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.

ztrmm_ltcopy_2.c 6.0 kB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240
  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 "common.h"
  40. int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLONG posY, FLOAT *b){
  41. BLASLONG i, js;
  42. BLASLONG X;
  43. FLOAT data1, data2, data3, data4, data5, data6, data7, data8;
  44. FLOAT *ao1, *ao2;
  45. lda += lda;
  46. js = (n >> 1);
  47. if (js > 0){
  48. do {
  49. X = posX;
  50. if (posX <= posY) {
  51. ao1 = a + posY * 2 + (posX + 0) * lda;
  52. ao2 = a + posY * 2 + (posX + 1) * lda;
  53. } else {
  54. ao1 = a + posX * 2 + (posY + 0) * lda;
  55. ao2 = a + posX * 2 + (posY + 1) * lda;
  56. }
  57. i = (m >> 1);
  58. if (i > 0) {
  59. do {
  60. if (X > posY) {
  61. ao1 += 4;
  62. ao2 += 4;
  63. b += 8;
  64. } else
  65. if (X < posY) {
  66. data1 = *(ao1 + 0);
  67. data2 = *(ao1 + 1);
  68. data3 = *(ao1 + 2);
  69. data4 = *(ao1 + 3);
  70. data5 = *(ao2 + 0);
  71. data6 = *(ao2 + 1);
  72. data7 = *(ao2 + 2);
  73. data8 = *(ao2 + 3);
  74. b[ 0] = data1;
  75. b[ 1] = data2;
  76. b[ 2] = data3;
  77. b[ 3] = data4;
  78. b[ 4] = data5;
  79. b[ 5] = data6;
  80. b[ 6] = data7;
  81. b[ 7] = data8;
  82. ao1 += 2 * lda;
  83. ao2 += 2 * lda;
  84. b += 8;
  85. } else {
  86. #ifdef UNIT
  87. data3 = *(ao1 + 2);
  88. data4 = *(ao1 + 3);
  89. b[ 0] = ONE;
  90. b[ 1] = ZERO;
  91. b[ 2] = data3;
  92. b[ 3] = data4;
  93. b[ 4] = ZERO;
  94. b[ 5] = ZERO;
  95. b[ 6] = ONE;
  96. b[ 7] = ZERO;
  97. #else
  98. data1 = *(ao1 + 0);
  99. data2 = *(ao1 + 1);
  100. data3 = *(ao1 + 2);
  101. data4 = *(ao1 + 3);
  102. data7 = *(ao2 + 2);
  103. data8 = *(ao2 + 3);
  104. b[ 0] = data1;
  105. b[ 1] = data2;
  106. b[ 2] = data3;
  107. b[ 3] = data4;
  108. b[ 4] = ZERO;
  109. b[ 5] = ZERO;
  110. b[ 6] = data7;
  111. b[ 7] = data8;
  112. #endif
  113. ao1 += 4;
  114. ao2 += 4;
  115. b += 8;
  116. }
  117. X += 2;
  118. i --;
  119. } while (i > 0);
  120. }
  121. if (m & 1) {
  122. if (X > posY) {
  123. ao1 += 2;
  124. ao2 += 2;
  125. b += 4;
  126. } else
  127. if (X < posY) {
  128. data1 = *(ao1 + 0);
  129. data2 = *(ao1 + 1);
  130. data3 = *(ao1 + 2);
  131. data4 = *(ao1 + 3);
  132. b[ 0] = data1;
  133. b[ 1] = data2;
  134. b[ 2] = data3;
  135. b[ 3] = data4;
  136. ao1 += lda;
  137. b += 4;
  138. } else {
  139. #ifdef UNIT
  140. data3 = *(ao1 + 2);
  141. data4 = *(ao1 + 3);
  142. b[ 0] = ONE;
  143. b[ 1] = ZERO;
  144. b[ 2] = data3;
  145. b[ 3] = data4;
  146. #else
  147. data1 = *(ao1 + 0);
  148. data2 = *(ao1 + 1);
  149. data3 = *(ao1 + 2);
  150. data4 = *(ao1 + 3);
  151. b[ 0] = data1;
  152. b[ 1] = data2;
  153. b[ 2] = data3;
  154. b[ 3] = data4;
  155. #endif
  156. b += 4;
  157. }
  158. }
  159. posY += 2;
  160. js --;
  161. } while (js > 0);
  162. } /* End of main loop */
  163. if (n & 1){
  164. X = posX;
  165. if (posX <= posY) {
  166. ao1 = a + posY * 2 + (posX + 0) * lda;
  167. } else {
  168. ao1 = a + posX * 2 + (posY + 0) * lda;
  169. }
  170. i = m;
  171. if (i > 0) {
  172. do {
  173. if (X > posY) {
  174. b += 2;
  175. ao1 += 2;
  176. } else
  177. if (X < posY) {
  178. data1 = *(ao1 + 0);
  179. data2 = *(ao1 + 1);
  180. b[ 0] = data1;
  181. b[ 1] = data2;
  182. b += 2;
  183. ao1 += lda;
  184. } else {
  185. #ifdef UNIT
  186. b[ 0] = ONE;
  187. b[ 1] = ZERO;
  188. #else
  189. data1 = *(ao1 + 0);
  190. data2 = *(ao1 + 1);
  191. b[ 0] = data1;
  192. b[ 1] = data2;
  193. #endif
  194. b += 2;
  195. ao1 += 2;
  196. }
  197. X ++;
  198. i --;
  199. } while (i > 0);
  200. }
  201. posY += 1;
  202. }
  203. return 0;
  204. }