arm_mat_trans_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_mat_trans_q15.c    
00009 *    
00010 * Description:  Q15 matrix transpose.    
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 groupMatrix    
00045  */
00046 
00047 /**    
00048  * @addtogroup MatrixTrans    
00049  * @{    
00050  */
00051 
00052 /*    
00053  * @brief Q15 matrix transpose.    
00054  * @param[in]  *pSrc points to the input matrix    
00055  * @param[out] *pDst points to the output matrix    
00056  * @return  The function returns either  <code>ARM_MATH_SIZE_MISMATCH</code>    
00057  * or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.    
00058  */
00059 
00060 arm_status arm_mat_trans_q15(
00061   const arm_matrix_instance_q15 * pSrc,
00062   arm_matrix_instance_q15 * pDst)
00063 {
00064   q15_t *pSrcA = pSrc->pData;                    /* input data matrix pointer */
00065   q15_t *pOut = pDst->pData;                     /* output data matrix pointer */
00066   uint16_t nRows = pSrc->numRows;                /* number of nRows */
00067   uint16_t nColumns = pSrc->numCols;             /* number of nColumns */
00068   uint16_t col, row = nRows, i = 0u;             /* row and column loop counters */
00069   arm_status status;                             /* status of matrix transpose */
00070 
00071 #ifndef ARM_MATH_CM0_FAMILY
00072 
00073   /* Run the below code for Cortex-M4 and Cortex-M3 */
00074 #ifndef UNALIGNED_SUPPORT_DISABLE
00075 
00076   q31_t in;                                      /* variable to hold temporary output  */
00077 
00078 #else
00079 
00080   q15_t in;
00081 
00082 #endif  /*  #ifndef UNALIGNED_SUPPORT_DISABLE   */
00083 
00084 #ifdef ARM_MATH_MATRIX_CHECK
00085 
00086 
00087   /* Check for matrix mismatch condition */
00088   if((pSrc->numRows != pDst->numCols) || (pSrc->numCols != pDst->numRows))
00089   {
00090     /* Set status as ARM_MATH_SIZE_MISMATCH */
00091     status = ARM_MATH_SIZE_MISMATCH;
00092   }
00093   else
00094 #endif /*      #ifdef ARM_MATH_MATRIX_CHECK    */
00095 
00096   {
00097     /* Matrix transpose by exchanging the rows with columns */
00098     /* row loop     */
00099     do
00100     {
00101 
00102       /* Apply loop unrolling and exchange the columns with row elements */
00103       col = nColumns >> 2u;
00104 
00105       /* The pointer pOut is set to starting address of the column being processed */
00106       pOut = pDst->pData + i;
00107 
00108       /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.    
00109        ** a second loop below computes the remaining 1 to 3 samples. */
00110       while(col > 0u)
00111       {
00112 #ifndef UNALIGNED_SUPPORT_DISABLE
00113 
00114         /* Read two elements from the row */
00115         in = *__SIMD32(pSrcA)++;
00116 
00117         /* Unpack and store one element in the destination */
00118 #ifndef ARM_MATH_BIG_ENDIAN
00119 
00120         *pOut = (q15_t) in;
00121 
00122 #else
00123 
00124         *pOut = (q15_t) ((in & (q31_t) 0xffff0000) >> 16);
00125 
00126 #endif /*    #ifndef ARM_MATH_BIG_ENDIAN    */
00127 
00128         /* Update the pointer pOut to point to the next row of the transposed matrix */
00129         pOut += nRows;
00130 
00131         /* Unpack and store the second element in the destination */
00132 
00133 #ifndef ARM_MATH_BIG_ENDIAN
00134 
00135         *pOut = (q15_t) ((in & (q31_t) 0xffff0000) >> 16);
00136 
00137 #else
00138 
00139         *pOut = (q15_t) in;
00140 
00141 #endif /*    #ifndef ARM_MATH_BIG_ENDIAN    */
00142 
00143         /* Update the pointer pOut to point to the next row of the transposed matrix */
00144         pOut += nRows;
00145 
00146         /* Read two elements from the row */
00147 #ifndef ARM_MATH_BIG_ENDIAN
00148 
00149         in = *__SIMD32(pSrcA)++;
00150 
00151 #else
00152 
00153         in = *__SIMD32(pSrcA)++;
00154 
00155 #endif /*    #ifndef ARM_MATH_BIG_ENDIAN    */
00156 
00157         /* Unpack and store one element in the destination */
00158 #ifndef ARM_MATH_BIG_ENDIAN
00159 
00160         *pOut = (q15_t) in;
00161 
00162 #else
00163 
00164         *pOut = (q15_t) ((in & (q31_t) 0xffff0000) >> 16);
00165 
00166 #endif /*    #ifndef ARM_MATH_BIG_ENDIAN    */
00167 
00168         /* Update the pointer pOut to point to the next row of the transposed matrix */
00169         pOut += nRows;
00170 
00171         /* Unpack and store the second element in the destination */
00172 #ifndef ARM_MATH_BIG_ENDIAN
00173 
00174         *pOut = (q15_t) ((in & (q31_t) 0xffff0000) >> 16);
00175 
00176 #else
00177 
00178         *pOut = (q15_t) in;
00179 
00180 #endif /*    #ifndef ARM_MATH_BIG_ENDIAN    */
00181 
00182 #else    
00183         /* Read one element from the row */
00184         in = *pSrcA++;
00185 
00186         /* Store one element in the destination */
00187         *pOut = in;
00188  
00189         /* Update the pointer px to point to the next row of the transposed matrix */
00190         pOut += nRows;
00191 
00192         /* Read one element from the row */
00193         in = *pSrcA++;
00194 
00195         /* Store one element in the destination */
00196         *pOut = in;
00197  
00198         /* Update the pointer px to point to the next row of the transposed matrix */
00199         pOut += nRows;
00200 
00201         /* Read one element from the row */
00202         in = *pSrcA++;
00203 
00204         /* Store one element in the destination */
00205         *pOut = in;
00206  
00207         /* Update the pointer px to point to the next row of the transposed matrix */
00208         pOut += nRows;
00209 
00210         /* Read one element from the row */
00211         in = *pSrcA++;
00212 
00213         /* Store one element in the destination */
00214         *pOut = in;
00215 
00216 #endif  /*  #ifndef UNALIGNED_SUPPORT_DISABLE   */
00217 
00218         /* Update the pointer pOut to point to the next row of the transposed matrix */
00219         pOut += nRows;
00220 
00221         /* Decrement the column loop counter */
00222         col--;
00223       }
00224 
00225       /* Perform matrix transpose for last 3 samples here. */
00226       col = nColumns % 0x4u;
00227 
00228 #else
00229 
00230   /* Run the below code for Cortex-M0 */
00231 
00232 #ifdef ARM_MATH_MATRIX_CHECK
00233 
00234   /* Check for matrix mismatch condition */
00235   if((pSrc->numRows != pDst->numCols) || (pSrc->numCols != pDst->numRows))
00236   {
00237     /* Set status as ARM_MATH_SIZE_MISMATCH */
00238     status = ARM_MATH_SIZE_MISMATCH;
00239   }
00240   else
00241 #endif /*    #ifdef ARM_MATH_MATRIX_CHECK    */
00242 
00243   {
00244     /* Matrix transpose by exchanging the rows with columns */
00245     /* row loop     */
00246     do
00247     {
00248       /* The pointer pOut is set to starting address of the column being processed */
00249       pOut = pDst->pData + i;
00250 
00251       /* Initialize column loop counter */
00252       col = nColumns;
00253 
00254 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
00255 
00256       while(col > 0u)
00257       {
00258         /* Read and store the input element in the destination */
00259         *pOut = *pSrcA++;
00260 
00261         /* Update the pointer pOut to point to the next row of the transposed matrix */
00262         pOut += nRows;
00263 
00264         /* Decrement the column loop counter */
00265         col--;
00266       }
00267 
00268       i++;
00269 
00270       /* Decrement the row loop counter */
00271       row--;
00272 
00273     } while(row > 0u);
00274 
00275     /* set status as ARM_MATH_SUCCESS */
00276     status = ARM_MATH_SUCCESS;
00277   }
00278   /* Return to application */
00279   return (status);
00280 }
00281 
00282 /**    
00283  * @} end of MatrixTrans group    
00284  */