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.

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207
  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 "QSPMV "
  46. #elif defined(DOUBLE)
  47. #define ERROR_NAME "DSPMV "
  48. #else
  49. #define ERROR_NAME "SSPMV "
  50. #endif
  51. static int (*spmv[])(BLASLONG, FLOAT, FLOAT *, FLOAT *, BLASLONG, FLOAT *, BLASLONG, void *) = {
  52. #ifdef XDOUBLE
  53. qspmv_U, qspmv_L,
  54. #elif defined(DOUBLE)
  55. dspmv_U, dspmv_L,
  56. #else
  57. sspmv_U, sspmv_L,
  58. #endif
  59. };
  60. #ifdef SMPTEST
  61. static int (*spmv_thread[])(BLASLONG, FLOAT, FLOAT *, FLOAT *, BLASLONG, FLOAT *, BLASLONG, FLOAT *, int) = {
  62. #ifdef XDOUBLE
  63. qspmv_thread_U, qspmv_thread_L,
  64. #elif defined(DOUBLE)
  65. dspmv_thread_U, dspmv_thread_L,
  66. #else
  67. sspmv_thread_U, sspmv_thread_L,
  68. #endif
  69. };
  70. #endif
  71. #ifndef CBLAS
  72. void NAME(char *UPLO, blasint *N, FLOAT *ALPHA, FLOAT *a,
  73. FLOAT *x, blasint *INCX, FLOAT *BETA, FLOAT *y, blasint *INCY){
  74. char uplo_arg = *UPLO;
  75. blasint n = *N;
  76. FLOAT alpha = *ALPHA;
  77. blasint incx = *INCX;
  78. FLOAT beta = *BETA;
  79. blasint incy = *INCY;
  80. blasint info;
  81. int uplo;
  82. FLOAT *buffer;
  83. #ifdef SMPTEST
  84. int nthreads;
  85. #endif
  86. PRINT_DEBUG_NAME;
  87. TOUPPER(uplo_arg);
  88. uplo = -1;
  89. if (uplo_arg == 'U') uplo = 0;
  90. if (uplo_arg == 'L') uplo = 1;
  91. info = 0;
  92. if (incy == 0) info = 9;
  93. if (incx == 0) info = 6;
  94. if (n < 0) info = 2;
  95. if (uplo < 0) info = 1;
  96. if (info != 0) {
  97. BLASFUNC(xerbla)(ERROR_NAME, &info, sizeof(ERROR_NAME));
  98. return;
  99. }
  100. #else
  101. void CNAME(enum CBLAS_ORDER order,
  102. enum CBLAS_UPLO Uplo,
  103. blasint n,
  104. FLOAT alpha,
  105. FLOAT *a,
  106. FLOAT *x, blasint incx,
  107. FLOAT beta,
  108. FLOAT *y, blasint incy){
  109. FLOAT *buffer;
  110. int uplo;
  111. blasint info;
  112. #ifdef SMPTEST
  113. int nthreads;
  114. #endif
  115. PRINT_DEBUG_CNAME;
  116. uplo = -1;
  117. info = 0;
  118. if (order == CblasColMajor) {
  119. if (Uplo == CblasUpper) uplo = 0;
  120. if (Uplo == CblasLower) uplo = 1;
  121. info = -1;
  122. if (incy == 0) info = 9;
  123. if (incx == 0) info = 6;
  124. if (n < 0) info = 2;
  125. if (uplo < 0) info = 1;
  126. }
  127. if (order == CblasRowMajor) {
  128. if (Uplo == CblasUpper) uplo = 1;
  129. if (Uplo == CblasLower) uplo = 0;
  130. info = -1;
  131. if (incy == 0) info = 9;
  132. if (incx == 0) info = 6;
  133. if (n < 0) info = 2;
  134. if (uplo < 0) info = 1;
  135. }
  136. if (info >= 0) {
  137. BLASFUNC(xerbla)(ERROR_NAME, &info, sizeof(ERROR_NAME));
  138. return;
  139. }
  140. #endif
  141. if (n == 0) return;
  142. if (beta != ONE) SCAL_K(n, 0, 0, beta, y, blasabs(incy), NULL, 0, NULL, 0);
  143. if (alpha == ZERO) return;
  144. IDEBUG_START;
  145. FUNCTION_PROFILE_START();
  146. if (incx < 0 ) x -= (n - 1) * incx;
  147. if (incy < 0 ) y -= (n - 1) * incy;
  148. buffer = (FLOAT *)blas_memory_alloc(1);
  149. #ifdef SMPTEST
  150. nthreads = num_cpu_avail(2);
  151. if (nthreads == 1) {
  152. #endif
  153. (spmv[uplo])(n, alpha, a, x, incx, y, incy, buffer);
  154. #ifdef SMPTEST
  155. } else {
  156. (spmv_thread[uplo])(n, alpha, a, x, incx, y, incy, buffer, nthreads);
  157. }
  158. #endif
  159. blas_memory_free(buffer);
  160. FUNCTION_PROFILE_END(1, n * n / 2 + n, n * n);
  161. IDEBUG_END;
  162. return;
  163. }