OSchip
/
OpenBLAS
Not watched
1
0
Fork 0
Code
Releases 66 Wiki evaluate Activity
Issues 0 Pull Requests 0 Datasets Model Cloudbrain HPC
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.
Branch: develop
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from 'develop'
${ noResults }
OpenBLAS/kernel/power/dgemm_ncopy_logic_4_power8.S

238 lines
4.0 kB
Raw Permalink Blame History 

  1. /***************************************************************************

  2. Copyright (c) 2013-2016, The OpenBLAS Project

  3. All rights reserved.

  4. Redistribution and use in source and binary forms, with or without

  5. modification, are permitted provided that the following conditions are

  6. met:

  7. 1. Redistributions of source code must retain the above copyright

  8. notice, this list of conditions and the following disclaimer.

  9. 2. Redistributions in binary form must reproduce the above copyright

  10. notice, this list of conditions and the following disclaimer in

  11. the documentation and/or other materials provided with the

  12. distribution.

  13. 3. Neither the name of the OpenBLAS project nor the names of

  14. its contributors may be used to endorse or promote products

  15. derived from this software without specific prior written permission.

  16. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"

  17. AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE

  18. IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE

  19. ARE DISCLAIMED. IN NO EVENT SHALL THE OPENBLAS PROJECT OR CONTRIBUTORS BE

  20. LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL

  21. DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR

  22. SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER

  23. CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,

  24. OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE

  25. USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

  26. *****************************************************************************/

  27. 

  28. /**************************************************************************************

  29. * 2016/04/28 Werner Saar (wernsaar@googlemail.com)

  30. * 	 BLASTEST 		: OK

  31. * 	 CTEST			: OK

  32. * 	 TEST			: OK

  33. *	 LAPACK-TEST		: OK

  34. **************************************************************************************/

  35. 

  36. 

  37. 	mr		BO,	B

  38. 	srawi.		I,	N,	2

  39. 	ble		DCOPYN_L2_BEGIN

  40. 

  41. 

  42. DCOPYN_L4_BEGIN:

  43. 

  44. 

  45. DCOPYN_L4_LOOP:

  46. 

  47. 	mr		A0,	A

  48. 	add		A1,	A0,	LDA

  49. 	add		A2,	A1,	LDA

  50. 	add		A3,	A2,	LDA

  51. 	add		A,	A3,	LDA

  52. 

  53. DCOPYN_L4x16_BEGIN:

  54. 

  55. 	srawi.		J,	M,	4

  56. 	ble		DCOPYN_L4x16_END

  57. 

  58. DCOPYN_L4x16_LOOP:

  59. 

  60. 	dcbt		A0,	PREA

  61. 	dcbt		A1,	PREA

  62. 	dcbt		A2,	PREA

  63. 	dcbt		A3,	PREA

  64. 	COPY_4x16

  65. 	addic.		J,	J,	-1

  66. 	bgt		DCOPYN_L4x16_LOOP

  67. 

  68. DCOPYN_L4x16_END:

  69. 

  70. 

  71. DCOPYN_L4x8_BEGIN:

  72. 

  73. 	andi.		J,	M,	8

  74. 	ble		DCOPYN_L4x8_END

  75. 	COPY_4x8

  76. 

  77. DCOPYN_L4x8_END:

  78. 

  79. 

  80. DCOPYN_L4x4_BEGIN:

  81. 

  82. 	andi.		J,	M,	4

  83. 	ble		DCOPYN_L4x4_END

  84. 	COPY_4x4

  85. 

  86. DCOPYN_L4x4_END:

  87. 

  88. 

  89. DCOPYN_L4x2_BEGIN:

  90. 

  91. 	andi.		J,	M,	2

  92. 	ble		DCOPYN_L4x2_END

  93. 	COPY_4x2

  94. 

  95. DCOPYN_L4x2_END:

  96. 

  97. 

  98. DCOPYN_L4x1_BEGIN:

  99. 

  100. 	andi.		J,	M,	1

  101. 	ble		DCOPYN_L4x1_END

  102. 	COPY_4x1

  103. 

  104. DCOPYN_L4x1_END:

  105. 

  106. 

  107. DCOPYN_L4_END:

  108. 

  109. 	addic.		I,	I,	-1

  110. 	bgt		DCOPYN_L4_LOOP

  111. 

  112. DCOPYN_L2_BEGIN:

  113. 

  114. 	andi.		T1,	4,	2

  115. 	ble		DCOPYN_L2_END

  116. 

  117. DCOPYN_L2_LOOP:

  118. 

  119. 	mr		A0,	A

  120. 	add		A1,	A0,	LDA

  121. 	add		A,	A1,	LDA

  122. 

  123. DCOPYN_L2x16_BEGIN:

  124. 

  125. 	srawi.		J,	M,	4

  126. 	ble		DCOPYN_L2x16_END

  127. 

  128. DCOPYN_L2x16_LOOP:

  129. 

  130. 	COPY_2x16

  131. 	addic.		J,	J,	-1

  132. 	bgt		DCOPYN_L2x16_LOOP

  133. 

  134. DCOPYN_L2x16_END:

  135. 

  136. 

  137. DCOPYN_L2x8_BEGIN:

  138. 

  139. 	andi.		J,	M,	8

  140. 	ble		DCOPYN_L2x8_END

  141. 	COPY_2x8

  142. 

  143. DCOPYN_L2x8_END:

  144. 

  145. 

  146. DCOPYN_L2x4_BEGIN:

  147. 

  148. 	andi.		J,	M,	4

  149. 	ble		DCOPYN_L2x4_END

  150. 	COPY_2x4

  151. 

  152. DCOPYN_L2x4_END:

  153. 

  154. 

  155. DCOPYN_L2x2_BEGIN:

  156. 

  157. 	andi.		J,	M,	2

  158. 	ble		DCOPYN_L2x2_END

  159. 	COPY_2x2

  160. 

  161. DCOPYN_L2x2_END:

  162. 

  163. 

  164. DCOPYN_L2x1_BEGIN:

  165. 

  166. 	andi.		J,	M,	1

  167. 	ble		DCOPYN_L2x1_END

  168. 	COPY_2x1

  169. 

  170. DCOPYN_L2x1_END:

  171. 

  172. 

  173. DCOPYN_L2_END:

  174. 

  175. 

  176. DCOPYN_L1_BEGIN:

  177. 

  178. 	andi.		T1,	4,	1

  179. 	ble		DCOPYN_L1_END

  180. 

  181. DCOPYN_L1_LOOP:

  182. 

  183. 	mr		A0,	A

  184. 	add		A,	A0,	LDA

  185. 

  186. DCOPYN_L1x16_BEGIN:

  187. 

  188. 	srawi.		J,	M,	4

  189. 	ble		DCOPYN_L1x16_END

  190. 

  191. DCOPYN_L1x16_LOOP:

  192. 

  193. 	COPY_1x16

  194. 	addic.		J,	J,	-1

  195. 	bgt		DCOPYN_L1x16_LOOP

  196. 

  197. DCOPYN_L1x16_END:

  198. 

  199. 

  200. DCOPYN_L1x8_BEGIN:

  201. 

  202. 	andi.		J,	M,	8

  203. 	ble		DCOPYN_L1x8_END

  204. 	COPY_1x8

  205. 

  206. DCOPYN_L1x8_END:

  207. 

  208. 

  209. DCOPYN_L1x4_BEGIN:

  210. 

  211. 	andi.		J,	M,	4

  212. 	ble		DCOPYN_L1x4_END

  213. 	COPY_1x4

  214. 

  215. DCOPYN_L1x4_END:

  216. 

  217. 

  218. DCOPYN_L1x2_BEGIN:

  219. 

  220. 	andi.		J,	M,	2

  221. 	ble		DCOPYN_L1x2_END

  222. 	COPY_1x2

  223. 

  224. DCOPYN_L1x2_END:

  225. 

  226. 

  227. DCOPYN_L1x1_BEGIN:

  228. 

  229. 	andi.		J,	M,	1

  230. 	ble		DCOPYN_L1x1_END

  231. 	COPY_1x1

  232. 

  233. DCOPYN_L1x1_END:

  234. 

  235. 

  236. DCOPYN_L1_END: