You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.

zgbmv_k.c 5.0 kB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145
  1. /*********************************************************************/
  2. /* Copyright 2009, 2010 The University of Texas at Austin. */
  3. /* All rights reserved. */
  4. /* */
  5. /* Redistribution and use in source and binary forms, with or */
  6. /* without modification, are permitted provided that the following */
  7. /* conditions are met: */
  8. /* */
  9. /* 1. Redistributions of source code must retain the above */
  10. /* copyright notice, this list of conditions and the following */
  11. /* disclaimer. */
  12. /* */
  13. /* 2. Redistributions in binary form must reproduce the above */
  14. /* copyright notice, this list of conditions and the following */
  15. /* disclaimer in the documentation and/or other materials */
  16. /* provided with the distribution. */
  17. /* */
  18. /* THIS SOFTWARE IS PROVIDED BY THE UNIVERSITY OF TEXAS AT */
  19. /* AUSTIN ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, */
  20. /* INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF */
  21. /* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE */
  22. /* DISCLAIMED. IN NO EVENT SHALL THE UNIVERSITY OF TEXAS AT */
  23. /* AUSTIN OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, */
  24. /* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES */
  25. /* (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE */
  26. /* GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR */
  27. /* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF */
  28. /* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT */
  29. /* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT */
  30. /* OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE */
  31. /* POSSIBILITY OF SUCH DAMAGE. */
  32. /* */
  33. /* The views and conclusions contained in the software and */
  34. /* documentation are those of the authors and should not be */
  35. /* interpreted as representing official policies, either expressed */
  36. /* or implied, of The University of Texas at Austin. */
  37. /*********************************************************************/
  38. #include <stdio.h>
  39. #include <ctype.h>
  40. #include "common.h"
  41. #ifndef XCONJ
  42. #ifndef CONJ
  43. #define ZAXPY AXPYU_K
  44. #define ZDOT DOTU_K
  45. #else
  46. #define ZAXPY AXPYC_K
  47. #define ZDOT DOTC_K
  48. #endif
  49. #else
  50. #ifndef CONJ
  51. #define ZAXPY AXPYU_K
  52. #define ZDOT DOTC_K
  53. #else
  54. #define ZAXPY AXPYC_K
  55. #define ZDOT DOTU_K
  56. #endif
  57. #endif
  58. #ifndef TRANS
  59. #define M m
  60. #define N n
  61. #else
  62. #define N m
  63. #define M n
  64. #endif
  65. void CNAME(BLASLONG m, BLASLONG n, BLASLONG ku, BLASLONG kl, FLOAT alpha_r, FLOAT alpha_i,
  66. FLOAT *a, BLASLONG lda,
  67. FLOAT *x, BLASLONG incx, FLOAT *y, BLASLONG incy, void *buffer){
  68. BLASLONG i, offset_u, offset_l, start, end, length;
  69. FLOAT *X = x;
  70. FLOAT *Y = y;
  71. FLOAT *gemvbuffer = (FLOAT *)buffer;
  72. FLOAT *bufferY = gemvbuffer;
  73. FLOAT *bufferX = gemvbuffer;
  74. #ifdef TRANS
  75. OPENBLAS_COMPLEX_FLOAT temp;
  76. #endif
  77. if (incy != 1) {
  78. Y = bufferY;
  79. bufferX = (FLOAT *)(((BLASLONG)bufferY + M * sizeof(FLOAT) * 2 + 4095) & ~4095);
  80. // gemvbuffer = bufferX;
  81. COPY_K(M, y, incy, Y, 1);
  82. }
  83. if (incx != 1) {
  84. X = bufferX;
  85. // gemvbuffer = (FLOAT *)(((BLASLONG)bufferX + N * sizeof(FLOAT) * 2 + 4095) & ~4095);
  86. COPY_K(N, x, incx, X, 1);
  87. }
  88. offset_u = ku;
  89. offset_l = ku + m;
  90. for (i = 0; i < MIN(n, m + ku); i++) {
  91. start = MAX(offset_u, 0);
  92. end = MIN(offset_l, ku + kl + 1);
  93. length = end - start;
  94. #ifndef TRANS
  95. ZAXPY(length, 0, 0,
  96. #ifndef XCONJ
  97. alpha_r * X[i * 2 + 0] - alpha_i * X[i * 2 + 1],
  98. alpha_i * X[i * 2 + 0] + alpha_r * X[i * 2 + 1],
  99. #else
  100. alpha_r * X[i * 2 + 0] + alpha_i * X[i * 2 + 1],
  101. alpha_i * X[i * 2 + 0] - alpha_r * X[i * 2 + 1],
  102. #endif
  103. a + start * 2, 1, Y + (start - offset_u) * 2, 1, NULL, 0);
  104. #else
  105. #ifndef XCONJ
  106. temp = ZDOT(length, a + start * 2, 1, X + (start - offset_u) * 2, 1);
  107. #else
  108. temp = ZDOT(length, X + (start - offset_u) * 2, 1, a + start * 2, 1);
  109. #endif
  110. #if !defined(XCONJ) || !defined(CONJ)
  111. Y[i * 2 + 0] += alpha_r * CREAL(temp) - alpha_i * CIMAG(temp);
  112. Y[i * 2 + 1] += alpha_i * CREAL(temp) + alpha_r * CIMAG(temp);
  113. #else
  114. Y[i * 2 + 0] += alpha_r * CREAL(temp) + alpha_i * CIMAG(temp);
  115. Y[i * 2 + 1] += alpha_i * CREAL(temp) - alpha_r * CIMAG(temp);
  116. #endif
  117. #endif
  118. offset_u --;
  119. offset_l --;
  120. a += lda * 2;
  121. }
  122. if (incy != 1) {
  123. COPY_K(M, Y, 1, y, incy);
  124. }
  125. return;
  126. }