Subversion Repositories DashDisplay

Rev

Rev 28 | Blame | Compare with Previous | Last modification | View Log | Download | RSS feed

  1. /* ----------------------------------------------------------------------    
  2. * Copyright (C) 2010-2014 ARM Limited. All rights reserved.    
  3. *    
  4. * $Date:        19. March 2015
  5. * $Revision:    V.1.4.5
  6. *    
  7. * Project:          CMSIS DSP Library    
  8. * Title:            arm_mat_trans_q15.c    
  9. *    
  10. * Description:  Q15 matrix transpose.    
  11. *    
  12. * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
  13. *  
  14. * Redistribution and use in source and binary forms, with or without
  15. * modification, are permitted provided that the following conditions
  16. * are met:
  17. *   - Redistributions of source code must retain the above copyright
  18. *     notice, this list of conditions and the following disclaimer.
  19. *   - Redistributions in binary form must reproduce the above copyright
  20. *     notice, this list of conditions and the following disclaimer in
  21. *     the documentation and/or other materials provided with the
  22. *     distribution.
  23. *   - Neither the name of ARM LIMITED nor the names of its contributors
  24. *     may be used to endorse or promote products derived from this
  25. *     software without specific prior written permission.
  26. *
  27. * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
  28. * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
  29. * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
  30. * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
  31. * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
  32. * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
  33. * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
  34. * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
  35. * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  36. * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
  37. * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
  38. * POSSIBILITY OF SUCH DAMAGE.  
  39. * -------------------------------------------------------------------- */
  40.  
  41. #include "arm_math.h"
  42.  
  43. /**    
  44.  * @ingroup groupMatrix    
  45.  */
  46.  
  47. /**    
  48.  * @addtogroup MatrixTrans    
  49.  * @{    
  50.  */
  51.  
  52. /*    
  53.  * @brief Q15 matrix transpose.    
  54.  * @param[in]  *pSrc points to the input matrix    
  55.  * @param[out] *pDst points to the output matrix    
  56.  * @return      The function returns either  <code>ARM_MATH_SIZE_MISMATCH</code>    
  57.  * or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.    
  58.  */
  59.  
  60. arm_status arm_mat_trans_q15(
  61.   const arm_matrix_instance_q15 * pSrc,
  62.   arm_matrix_instance_q15 * pDst)
  63. {
  64.   q15_t *pSrcA = pSrc->pData;                    /* input data matrix pointer */
  65.   q15_t *pOut = pDst->pData;                     /* output data matrix pointer */
  66.   uint16_t nRows = pSrc->numRows;                /* number of nRows */
  67.   uint16_t nColumns = pSrc->numCols;             /* number of nColumns */
  68.   uint16_t col, row = nRows, i = 0u;             /* row and column loop counters */
  69.   arm_status status;                             /* status of matrix transpose */
  70.  
  71. #ifndef ARM_MATH_CM0_FAMILY
  72.  
  73.   /* Run the below code for Cortex-M4 and Cortex-M3 */
  74. #ifndef UNALIGNED_SUPPORT_DISABLE
  75.  
  76.   q31_t in;                                      /* variable to hold temporary output  */
  77.  
  78. #else
  79.  
  80.   q15_t in;
  81.  
  82. #endif  /*      #ifndef UNALIGNED_SUPPORT_DISABLE       */
  83.  
  84. #ifdef ARM_MATH_MATRIX_CHECK
  85.  
  86.  
  87.   /* Check for matrix mismatch condition */
  88.   if((pSrc->numRows != pDst->numCols) || (pSrc->numCols != pDst->numRows))
  89.   {
  90.     /* Set status as ARM_MATH_SIZE_MISMATCH */
  91.     status = ARM_MATH_SIZE_MISMATCH;
  92.   }
  93.   else
  94. #endif /*      #ifdef ARM_MATH_MATRIX_CHECK    */
  95.  
  96.   {
  97.     /* Matrix transpose by exchanging the rows with columns */
  98.     /* row loop     */
  99.     do
  100.     {
  101.  
  102.       /* Apply loop unrolling and exchange the columns with row elements */
  103.       col = nColumns >> 2u;
  104.  
  105.       /* The pointer pOut is set to starting address of the column being processed */
  106.       pOut = pDst->pData + i;
  107.  
  108.       /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.    
  109.        ** a second loop below computes the remaining 1 to 3 samples. */
  110.       while(col > 0u)
  111.       {
  112. #ifndef UNALIGNED_SUPPORT_DISABLE
  113.  
  114.         /* Read two elements from the row */
  115.         in = *__SIMD32(pSrcA)++;
  116.  
  117.         /* Unpack and store one element in the destination */
  118. #ifndef ARM_MATH_BIG_ENDIAN
  119.  
  120.         *pOut = (q15_t) in;
  121.  
  122. #else
  123.  
  124.         *pOut = (q15_t) ((in & (q31_t) 0xffff0000) >> 16);
  125.  
  126. #endif /*    #ifndef ARM_MATH_BIG_ENDIAN    */
  127.  
  128.         /* Update the pointer pOut to point to the next row of the transposed matrix */
  129.         pOut += nRows;
  130.  
  131.         /* Unpack and store the second element in the destination */
  132.  
  133. #ifndef ARM_MATH_BIG_ENDIAN
  134.  
  135.         *pOut = (q15_t) ((in & (q31_t) 0xffff0000) >> 16);
  136.  
  137. #else
  138.  
  139.         *pOut = (q15_t) in;
  140.  
  141. #endif /*    #ifndef ARM_MATH_BIG_ENDIAN    */
  142.  
  143.         /* Update the pointer pOut to point to the next row of the transposed matrix */
  144.         pOut += nRows;
  145.  
  146.         /* Read two elements from the row */
  147. #ifndef ARM_MATH_BIG_ENDIAN
  148.  
  149.         in = *__SIMD32(pSrcA)++;
  150.  
  151. #else
  152.  
  153.         in = *__SIMD32(pSrcA)++;
  154.  
  155. #endif /*    #ifndef ARM_MATH_BIG_ENDIAN    */
  156.  
  157.         /* Unpack and store one element in the destination */
  158. #ifndef ARM_MATH_BIG_ENDIAN
  159.  
  160.         *pOut = (q15_t) in;
  161.  
  162. #else
  163.  
  164.         *pOut = (q15_t) ((in & (q31_t) 0xffff0000) >> 16);
  165.  
  166. #endif /*    #ifndef ARM_MATH_BIG_ENDIAN    */
  167.  
  168.         /* Update the pointer pOut to point to the next row of the transposed matrix */
  169.         pOut += nRows;
  170.  
  171.         /* Unpack and store the second element in the destination */
  172. #ifndef ARM_MATH_BIG_ENDIAN
  173.  
  174.         *pOut = (q15_t) ((in & (q31_t) 0xffff0000) >> 16);
  175.  
  176. #else
  177.  
  178.         *pOut = (q15_t) in;
  179.  
  180. #endif /*    #ifndef ARM_MATH_BIG_ENDIAN    */
  181.  
  182. #else    
  183.         /* Read one element from the row */
  184.         in = *pSrcA++;
  185.  
  186.         /* Store one element in the destination */
  187.         *pOut = in;
  188.  
  189.         /* Update the pointer px to point to the next row of the transposed matrix */
  190.         pOut += nRows;
  191.  
  192.         /* Read one element from the row */
  193.         in = *pSrcA++;
  194.  
  195.         /* Store one element in the destination */
  196.         *pOut = in;
  197.  
  198.         /* Update the pointer px to point to the next row of the transposed matrix */
  199.         pOut += nRows;
  200.  
  201.         /* Read one element from the row */
  202.         in = *pSrcA++;
  203.  
  204.         /* Store one element in the destination */
  205.         *pOut = in;
  206.  
  207.         /* Update the pointer px to point to the next row of the transposed matrix */
  208.         pOut += nRows;
  209.  
  210.         /* Read one element from the row */
  211.         in = *pSrcA++;
  212.  
  213.         /* Store one element in the destination */
  214.         *pOut = in;
  215.  
  216. #endif  /*      #ifndef UNALIGNED_SUPPORT_DISABLE       */
  217.  
  218.         /* Update the pointer pOut to point to the next row of the transposed matrix */
  219.         pOut += nRows;
  220.  
  221.         /* Decrement the column loop counter */
  222.         col--;
  223.       }
  224.  
  225.       /* Perform matrix transpose for last 3 samples here. */
  226.       col = nColumns % 0x4u;
  227.  
  228. #else
  229.  
  230.   /* Run the below code for Cortex-M0 */
  231.  
  232. #ifdef ARM_MATH_MATRIX_CHECK
  233.  
  234.   /* Check for matrix mismatch condition */
  235.   if((pSrc->numRows != pDst->numCols) || (pSrc->numCols != pDst->numRows))
  236.   {
  237.     /* Set status as ARM_MATH_SIZE_MISMATCH */
  238.     status = ARM_MATH_SIZE_MISMATCH;
  239.   }
  240.   else
  241. #endif /*    #ifdef ARM_MATH_MATRIX_CHECK    */
  242.  
  243.   {
  244.     /* Matrix transpose by exchanging the rows with columns */
  245.     /* row loop     */
  246.     do
  247.     {
  248.       /* The pointer pOut is set to starting address of the column being processed */
  249.       pOut = pDst->pData + i;
  250.  
  251.       /* Initialize column loop counter */
  252.       col = nColumns;
  253.  
  254. #endif /* #ifndef ARM_MATH_CM0_FAMILY */
  255.  
  256.       while(col > 0u)
  257.       {
  258.         /* Read and store the input element in the destination */
  259.         *pOut = *pSrcA++;
  260.  
  261.         /* Update the pointer pOut to point to the next row of the transposed matrix */
  262.         pOut += nRows;
  263.  
  264.         /* Decrement the column loop counter */
  265.         col--;
  266.       }
  267.  
  268.       i++;
  269.  
  270.       /* Decrement the row loop counter */
  271.       row--;
  272.  
  273.     } while(row > 0u);
  274.  
  275.     /* set status as ARM_MATH_SUCCESS */
  276.     status = ARM_MATH_SUCCESS;
  277.   }
  278.   /* Return to application */
  279.   return (status);
  280. }
  281.  
  282. /**    
  283.  * @} end of MatrixTrans group    
  284.  */
  285.