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.

zhpmv_k.c 6.3 kB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179
  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. int CNAME(BLASLONG m, FLOAT alpha_r, FLOAT alpha_i,
  42. FLOAT *a, FLOAT *x, BLASLONG incx, FLOAT *y, BLASLONG incy, void *buffer){
  43. BLASLONG i;
  44. FLOAT *X = x;
  45. FLOAT *Y = y;
  46. FLOAT *gemvbuffer = (FLOAT *)buffer;
  47. FLOAT *bufferY = gemvbuffer;
  48. FLOAT *bufferX = gemvbuffer;
  49. FLOAT temp[2];
  50. OPENBLAS_COMPLEX_FLOAT result;
  51. if (incy != 1) {
  52. Y = bufferY;
  53. bufferX = (FLOAT *)(((BLASLONG)bufferY + m * sizeof(FLOAT) * 2 + 4095) & ~4095);
  54. // gemvbuffer = bufferX;
  55. COPY_K(m, y, incy, Y, 1);
  56. }
  57. if (incx != 1) {
  58. X = bufferX;
  59. // gemvbuffer = (FLOAT *)(((BLASLONG)bufferX + m * sizeof(FLOAT) * 2 + 4095) & ~4095);
  60. COPY_K(m, x, incx, X, 1);
  61. }
  62. for (i = 0; i < m; i++) {
  63. #ifndef HEMVREV
  64. #ifndef LOWER
  65. if (i > 0) {
  66. result = DOTC_K(i, a, 1, X, 1);
  67. Y[i * 2 + 0] += alpha_r * CREAL(result) - alpha_i * CIMAG(result);
  68. Y[i * 2 + 1] += alpha_r * CIMAG(result) + alpha_i * CREAL(result);
  69. }
  70. temp[0] = a[i * 2 + 0] * X[i * 2 + 0];
  71. temp[1] = a[i * 2 + 0] * X[i * 2 + 1];
  72. Y[i * 2 + 0] += alpha_r * temp[0] - alpha_i * temp[1];
  73. Y[i * 2 + 1] += alpha_r * temp[1] + alpha_i * temp[0];
  74. if (i > 0) {
  75. AXPYU_K(i, 0, 0,
  76. alpha_r * X[i * 2 + 0] - alpha_i * X[i * 2 + 1],
  77. alpha_r * X[i * 2 + 1] + alpha_i * X[i * 2 + 0],
  78. a, 1, Y, 1, NULL, 0);
  79. }
  80. a += (i + 1) * 2;
  81. #else
  82. if (m - i > 1) {
  83. result = DOTC_K(m - i - 1, a + (i + 1) * 2, 1, X + (i + 1) * 2, 1);
  84. Y[i * 2 + 0] += alpha_r * CREAL(result) - alpha_i * CIMAG(result);
  85. Y[i * 2 + 1] += alpha_r * CIMAG(result) + alpha_i * CREAL(result);
  86. }
  87. temp[0] = a[i * 2 + 0] * X[i * 2 + 0];
  88. temp[1] = a[i * 2 + 0] * X[i * 2 + 1];
  89. Y[i * 2 + 0] += alpha_r * temp[0] - alpha_i * temp[1];
  90. Y[i * 2 + 1] += alpha_r * temp[1] + alpha_i * temp[0];
  91. if (m - i > 1) {
  92. AXPYU_K(m - i - 1, 0, 0,
  93. alpha_r * X[i * 2 + 0] - alpha_i * X[i * 2 + 1],
  94. alpha_r * X[i * 2 + 1] + alpha_i * X[i * 2 + 0],
  95. a + (i + 1) * 2, 1, Y + (i + 1) * 2, 1, NULL, 0);
  96. }
  97. a += (m - i - 1) * 2;
  98. #endif
  99. #else
  100. #ifndef LOWER
  101. if (i > 0) {
  102. result = DOTU_K(i, a, 1, X, 1);
  103. Y[i * 2 + 0] += alpha_r * CREAL(result) - alpha_i * CIMAG(result);
  104. Y[i * 2 + 1] += alpha_r * CIMAG(result) + alpha_i * CREAL(result);
  105. }
  106. temp[0] = a[i * 2 + 0] * X[i * 2 + 0];
  107. temp[1] = a[i * 2 + 0] * X[i * 2 + 1];
  108. Y[i * 2 + 0] += alpha_r * temp[0] - alpha_i * temp[1];
  109. Y[i * 2 + 1] += alpha_r * temp[1] + alpha_i * temp[0];
  110. if (i > 0) {
  111. AXPYC_K(i, 0, 0,
  112. alpha_r * X[i * 2 + 0] - alpha_i * X[i * 2 + 1],
  113. alpha_r * X[i * 2 + 1] + alpha_i * X[i * 2 + 0],
  114. a, 1, Y, 1, NULL, 0);
  115. }
  116. a += (i + 1) * 2;
  117. #else
  118. if (m - i > 1) {
  119. result = DOTU_K(m - i - 1, a + (i + 1) * 2, 1, X + (i + 1) * 2, 1);
  120. Y[i * 2 + 0] += alpha_r * CREAL(result) - alpha_i * CIMAG(result);
  121. Y[i * 2 + 1] += alpha_r * CIMAG(result) + alpha_i * CREAL(result);
  122. }
  123. temp[0] = a[i * 2 + 0] * X[i * 2 + 0];
  124. temp[1] = a[i * 2 + 0] * X[i * 2 + 1];
  125. Y[i * 2 + 0] += alpha_r * temp[0] - alpha_i * temp[1];
  126. Y[i * 2 + 1] += alpha_r * temp[1] + alpha_i * temp[0];
  127. if (m - i > 1) {
  128. AXPYC_K(m - i - 1, 0, 0,
  129. alpha_r * X[i * 2 + 0] - alpha_i * X[i * 2 + 1],
  130. alpha_r * X[i * 2 + 1] + alpha_i * X[i * 2 + 0],
  131. a + (i + 1) * 2, 1, Y + (i + 1) * 2, 1, NULL, 0);
  132. }
  133. a += (m - i - 1) * 2;
  134. #endif
  135. #endif
  136. }
  137. if (incy != 1) {
  138. COPY_K(m, Y, 1, y, incy);
  139. }
  140. return 0;
  141. }