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.

zher.c 5.7 kB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202
  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. #ifdef FUNCTION_PROFILE
  42. #include "functable.h"
  43. #endif
  44. #ifdef XDOUBLE
  45. #define ERROR_NAME "XHER "
  46. #elif defined(DOUBLE)
  47. #define ERROR_NAME "ZHER "
  48. #else
  49. #define ERROR_NAME "CHER "
  50. #endif
  51. static int (*her[])(BLASLONG, FLOAT, FLOAT *, BLASLONG, FLOAT *, BLASLONG, FLOAT *) = {
  52. #ifdef XDOUBLE
  53. xher_U, xher_L, xher_V, xher_M,
  54. #elif defined(DOUBLE)
  55. zher_U, zher_L, zher_V, zher_M,
  56. #else
  57. cher_U, cher_L, cher_V, cher_M,
  58. #endif
  59. };
  60. #ifdef SMP
  61. static int (*her_thread[])(BLASLONG, FLOAT, FLOAT *, BLASLONG, FLOAT *, BLASLONG, FLOAT *, int) = {
  62. #ifdef XDOUBLE
  63. xher_thread_U, xher_thread_L, xher_thread_V, xher_thread_M,
  64. #elif defined(DOUBLE)
  65. zher_thread_U, zher_thread_L, zher_thread_V, zher_thread_M,
  66. #else
  67. cher_thread_U, cher_thread_L, cher_thread_V, cher_thread_M,
  68. #endif
  69. };
  70. #endif
  71. #ifndef CBLAS
  72. void NAME(char *UPLO, blasint *N, FLOAT *ALPHA,
  73. FLOAT *x, blasint *INCX, FLOAT *a, blasint *LDA){
  74. char uplo_arg = *UPLO;
  75. blasint n = *N;
  76. FLOAT alpha = *ALPHA;
  77. blasint lda = *LDA;
  78. blasint incx = *INCX;
  79. blasint info;
  80. int uplo;
  81. FLOAT *buffer;
  82. #ifdef SMP
  83. int nthreads;
  84. #endif
  85. PRINT_DEBUG_NAME;
  86. TOUPPER(uplo_arg);
  87. uplo = -1;
  88. if (uplo_arg == 'U') uplo = 0;
  89. if (uplo_arg == 'L') uplo = 1;
  90. info = 0;
  91. if (lda < MAX(1, n)) info = 7;
  92. if (incx == 0) info = 5;
  93. if (n < 0) info = 2;
  94. if (uplo < 0) info = 1;
  95. if (info != 0) {
  96. BLASFUNC(xerbla)(ERROR_NAME, &info, sizeof(ERROR_NAME));
  97. return;
  98. }
  99. #else
  100. void CNAME(enum CBLAS_ORDER order, enum CBLAS_UPLO Uplo, blasint n, FLOAT alpha, void *vx, blasint incx, void *va, blasint lda) {
  101. FLOAT* a = (FLOAT*) va;
  102. FLOAT* x = (FLOAT*) vx;
  103. FLOAT *buffer;
  104. int uplo;
  105. blasint info;
  106. #ifdef SMP
  107. int nthreads;
  108. #endif
  109. PRINT_DEBUG_CNAME;
  110. uplo = -1;
  111. info = 0;
  112. if (order == CblasColMajor) {
  113. if (Uplo == CblasUpper) uplo = 0;
  114. if (Uplo == CblasLower) uplo = 1;
  115. info = -1;
  116. if (lda < MAX(1, n)) info = 7;
  117. if (incx == 0) info = 5;
  118. if (n < 0) info = 2;
  119. if (uplo < 0) info = 1;
  120. }
  121. if (order == CblasRowMajor) {
  122. if (Uplo == CblasUpper) uplo = 3;
  123. if (Uplo == CblasLower) uplo = 2;
  124. info = -1;
  125. if (lda < MAX(1, n)) info = 7;
  126. if (incx == 0) info = 5;
  127. if (n < 0) info = 2;
  128. if (uplo < 0) info = 1;
  129. }
  130. if (info >= 0) {
  131. BLASFUNC(xerbla)(ERROR_NAME, &info, sizeof(ERROR_NAME));
  132. return;
  133. }
  134. #endif
  135. if (n == 0) return;
  136. if (alpha == ZERO) return;
  137. IDEBUG_START;
  138. FUNCTION_PROFILE_START();
  139. if (incx < 0 ) x -= (n - 1) * incx * 2;
  140. buffer = (FLOAT *)blas_memory_alloc(1);
  141. #ifdef SMP
  142. nthreads = num_cpu_avail(2);
  143. if (nthreads == 1) {
  144. #endif
  145. (her[uplo])(n, alpha, x, incx, a, lda, buffer);
  146. #ifdef SMP
  147. } else {
  148. (her_thread[uplo])(n, alpha, x, incx, a, lda, buffer, nthreads);
  149. }
  150. #endif
  151. blas_memory_free(buffer);
  152. FUNCTION_PROFILE_END(4, n * n / 2 + n, n * n);
  153. IDEBUG_END;
  154. return;
  155. }