开通博客赚积分
发布资源赚积分
分类

源码开发语言/平台
当前位置：首页 > 源码/资料 > 嵌入式/单片机编程 > Windows CE > 查看源码
mcomp_mips64.c
资源名称：tcpmp.rar　[点击查看]
上传用户：wstnjxml
上传日期：2014-04-03
资源大小：7248k
文件大小：12k
源码类别：

Windows CE

开发平台：

C/C++

mcomp_mips64.c：源码内容
  1. /*****************************************************************************
  2.  *
  3.  * This program is free software ; you can redistribute it and/or modify
  4.  * it under the terms of the GNU General Public License as published by
  5.  * the Free Software Foundation; either version 2 of the License, or
  6.  * (at your option) any later version.
  7.  *
  8.  * This program is distributed in the hope that it will be useful,
  9.  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10.  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  11.  * GNU General Public License for more details.
  12.  *
  13.  * You should have received a copy of the GNU General Public License
  14.  * along with this program; if not, write to the Free Software
  15.  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
  16.  *
  17.  * $Id: mcomp_mips64.c 284 2005-10-04 08:54:26Z picard $
  18.  *
  19.  * The Core Pocket Media Player
  20.  * Copyright (c) 2004-2005 Gabor Kovacs
  21.  *
  22.  ****************************************************************************/
  24. #include "../common.h"
  25. #include "softidct.h"
  27. #if defined(MIPS64)
  29. // important: disable interrupts before using 64bit registers (but not too long, it could freeze)
  30. // important: $8 can't be used as 64bit register (trashed by some kernel routine)
  32. // $8 src end pointer
  33. // $4 src pointer
  34. // $5 dst pointer
  35. // $6 src stride
  36. // $7 dst stride
  37. // $2,$9 first item lower 8 bytes (in two forms)
  38. // $10,$11 first item upper 8 bytes (in two forms) - for 16x16 macroblocks
  39. // $12,$13 second item lower 8 bytes (in two forms)
  40. // $14,$15 second item upper 8 bytes (in two forms) - for 16x16 macroblocks
  41. // $24 0x0101 0101 0101 0101 - for non horver
  42. // $25 0xFEFE FEFE FEFE FEFE - for non horver
  43. // $24 rounding   - for horver
  44. // $25 temporary   - for 16x16 horver
  45. // $3 0x0303 0303 0303 0303 - for horver
  46. // $1 0xFCFC FCFC FCFC FCFC - for horver
  47. // $16 temporary   - for 16x16 horver (must be saved/restored)
  49. #define SET_SRCEND8
  50. "sll $8,$6,3;"
  51. "addu $8,$4,$8;"
  53. #define SET_SRCEND16
  54. "sll $8,$6,4;"
  55. "addu $8,$4,$8;"
  57. #define SET_MASKS
  58.     "li $24,0x01010101;"
  59. "dsll $25,$24,32;"
  60. "or $24,$24,$25;"
  61. "nor $25,$24,$0;"
  63. #define SET_MASKS2
  64. ".set noat;"
  65.     "li $3,0x03030303;"
  66. "dsll $1,$3,32;"
  67. "or $3,$3,$1;"
  68. "nor $1,$3,$0;"
  70. #define LOAD_FIRST8(ofs)
  71. "uld $2, " #ofs "($4);"
  72. "and $9,$2,$25;"
  73. "dsrl $9,$9,1;"
  75. #define LOAD_FIRST16(ofs)
  76. "uld $2, " #ofs "($4);"
  77. "uld $10," #ofs "+8($4);"
  78. "and $9,$2,$25;"
  79. "and $11,$10,$25;"
  80. "dsrl $9,$9,1;"
  81. "dsrl $11,$11,1;"
  83. #define LOAD_SECOND8(ofs)
  84. "uld $12," #ofs "($4);"
  85. "and $13,$12,$25;"
  86. "dsrl $13,$13,1;"
  88. #define LOAD_SECOND16(ofs)
  89. "uld $12," #ofs "($4);"
  90. "uld $14," #ofs "+8($4);"
  91. "and $13,$12,$25;"
  92. "and $15,$14,$25;"
  93. "dsrl $13,$13,1;"
  94. "dsrl $15,$15,1;"
  96. #define LOAD_FIRST8_HV
  97. "uld $2,0($4);"
  98. "uld $9,1($4);"
  99. "and $10,$2,$1;"
  100. "and $11,$9,$1;"
  101. "and $2,$2,$3;"
  102. "and $9,$9,$3;"
  103. "dsrl $10,$10,2;"
  104. "dsrl $11,$11,2;"
  105. "daddu $2,$2,$9;"
  106. "daddu $9,$10,$11;"
  108. #define LOAD_FIRST16_HV
  109. "uld $2,0($4);"
  110. "uld $9,1($4);"
  111. "and $16,$2,$1;"
  112. "and $25,$9,$1;"
  113. "and $2,$2,$3;"
  114. "and $9,$9,$3;"
  115. "dsrl $16,$16,2;"
  116. "dsrl $25,$25,2;"
  117. "daddu $2,$2,$9;"
  118. "daddu $9,$16,$25;"
  120. "uld $10,8($4);"
  121. "uld $11,9($4);"
  122. "and $16,$10,$1;"
  123. "and $25,$11,$1;"
  124. "and $10,$10,$3;"
  125. "and $11,$11,$3;"
  126. "dsrl $16,$16,2;"
  127. "dsrl $25,$25,2;"
  128. "daddu $10,$10,$11;"
  129. "daddu $11,$16,$25;"
  131. #define LOAD_SECOND8_HV
  132. "uld $12,0($4);"
  133. "uld $13,1($4);"
  134. "and $14,$12,$1;"
  135. "and $15,$13,$1;"
  136. "and $12,$12,$3;"
  137. "and $13,$13,$3;"
  138. "dsrl $14,$14,2;"
  139. "dsrl $15,$15,2;"
  140. "daddu $12,$12,$13;"
  141. "daddu $13,$14,$15;"
  143. #define LOAD_SECOND16_HV
  144. "uld $12,0($4);"
  145. "uld $13,1($4);"
  146. "and $16,$12,$1;"
  147. "and $25,$13,$1;"
  148. "and $12,$12,$3;"
  149. "and $13,$13,$3;"
  150. "dsrl $16,$16,2;"
  151. "dsrl $25,$25,2;"
  152. "daddu $12,$12,$13;"
  153. "daddu $13,$16,$25;"
  155. "uld $14,8($4);"
  156. "uld $15,9($4);"
  157. "and $16,$14,$1;"
  158. "and $25,$15,$1;"
  159. "and $14,$14,$3;"
  160. "and $15,$15,$3;"
  161. "dsrl $16,$16,2;"
  162. "dsrl $25,$25,2;"
  163. "daddu $14,$14,$15;"
  164. "daddu $15,$16,$25;"
  166. #define AVG8
  167. "or $2,$2,$12;"
  168. "and $2,$2,$24;"
  169. "daddu $2,$2,$9;"
  170. "daddu $2,$2,$13;"
  172. #define AVG16
  173. "or $2,$2,$12;"
  174. "or $10,$10,$14;"
  175. "and $2,$2,$24;"
  176. "and $10,$10,$24;"
  177. "daddu $2,$2,$9;"
  178. "daddu $10,$10,$11;"
  179. "daddu $2,$2,$13;"
  180. "daddu $10,$10,$15;"
  182. #define AVGROUND8
  183. "and $2,$2,$12;"
  184. "and $2,$2,$24;"
  185. "daddu $2,$2,$9;"
  186. "daddu $2,$2,$13;"
  188. #define AVGROUND16
  189. "and $2,$2,$12;"
  190. "and $10,$10,$14;"
  191. "and $2,$2,$24;"
  192. "and $10,$10,$24;"
  193. "daddu $2,$2,$9;"
  194. "daddu $10,$10,$11;"
  195. "daddu $2,$2,$13;"
  196. "daddu $10,$10,$15;"
  198. #define SWAPSET8
  199. "move $2,$12;"
  200. "move $9,$13;"
  202. #define SWAPSET16
  203. "move $2,$12;"
  204. "move $9,$13;"
  205. "move $10,$14;"
  206. "move $11,$15;"
  208. #define WRITE8
  209. "sdr $2,0($5);"
  210. "addu $5,$5,$7;" 
  212. #define WRITE16
  213. "sdr $2,0($5);"
  214. "sdr $10,8($5);"
  215. "addu $5,$5,$7;" 
  217. #define SAVE
  218. "addiu $sp,$sp,-4;"
  219. "sw $16,0(sp);"
  221. #define RESTORE
  222. "lw $16,0(sp);"
  223. "addiu $sp,$sp,4;"
  225. #ifdef MIPSVR41XX
  226. //cache without loading
  227. #define CACHE16
  228. ".set noreorder;"
  229. "cache 13,0($5);"
  230. ".set reorder;"
  231. #else
  232. #define CACHE16
  233. #endif
  235. void CopyBlock(unsigned char * Src, unsigned char * Dst, int SrcStride, int DstStride) 
  236. { 
  237. __asm( SET_SRCEND8
  239. "loop:"
  240. "uld $2,0($4);" 
  241. "addu $4,$4,$6;"
  242. "sdr $2,0($5);"
  243. "addu $5,$5,$7;" 
  244. "bne $4,$8,loop;"); 
  245. } 
  247. void CopyBlockHor(unsigned char * Src, unsigned char * Dst, int SrcStride, int DstStride) 
  248. { 
  249. __asm ( SET_SRCEND8
  250. SET_MASKS
  252. "loophor:"
  253. LOAD_FIRST8(0)
  254. LOAD_SECOND8(1)
  255. "addu $4,$4,$6;" 
  257. AVG8
  258. WRITE8
  260. "bne $4,$8,loophor;");
  261. } 
  263. void CopyBlockHorRound(unsigned char * Src, unsigned char * Dst, int SrcStride, int DstStride) 
  264. { 
  265. __asm ( SET_SRCEND8
  266. SET_MASKS
  268. "loophorround:"
  269. LOAD_FIRST8(0)
  270. LOAD_SECOND8(1)
  272. "addu $4,$4,$6;" 
  274. AVGROUND8
  275. WRITE8
  277. "bne $4,$8,loophorround;");
  278. } 
  280. void CopyBlockVer(unsigned char * Src, unsigned char * Dst, int SrcStride, int DstStride) 
  281. { 
  282. __asm ( SET_SRCEND8
  283. SET_MASKS
  285. LOAD_FIRST8(0)
  287. "loopver:"
  288. "addu $4,$4,$6;" 
  290. LOAD_SECOND8(0)
  292. AVG8
  293. WRITE8
  294. SWAPSET8
  296. "bne $4,$8,loopver;");
  297. } 
  299. void CopyBlockVerRound(unsigned char * Src, unsigned char * Dst, int SrcStride, int DstStride) 
  300. { 
  301. __asm ( SET_SRCEND8
  302. SET_MASKS
  304. LOAD_FIRST8(0)
  306. "loopverround:"
  307. "addu $4,$4,$6;" 
  309. LOAD_SECOND8(0)
  311. AVGROUND8
  312. WRITE8
  313. SWAPSET8
  315. "bne $4,$8,loopverround;");
  316. }
  318. void CopyBlockHorVer(unsigned char * Src, unsigned char * Dst, int SrcStride, int DstStride) 
  319. { 
  320. __asm ( SET_SRCEND8
  321. SET_MASKS2
  323. "dsll $24,$3,1;"  
  324. "and $24,$24,$3;" // 0x0202 0202 0202 0202
  326. //preprocessing
  328. LOAD_FIRST8_HV
  330. "loophorver:"
  331. "addu $4,$4,$6;"
  333. LOAD_SECOND8_HV
  335. "daddu $2,$2,$12;"
  336. "daddu $9,$9,$13;"
  338. "daddu $2,$2,$24;"
  340. "and $2,$2,$1;"
  341. "dsrl $2,$2,2;"
  342. "daddu $2,$2,$9;"
  344. WRITE8
  345. SWAPSET8
  347. "bne $4,$8,loophorver;");
  348. } 
  350. void CopyBlockHorVerRound(unsigned char * Src, unsigned char * Dst, int SrcStride, int DstStride) 
  351. { 
  352. __asm ( SET_SRCEND8
  353. SET_MASKS2
  355. "dsrl $24,$3,1;"  
  356. "and $24,$24,$3;" // 0x0101 0101 0101 0101
  358. //preprocessing
  360. LOAD_FIRST8_HV
  362. "loophorverround:"
  363. "addu $4,$4,$6;"
  365. LOAD_SECOND8_HV
  367. "daddu $2,$2,$12;"
  368. "daddu $9,$9,$13;"
  370. "daddu $2,$2,$24;"
  372. "and $2,$2,$1;"
  373. "dsrl $2,$2,2;"
  374. "daddu $2,$2,$9;"
  376. WRITE8
  377. SWAPSET8
  379. "bne $4,$8,loophorverround;");
  380. }
  382. void CopyMBlock(unsigned char * Src, unsigned char * Dst, int SrcStride, int DstStride) 
  383. { 
  384. __asm( SET_SRCEND16
  386. "loopm:"
  387. CACHE16
  389. "uld $2,0($4);" 
  390. "uld $10,8($4);" 
  391. "addu $4,$4,$6;" 
  393. WRITE16
  395. "bne $4,$8,loopm;"); 
  396. } 
  398. void CopyMBlockHor(unsigned char * Src, unsigned char * Dst, int SrcStride, int DstStride) 
  399. { 
  400. __asm ( SET_SRCEND16
  401. SET_MASKS
  403. "loopmhor:"
  404. LOAD_FIRST16(0)
  405. LOAD_SECOND16(1)
  406. "addu $4,$4,$6;" 
  408. CACHE16
  409. AVG16
  410. WRITE16
  412. "bne $4,$8,loopmhor;");
  413. } 
  415. void CopyMBlockHorRound(unsigned char * Src, unsigned char * Dst, int SrcStride, int DstStride) 
  416. { 
  417. __asm ( SET_SRCEND16
  418. SET_MASKS
  420. "loopmhorround:"
  421. LOAD_FIRST16(0)
  422. LOAD_SECOND16(1)
  423. "addu $4,$4,$6;" 
  425. CACHE16
  426. AVGROUND16
  427. WRITE16
  429. "bne $4,$8,loopmhorround;");
  430. } 
  432. void CopyMBlockVer(unsigned char * Src, unsigned char * Dst, int SrcStride, int DstStride) 
  433. { 
  434. __asm ( SET_SRCEND16
  435. SET_MASKS
  437. LOAD_FIRST16(0)
  439. "loopmver:"
  440. "addu $4,$4,$6;" 
  442. LOAD_SECOND16(0)
  444. CACHE16
  445. AVG16
  446. WRITE16
  447. SWAPSET16
  449. "bne $4,$8,loopmver;"
  451. );
  452. } 
  454. void CopyMBlockVerRound(unsigned char * Src, unsigned char * Dst, int SrcStride, int DstStride) 
  455. { 
  456. __asm ( SET_SRCEND16
  457. SET_MASKS
  459. LOAD_FIRST16(0)
  461. "loopmverround:"
  462. "addu $4,$4,$6;" 
  464. LOAD_SECOND16(0)
  466. CACHE16
  467. AVGROUND16
  468. WRITE16
  469. SWAPSET16
  471. "bne $4,$8,loopmverround;");
  472. } 
  474. void CopyMBlockHorVer(unsigned char * Src, unsigned char * Dst, int SrcStride, int DstStride) 
  475. { 
  476. __asm ( SAVE
  477.     SET_SRCEND16
  478. SET_MASKS2
  480. "dsll $24,$3,1;"  
  481. "and $24,$24,$3;" // 0x0202 0202 0202 0202
  483. //preprocessing
  485. LOAD_FIRST16_HV
  487. "loopmhorver:"
  488. "addu $4,$4,$6;"
  490. LOAD_SECOND16_HV 
  491. CACHE16);
  493. __asm ( "daddu $2,$2,$12;"
  494. "daddu $9,$9,$13;"
  495. "daddu $10,$10,$14;"
  496. "daddu $11,$11,$15;"
  498. "daddu $2,$2,$24;"
  499. "daddu $10,$10,$24;"
  501. "and $2,$2,$1;"
  502. "and $10,$10,$1;"
  503. "dsrl $2,$2,2;"
  504. "dsrl $10,$10,2;"
  505. "daddu $2,$2,$9;"
  506. "daddu $10,$10,$11;"
  508. WRITE16
  509. SWAPSET16
  511. "bne $4,$8,loopmhorver;"
  513. RESTORE);
  514. } 
  516. void CopyMBlockHorVerRound(unsigned char * Src, unsigned char * Dst, int SrcStride, int DstStride) 
  517. { 
  518. __asm ( SAVE
  519.     SET_SRCEND16
  520. SET_MASKS2
  522. "dsrl $24,$3,1;"  
  523. "and $24,$24,$3;" // 0x0101 0101 0101 0101
  525. //preprocessing
  527. LOAD_FIRST16_HV
  529. "loopmhorverround:"
  530. "addu $4,$4,$6;");
  532. __asm ( LOAD_SECOND16_HV
  533. CACHE16
  535. "daddu $2,$2,$12;"
  536. "daddu $9,$9,$13;"
  537. "daddu $10,$10,$14;"
  538. "daddu $11,$11,$15;"
  540. "daddu $2,$2,$24;"
  541. "daddu $10,$10,$24;"
  543. "and $2,$2,$1;"
  544. "and $10,$10,$1;"
  545. "dsrl $2,$2,2;"
  546. "dsrl $10,$10,2;"
  547. "daddu $2,$2,$9;"
  548. "daddu $10,$10,$11;"
  550. WRITE16
  551. SWAPSET16
  553. "bne $4,$8,loopmhorverround;"
  555. RESTORE);
  556. }
  558. void AddBlock8x8(unsigned char * Src, unsigned char * Dst, int SrcStride, int DstStride) 
  559. { 
  560. __asm( SET_SRCEND8
  561. SET_MASKS
  563. "loopadd:"
  564. "ldr $2,0($4);"
  565. "addu $4,$4,$6;"
  567. "ldr $9,0($5);"
  568. "and $11,$2,$25;"
  569. "or $2,$2,$9;"
  570. "and $2,$2,$24;"
  571. "dsrl $11,$11,1;"
  572. "daddu $2,$2,$11;"
  573. "and $9,$9,$25;"
  574. "dsrl $9,$9,1;"
  575. "daddu $2,$2,$9;"
  577. "sdr $2,0($5);"
  578. "addu $5,$5,$7;" 
  579. "bne $4,$8,loopadd;"); 
  580. } 
  582. void AddBlock16x16(unsigned char * Src, unsigned char * Dst, int SrcStride, int DstStride) 
  583. { 
  584. __asm( SET_SRCEND16
  585. SET_MASKS
  587. "loopadd16:"
  588. "ldr $2,0($4);" 
  589. "ldr $10,8($4);" 
  590. #ifdef MIPSVR41XX
  591. ".set noreorder;"
  592. "cache 17,0($4);" // hit invalidate (lose changes)
  593. ".set reorder;"
  594. #endif
  595. "addu $4,$4,$6;" 
  597. "ldr $9,0($5);"
  598. "and $11,$2,$25;"
  599. "or $2,$2,$9;"
  600. "and $2,$2,$24;"
  601. "dsrl $11,$11,1;"
  602. "daddu $2,$2,$11;"
  603. "and $9,$9,$25;"
  604. "dsrl $9,$9,1;"
  605. "daddu $2,$2,$9;"
  607. "ldr $11,8($5);"
  608. "and $9,$10,$25;"
  609. "or $10,$10,$11;"
  610. "and $10,$10,$24;"
  611. "dsrl $9,$9,1;"
  612. "daddu $10,$10,$9;"
  613. "and $11,$11,$25;"
  614. "dsrl $11,$11,1;"
  615. "daddu $10,$10,$11;"
  617. "sdr $2,0($5);"
  618. "sdr $10,8($5);"
  619. "addu $5,$5,$7;" 
  621. "bne $4,$8,loopadd16;"); 
  622. } 
  624. #endif