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.
Tree: 855945befb
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '855945befb'
${ noResults }
OpenBLAS/kernel/power/cgemm_tcopy_logic_8_power8.S

248 lines
4.0 kB
Raw 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. * 2016/04/23 Werner Saar (wernsaar@googlemail.com)

  30. * 	 BLASTEST 		: OK

  31. * 	 CTEST			: OK

  32. * 	 TEST			: OK

  33. *	 LAPACK-TEST		: OK

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

  35. 

  36. 

  37. 	srawi.		I,	M,	2

  38. 	ble		CCOPYT_L2_BEGIN

  39. 

  40. 

  41. CCOPYT_L4_BEGIN:

  42. 

  43. 	mr		A0,	A

  44. 	add		A1,	A0,	LDA

  45. 	add		A2,	A1,	LDA

  46. 	add		A3,	A2,	LDA

  47. 	add		A,	A3,	LDA

  48. 	mr		B8,	B

  49. 	addi		B,	B,	64*SIZE

  50. 

  51. 	sradi.		J,	N,	3

  52. 	ble		CCOPYT_L4x4_BEGIN

  53. 

  54. 	mr		BO,	B8

  55. 

  56. CCOPYT_L4x8_LOOP:

  57. 

  58. 	dcbt		A0, PREA

  59. 	dcbt		A1, PREA

  60. 	dcbt		A2, PREA

  61. 	dcbt		A3, PREA

  62. 	dcbtst		BO, M8

  63. 	dcbtst		BO, PREB

  64. 	COPY_4x8

  65. 

  66. 	add		BO,	BO,	M8

  67. 

  68. 	addic.		J,	J,	-1

  69. 	ble		CCOPYT_L4x4_BEGIN

  70. 

  71. 

  72. 	COPY_4x8

  73. 

  74. 	add		BO,	BO,	M8

  75. 

  76. 	addic.		J,	J,	-1

  77. 	bgt		CCOPYT_L4x8_LOOP

  78. 

  79. CCOPYT_L4x4_BEGIN:

  80. 

  81. 	andi.		T1,	N,	4

  82. 	ble		CCOPYT_L4x2_BEGIN

  83. 

  84. 	mr		BO,	B4

  85. 

  86. 	COPY_4x4

  87. 

  88. 

  89. 	addi		B4,	B4,	32*SIZE

  90. 

  91. CCOPYT_L4x2_BEGIN:

  92. 

  93. 	andi.		T1,	N,	2

  94. 	ble		CCOPYT_L4x1_BEGIN

  95. 

  96. 	mr		BO,	B2

  97. 

  98. 	COPY_4x2

  99. 

  100. 

  101. 	addi		B2,	B2,	16*SIZE

  102. 

  103. CCOPYT_L4x1_BEGIN:

  104. 

  105. 	andi.		T1,	N,	1

  106. 	ble		CCOPYT_L4_END

  107. 

  108. 	mr		BO,	B1

  109. 

  110. 	COPY_4x1

  111. 

  112. 

  113. 	addi		B1,	B1,	8*SIZE

  114. 

  115. CCOPYT_L4_END:

  116. 

  117. 	addic.		I,	I,	-1

  118. 	bgt		CCOPYT_L4_BEGIN

  119. 

  120. 

  121. 

  122. CCOPYT_L2_BEGIN:

  123. 

  124. 	andi.		T1,	M,	2

  125. 	ble		CCOPYT_L1_BEGIN

  126. 

  127. 	mr		A0,	A

  128. 	add		A1,	A0,	LDA

  129. 	add		A,	A1,	LDA

  130. 	mr		B8,	B

  131. 	addi		B,	B,	32*SIZE

  132. 

  133. 	sradi.		J,	N,	3

  134. 	ble		CCOPYT_L2x4_BEGIN

  135. 

  136. 	mr		BO,	B8

  137. 

  138. CCOPYT_L2x8_LOOP:

  139. 

  140. 	COPY_2x8

  141. 

  142. 	add		BO,	BO,	M8

  143. 

  144. 	addic.		J,	J,	-1

  145. 	bgt		CCOPYT_L2x8_LOOP

  146. 

  147. CCOPYT_L2x4_BEGIN:

  148. 

  149. 	andi.		T1,	N,	4

  150. 	ble		CCOPYT_L2x2_BEGIN

  151. 

  152. 	mr		BO,	B4

  153. 

  154. 	COPY_2x4

  155. 

  156. 

  157. 	addi		B4,	B4,	16*SIZE

  158. 

  159. CCOPYT_L2x2_BEGIN:

  160. 

  161. 	andi.		T1,	N,	2

  162. 	ble		CCOPYT_L2x1_BEGIN

  163. 

  164. 	mr		BO,	B2

  165. 

  166. 	COPY_2x2

  167. 

  168. 

  169. 	addi		B2,	B2,	8*SIZE

  170. 

  171. CCOPYT_L2x1_BEGIN:

  172. 

  173. 	andi.		T1,	N,	1

  174. 	ble		CCOPYT_L2_END

  175. 

  176. 	mr		BO,	B1

  177. 

  178. 	COPY_2x1

  179. 

  180. 

  181. 	addi		B1,	B1,	4*SIZE

  182. 

  183. CCOPYT_L2_END:

  184. 

  185. 

  186. CCOPYT_L1_BEGIN:

  187. 

  188. 	andi.		T1,	M,	1

  189. 	ble		L999

  190. 

  191. 	mr		A0,	A

  192. 	add		A,	A0,	LDA

  193. 	mr		B8,	B

  194. 	addi		B,	B,	16*SIZE

  195. 

  196. 	sradi.		J,	N,	3

  197. 	ble		CCOPYT_L1x4_BEGIN

  198. 

  199. 	mr		BO,	B8

  200. 

  201. CCOPYT_L1x8_LOOP:

  202. 

  203. 	COPY_1x8

  204. 

  205. 	add		BO,	BO,	M8

  206. 

  207. 	addic.		J,	J,	-1

  208. 	bgt		CCOPYT_L1x8_LOOP

  209. 

  210. CCOPYT_L1x4_BEGIN:

  211. 

  212. 	andi.		T1,	N,	4

  213. 	ble		CCOPYT_L1x2_BEGIN

  214. 

  215. 	mr		BO,	B4

  216. 

  217. 	COPY_1x4

  218. 

  219. 

  220. 	addi		B4,	B4,	8*SIZE

  221. 

  222. CCOPYT_L1x2_BEGIN:

  223. 

  224. 	andi.		T1,	N,	2

  225. 	ble		CCOPYT_L1x1_BEGIN

  226. 

  227. 	mr		BO,	B2

  228. 

  229. 	COPY_1x2

  230. 

  231. 

  232. 	addi		B2,	B2,	4*SIZE

  233. 

  234. CCOPYT_L1x1_BEGIN:

  235. 

  236. 	andi.		T1,	N,	1

  237. 	ble		CCOPYT_L1_END

  238. 

  239. 	mr		BO,	B1

  240. 

  241. 	COPY_1x1

  242. 

  243. 

  244. 	addi		B1,	B1,	2*SIZE

  245. 

  246. CCOPYT_L1_END: