00115     in4 = (q15_t) (__SSAT(out4, 16));
00116 
00117     /* store the result to destination */
00118     *__SIMD32(pDst)++ = __PKHBT(in2, in1, 16);
00119     *__SIMD32(pDst)++ = __PKHBT(in4, in3, 16);
00120 
00121     /* Decrement the loop counter */
00122     blkCnt--;
00123   }
00124 
00125   /* If the blockSize is not a multiple of 4, compute any remaining output samples here.    
00126    ** No loop unrolling is used. */
00127   blkCnt = blockSize % 0x4u;
00128 
00129   while(blkCnt > 0u)
00130   {
00131     /* C = A * scale */
00132     /* Scale the input and then store the result in the destination buffer. */
00133     *pDst++ = (q15_t) (__SSAT(((*pSrc++) * scaleFract) >> kShift, 16));
00134 
00135     /* Decrement the loop counter */
00136     blkCnt--;
00137   }
00138 
00139 #else
00140 
00141   /* Run the below code for Cortex-M0 */
00142 
00143   /* Initialize blkCnt with number of samples */
00144   blkCnt = blockSize;
00145 
00146   while(blkCnt > 0u)
00147   {
00148     /* C = A * scale */
00149     /* Scale the input and then store the result in the destination buffer. */
00150     *pDst++ = (q15_t) (__SSAT(((q31_t) * pSrc++ * scaleFract) >> kShift, 16));
00151 
00152     /* Decrement the loop counter */
00153     blkCnt--;
00154   }
00155 
00156 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
00157 
00158 }
00159 
00160 /**    
00161  * @} end of scale group    
00162  */