arm_mat_trans_f32.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_mat_trans_f32.c    
00009 *    
00010 * Description:  Floating-point 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 /**    
00042  * @defgroup MatrixTrans Matrix Transpose    
00043  *    
00044  * Tranposes a matrix.    
00045  * Transposing an <code>M x N</code> matrix flips it around the center diagonal and results in an <code>N x M</code> matrix.    
00046  * \image html MatrixTranspose.gif "Transpose of a 3 x 3 matrix"    
00047  */
00048 
00049 #include "arm_math.h"
00050 
00051 /**    
00052  * @ingroup groupMatrix    
00053  */
00054 
00055 /**    
00056  * @addtogroup MatrixTrans    
00057  * @{    
00058  */
00059 
00060 /**    
00061   * @brief Floating-point matrix transpose.    
00062   * @param[in]  *pSrc points to the input matrix    
00063   * @param[out] *pDst points to the output matrix    
00064   * @return     The function returns either  <code>ARM_MATH_SIZE_MISMATCH</code>    
00065   * or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.    
00066   */
00067 
00068 
00069 arm_status arm_mat_trans_f32(
00070   const arm_matrix_instance_f32 * pSrc,
00071   arm_matrix_instance_f32 * pDst)
00072 {
00073   float32_t *pIn = pSrc->pData;                  /* input data matrix pointer */
00074   float32_t *pOut = pDst->pData;                 /* output data matrix pointer */
00075   float32_t *px;                                 /* Temporary output data matrix pointer */
00076   uint16_t nRows = pSrc->numRows;                /* number of rows */
00077   uint16_t nColumns = pSrc->numCols;             /* number of columns */
00078 
00079 #ifndef ARM_MATH_CM0_FAMILY
00080 
00081   /* Run the below code for Cortex-M4 and Cortex-M3 */
00082 
00083   uint16_t blkCnt, i = 0u, row = nRows;          /* loop counters */
00084   arm_status status;                             /* status of matrix transpose  */
00085 
00086 
00087 #ifdef ARM_MATH_MATRIX_CHECK
00088 
00089 
00090   /* Check for matrix mismatch condition */
00091   if((pSrc->numRows != pDst->numCols) || (pSrc->numCols != pDst->numRows))
00092   {
00093     /* Set status as ARM_MATH_SIZE_MISMATCH */
00094     status = ARM_MATH_SIZE_MISMATCH;
00095   }
00096   else
00097 #endif /*    #ifdef ARM_MATH_MATRIX_CHECK    */
00098 
00099   {
00100     /* Matrix transpose by exchanging the rows with columns */
00101     /* row loop     */
00102     do
00103     {
00104       /* Loop Unrolling */
00105       blkCnt = nColumns >> 2;
00106 
00107       /* The pointer px is set to starting address of the column being processed */
00108       px = pOut + i;
00109 
00110       /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.    
00111        ** a second loop below computes the remaining 1 to 3 samples. */
00112       while(blkCnt > 0u)        /* column loop */
00113       {
00114         /* Read and store the input element in the destination */
00115         *px = *pIn++;
00116 
00117         /* Update the pointer px to point to the next row of the transposed matrix */
00118         px += nRows;
00119 
00120         /* Read and store the input element in the destination */
00121         *px = *pIn++;
00122 
00123         /* Update the pointer px to point to the next row of the transposed matrix */
00124         px += nRows;
00125 
00126         /* Read and store the input element in the destination */
00127         *px = *pIn++;
00128 
00129         /* Update the pointer px to point to the next row of the transposed matrix */
00130         px += nRows;
00131 
00132         /* Read and store the input element in the destination */
00133         *px = *pIn++;
00134 
00135         /* Update the pointer px to point to the next row of the transposed matrix */
00136         px += nRows;
00137 
00138         /* Decrement the column loop counter */
00139         blkCnt--;
00140       }
00141 
00142       /* Perform matrix transpose for last 3 samples here. */
00143       blkCnt = nColumns % 0x4u;
00144 
00145       while(blkCnt > 0u)
00146       {
00147         /* Read and store the input element in the destination */
00148         *px = *pIn++;
00149 
00150         /* Update the pointer px to point to the next row of the transposed matrix */
00151         px += nRows;
00152 
00153         /* Decrement the column loop counter */
00154         blkCnt--;
00155       }
00156 
00157 #else
00158 
00159   /* Run the below code for Cortex-M0 */
00160 
00161   uint16_t col, i = 0u, row = nRows;             /* loop counters */
00162   arm_status status;                             /* status of matrix transpose  */
00163 
00164 
00165 #ifdef ARM_MATH_MATRIX_CHECK
00166 
00167   /* Check for matrix mismatch condition */
00168   if((pSrc->numRows != pDst->numCols) || (pSrc->numCols != pDst->numRows))
00169   {
00170     /* Set status as ARM_MATH_SIZE_MISMATCH */
00171     status = ARM_MATH_SIZE_MISMATCH;
00172   }
00173   else
00174 #endif /*      #ifdef ARM_MATH_MATRIX_CHECK    */
00175 
00176   {
00177     /* Matrix transpose by exchanging the rows with columns */
00178     /* row loop     */
00179     do
00180     {
00181       /* The pointer px is set to starting address of the column being processed */
00182       px = pOut + i;
00183 
00184       /* Initialize column loop counter */
00185       col = nColumns;
00186 
00187       while(col > 0u)
00188       {
00189         /* Read and store the input element in the destination */
00190         *px = *pIn++;
00191 
00192         /* Update the pointer px to point to the next row of the transposed matrix */
00193         px += nRows;
00194 
00195         /* Decrement the column loop counter */
00196         col--;
00197       }
00198 
00199 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
00200 
00201       i++;
00202 
00203       /* Decrement the row loop counter */
00204       row--;
00205 
00206     } while(row > 0u);          /* row loop end  */
00207 
00208     /* Set status as ARM_MATH_SUCCESS */
00209     status = ARM_MATH_SUCCESS;
00210   }
00211 
00212   /* Return to application */
00213   return (status);
00214 }
00215 
00216 /**    
00217  * @} end of MatrixTrans group    
00218  */