Home Explore Help
Sign In
baoxu
/
frp
1
0
Fork 0
Files
Branch: dev
Branches Tags
frp
/
vendor
/
github.com
/
templexxx
/
xor
/
avx2_amd64.s

avx2_amd64.s 7.3 KB
Permalink History Raw

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438 
  1. #include "textflag.h"
    2. 

  2. 

  3. // addr of mem
    4. 

  4. #define DST BX
    5. 

  5. #define SRC SI
    6. 

  6. #define SRC0 TMP4
    7. 

  7. #define SRC1 TMP5
    8. 

  8. 

  9. // loop args
    10. 

  10. // num of vect
    11. 

  11. #define VECT CX
    12. 

  12. #define LEN DX
    13. 

  13. // pos of matrix
    14. 

  14. #define POS R8
    15. 

  15. 

  16. // tmp store
    17. 

  17. // num of vect or ...
    18. 

  18. #define TMP1 R9
    19. 

  19. // pos of matrix or ...
    20. 

  20. #define TMP2 R10
    21. 

  21. // store addr of data/parity or ...
    22. 

  22. #define TMP3 R11
    23. 

  23. #define TMP4 R12
    24. 

  24. #define TMP5 R13
    25. 

  25. #define TMP6 R14
    26. 

  26. 

  27. // func bytesAVX2mini(dst, src0, src1 []byte, size int)
    28. 

  28. TEXT ·bytesAVX2mini(SB), NOSPLIT, $0
    29. 

  29. 	MOVQ  len+72(FP), LEN
    30. 

  30. 	CMPQ  LEN, $0
    31. 

  31. 	JE    ret
    32. 

  32. 	MOVQ  dst+0(FP), DST
    33. 

  33. 	MOVQ  src0+24(FP), SRC0
    34. 

  34. 	MOVQ  src1+48(FP), SRC1
    35. 

  35. 	TESTQ $31, LEN
    36. 

  36. 	JNZ   not_aligned
    37. 

  37. 

  38. aligned:
    39. 

  39. 	MOVQ $0, POS
    40. 

  40. 

  41. loop32b:
    42. 

  42. 	VMOVDQU (SRC0)(POS*1), Y0
    43. 

  43. 	VPXOR   (SRC1)(POS*1), Y0, Y0
    44. 

  44. 	VMOVDQU Y0, (DST)(POS*1)
    45. 

  45. 	ADDQ    $32, POS
    46. 

  46. 	CMPQ    LEN, POS
    47. 

  47. 	JNE     loop32b
    48. 

  48. 	VZEROUPPER
    49. 

  49. 	RET
    50. 

  50. 

  51. loop_1b:
    52. 

  52. 	MOVB  -1(SRC0)(LEN*1), TMP1
    53. 

  53. 	MOVB  -1(SRC1)(LEN*1), TMP2
    54. 

  54. 	XORB  TMP1, TMP2
    55. 

  55. 	MOVB  TMP2, -1(DST)(LEN*1)
    56. 

  56. 	SUBQ  $1, LEN
    57. 

  57. 	TESTQ $7, LEN
    58. 

  58. 	JNZ   loop_1b
    59. 

  59. 	CMPQ  LEN, $0
    60. 

  60. 	JE    ret
    61. 

  61. 	TESTQ $31, LEN
    62. 

  62. 	JZ    aligned
    63. 

  63. 

  64. not_aligned:
    65. 

  65. 	TESTQ $7, LEN
    66. 

  66. 	JNE   loop_1b
    67. 

  67. 	MOVQ  LEN, TMP1
    68. 

  68. 	ANDQ  $31, TMP1
    69. 

  69. 

  70. loop_8b:
    71. 

  71. 	MOVQ -8(SRC0)(LEN*1), TMP2
    72. 

  72. 	MOVQ -8(SRC1)(LEN*1), TMP3
    73. 

  73. 	XORQ TMP2, TMP3
    74. 

  74. 	MOVQ TMP3, -8(DST)(LEN*1)
    75. 

  75. 	SUBQ $8, LEN
    76. 

  76. 	SUBQ $8, TMP1
    77. 

  77. 	JG   loop_8b
    78. 

  78. 

  79. 	CMPQ LEN, $32
    80. 

  80. 	JGE  aligned
    81. 

  81. 	RET
    82. 

  82. 

  83. ret:
    84. 

  84. 	RET
    85. 

  85. 

  86. // func bytesAVX2small(dst, src0, src1 []byte, size int)
    87. 

  87. TEXT ·bytesAVX2small(SB), NOSPLIT, $0
    88. 

  88. 	MOVQ  len+72(FP), LEN
    89. 

  89. 	CMPQ  LEN, $0
    90. 

  90. 	JE    ret
    91. 

  91. 	MOVQ  dst+0(FP), DST
    92. 

  92. 	MOVQ  src0+24(FP), SRC0
    93. 

  93. 	MOVQ  src1+48(FP), SRC1
    94. 

  94. 	TESTQ $127, LEN
    95. 

  95. 	JNZ   not_aligned
    96. 

  96. 

  97. aligned:
    98. 

  98. 	MOVQ $0, POS
    99. 

  99. 

  100. loop128b:
    101. 

  101. 	VMOVDQU (SRC0)(POS*1), Y0
    102. 

  102. 	VMOVDQU 32(SRC0)(POS*1), Y1
    103. 

  103. 	VMOVDQU 64(SRC0)(POS*1), Y2
    104. 

  104. 	VMOVDQU 96(SRC0)(POS*1), Y3
    105. 

  105. 	VPXOR   (SRC1)(POS*1), Y0, Y0
    106. 

  106. 	VPXOR   32(SRC1)(POS*1), Y1, Y1
    107. 

  107. 	VPXOR   64(SRC1)(POS*1), Y2, Y2
    108. 

  108. 	VPXOR   96(SRC1)(POS*1), Y3, Y3
    109. 

  109. 	VMOVDQU Y0, (DST)(POS*1)
    110. 

  110. 	VMOVDQU Y1, 32(DST)(POS*1)
    111. 

  111. 	VMOVDQU Y2, 64(DST)(POS*1)
    112. 

  112. 	VMOVDQU Y3, 96(DST)(POS*1)
    113. 

  113. 

  114. 	ADDQ $128, POS
    115. 

  115. 	CMPQ LEN, POS
    116. 

  116. 	JNE  loop128b
    117. 

  117. 	VZEROUPPER
    118. 

  118. 	RET
    119. 

  119. 

  120. loop_1b:
    121. 

  121. 	MOVB  -1(SRC0)(LEN*1), TMP1
    122. 

  122. 	MOVB  -1(SRC1)(LEN*1), TMP2
    123. 

  123. 	XORB  TMP1, TMP2
    124. 

  124. 	MOVB  TMP2, -1(DST)(LEN*1)
    125. 

  125. 	SUBQ  $1, LEN
    126. 

  126. 	TESTQ $7, LEN
    127. 

  127. 	JNZ   loop_1b
    128. 

  128. 	CMPQ  LEN, $0
    129. 

  129. 	JE    ret
    130. 

  130. 	TESTQ $127, LEN
    131. 

  131. 	JZ    aligned
    132. 

  132. 

  133. not_aligned:
    134. 

  134. 	TESTQ $7, LEN
    135. 

  135. 	JNE   loop_1b
    136. 

  136. 	MOVQ  LEN, TMP1
    137. 

  137. 	ANDQ  $127, TMP1
    138. 

  138. 

  139. loop_8b:
    140. 

  140. 	MOVQ -8(SRC0)(LEN*1), TMP2
    141. 

  141. 	MOVQ -8(SRC1)(LEN*1), TMP3
    142. 

  142. 	XORQ TMP2, TMP3
    143. 

  143. 	MOVQ TMP3, -8(DST)(LEN*1)
    144. 

  144. 	SUBQ $8, LEN
    145. 

  145. 	SUBQ $8, TMP1
    146. 

  146. 	JG   loop_8b
    147. 

  147. 

  148. 	CMPQ LEN, $128
    149. 

  149. 	JGE  aligned
    150. 

  150. 	RET
    151. 

  151. 

  152. ret:
    153. 

  153. 	RET
    154. 

  154. 

  155. // func bytesAVX2big(dst, src0, src1 []byte, size int)
    156. 

  156. TEXT ·bytesAVX2big(SB), NOSPLIT, $0
    157. 

  157. 	MOVQ  len+72(FP), LEN
    158. 

  158. 	CMPQ  LEN, $0
    159. 

  159. 	JE    ret
    160. 

  160. 	MOVQ  dst+0(FP), DST
    161. 

  161. 	MOVQ  src0+24(FP), SRC0
    162. 

  162. 	MOVQ  src1+48(FP), SRC1
    163. 

  163. 	TESTQ $127, LEN
    164. 

  164. 	JNZ   not_aligned
    165. 

  165. 

  166. aligned:
    167. 

  167. 	MOVQ $0, POS
    168. 

  168. 

  169. loop128b:
    170. 

  170. 	VMOVDQU (SRC0)(POS*1), Y0
    171. 

  171. 	VMOVDQU 32(SRC0)(POS*1), Y1
    172. 

  172. 	VMOVDQU 64(SRC0)(POS*1), Y2
    173. 

  173. 	VMOVDQU 96(SRC0)(POS*1), Y3
    174. 

  174. 	VPXOR   (SRC1)(POS*1), Y0, Y0
    175. 

  175. 	VPXOR   32(SRC1)(POS*1), Y1, Y1
    176. 

  176. 	VPXOR   64(SRC1)(POS*1), Y2, Y2
    177. 

  177. 	VPXOR   96(SRC1)(POS*1), Y3, Y3
    178. 

  178. 	LONG    $0xe77da1c4; WORD $0x0304
    179. 

  179. 	LONG    $0xe77da1c4; WORD $0x034c; BYTE $0x20
    180. 

  180. 	LONG    $0xe77da1c4; WORD $0x0354; BYTE $0x40
    181. 

  181. 	LONG    $0xe77da1c4; WORD $0x035c; BYTE $0x60
    182. 

  182. 

  183. 	ADDQ $128, POS
    184. 

  184. 	CMPQ LEN, POS
    185. 

  185. 	JNE  loop128b
    186. 

  186. 	SFENCE
    187. 

  187. 	VZEROUPPER
    188. 

  188. 	RET
    189. 

  189. 

  190. loop_1b:
    191. 

  191. 	MOVB  -1(SRC0)(LEN*1), TMP1
    192. 

  192. 	MOVB  -1(SRC1)(LEN*1), TMP2
    193. 

  193. 	XORB  TMP1, TMP2
    194. 

  194. 	MOVB  TMP2, -1(DST)(LEN*1)
    195. 

  195. 	SUBQ  $1, LEN
    196. 

  196. 	TESTQ $7, LEN
    197. 

  197. 	JNZ   loop_1b
    198. 

  198. 	CMPQ  LEN, $0
    199. 

  199. 	JE    ret
    200. 

  200. 	TESTQ $127, LEN
    201. 

  201. 	JZ    aligned
    202. 

  202. 

  203. not_aligned:
    204. 

  204. 	TESTQ $7, LEN
    205. 

  205. 	JNE   loop_1b
    206. 

  206. 	MOVQ  LEN, TMP1
    207. 

  207. 	ANDQ  $127, TMP1
    208. 

  208. 

  209. loop_8b:
    210. 

  210. 	MOVQ -8(SRC0)(LEN*1), TMP2
    211. 

  211. 	MOVQ -8(SRC1)(LEN*1), TMP3
    212. 

  212. 	XORQ TMP2, TMP3
    213. 

  213. 	MOVQ TMP3, -8(DST)(LEN*1)
    214. 

  214. 	SUBQ $8, LEN
    215. 

  215. 	SUBQ $8, TMP1
    216. 

  216. 	JG   loop_8b
    217. 

  217. 

  218. 	CMPQ LEN, $128
    219. 

  219. 	JGE  aligned
    220. 

  220. 	RET
    221. 

  221. 

  222. ret:
    223. 

  223. 	RET
    224. 

  224. 

  225. // func matrixAVX2small(dst []byte, src [][]byte)
    226. 

  226. TEXT ·matrixAVX2small(SB), NOSPLIT, $0
    227. 

  227. 	MOVQ  dst+0(FP), DST
    228. 

  228. 	MOVQ  src+24(FP), SRC
    229. 

  229. 	MOVQ  vec+32(FP), VECT
    230. 

  230. 	MOVQ  len+8(FP), LEN
    231. 

  231. 	TESTQ $127, LEN
    232. 

  232. 	JNZ   not_aligned
    233. 

  233. 

  234. aligned:
    235. 

  235. 	MOVQ $0, POS
    236. 

  236. 

  237. loop128b:
    238. 

  238. 	MOVQ    VECT, TMP1
    239. 

  239. 	SUBQ    $2, TMP1
    240. 

  240. 	MOVQ    $0, TMP2
    241. 

  241. 	MOVQ    (SRC)(TMP2*1), TMP3
    242. 

  242. 	MOVQ    TMP3, TMP4
    243. 

  243. 	VMOVDQU (TMP3)(POS*1), Y0
    244. 

  244. 	VMOVDQU 32(TMP4)(POS*1), Y1
    245. 

  245. 	VMOVDQU 64(TMP3)(POS*1), Y2
    246. 

  246. 	VMOVDQU 96(TMP4)(POS*1), Y3
    247. 

  247. 

  248. next_vect:
    249. 

  249. 	ADDQ    $24, TMP2
    250. 

  250. 	MOVQ    (SRC)(TMP2*1), TMP3
    251. 

  251. 	MOVQ    TMP3, TMP4
    252. 

  252. 	VMOVDQU (TMP3)(POS*1), Y4
    253. 

  253. 	VMOVDQU 32(TMP4)(POS*1), Y5
    254. 

  254. 	VMOVDQU 64(TMP3)(POS*1), Y6
    255. 

  255. 	VMOVDQU 96(TMP4)(POS*1), Y7
    256. 

  256. 	VPXOR   Y4, Y0, Y0
    257. 

  257. 	VPXOR   Y5, Y1, Y1
    258. 

  258. 	VPXOR   Y6, Y2, Y2
    259. 

  259. 	VPXOR   Y7, Y3, Y3
    260. 

  260. 	SUBQ    $1, TMP1
    261. 

  261. 	JGE     next_vect
    262. 

  262. 

  263. 	VMOVDQU Y0, (DST)(POS*1)
    264. 

  264. 	VMOVDQU Y1, 32(DST)(POS*1)
    265. 

  265. 	VMOVDQU Y2, 64(DST)(POS*1)
    266. 

  266. 	VMOVDQU Y3, 96(DST)(POS*1)
    267. 

  267. 

  268. 	ADDQ $128, POS
    269. 

  269. 	CMPQ LEN, POS
    270. 

  270. 	JNE  loop128b
    271. 

  271. 	VZEROUPPER
    272. 

  272. 	RET
    273. 

  273. 

  274. loop_1b:
    275. 

  275. 	MOVQ VECT, TMP1
    276. 

  276. 	MOVQ $0, TMP2
    277. 

  277. 	MOVQ (SRC)(TMP2*1), TMP3
    278. 

  278. 	SUBQ $2, TMP1
    279. 

  279. 	MOVB -1(TMP3)(LEN*1), TMP5
    280. 

  280. 

  281. next_vect_1b:
    282. 

  282. 	ADDQ $24, TMP2
    283. 

  283. 	MOVQ (SRC)(TMP2*1), TMP3
    284. 

  284. 	MOVB -1(TMP3)(LEN*1), TMP6
    285. 

  285. 	XORB TMP6, TMP5
    286. 

  286. 	SUBQ $1, TMP1
    287. 

  287. 	JGE  next_vect_1b
    288. 

  288. 

  289. 	MOVB  TMP5, -1(DST)(LEN*1)
    290. 

  290. 	SUBQ  $1, LEN
    291. 

  291. 	TESTQ $7, LEN
    292. 

  292. 	JNZ   loop_1b
    293. 

  293. 

  294. 	CMPQ  LEN, $0
    295. 

  295. 	JE    ret
    296. 

  296. 	TESTQ $127, LEN
    297. 

  297. 	JZ    aligned
    298. 

  298. 

  299. not_aligned:
    300. 

  300. 	TESTQ $7, LEN
    301. 

  301. 	JNE   loop_1b
    302. 

  302. 	MOVQ  LEN, TMP4
    303. 

  303. 	ANDQ  $127, TMP4
    304. 

  304. 

  305. loop_8b:
    306. 

  306. 	MOVQ VECT, TMP1
    307. 

  307. 	MOVQ $0, TMP2
    308. 

  308. 	MOVQ (SRC)(TMP2*1), TMP3
    309. 

  309. 	SUBQ $2, TMP1
    310. 

  310. 	MOVQ -8(TMP3)(LEN*1), TMP5
    311. 

  311. 

  312. next_vect_8b:
    313. 

  313. 	ADDQ $24, TMP2
    314. 

  314. 	MOVQ (SRC)(TMP2*1), TMP3
    315. 

  315. 	MOVQ -8(TMP3)(LEN*1), TMP6
    316. 

  316. 	XORQ TMP6, TMP5
    317. 

  317. 	SUBQ $1, TMP1
    318. 

  318. 	JGE  next_vect_8b
    319. 

  319. 

  320. 	MOVQ TMP5, -8(DST)(LEN*1)
    321. 

  321. 	SUBQ $8, LEN
    322. 

  322. 	SUBQ $8, TMP4
    323. 

  323. 	JG   loop_8b
    324. 

  324. 

  325. 	CMPQ LEN, $128
    326. 

  326. 	JGE  aligned
    327. 

  327. 	RET
    328. 

  328. 

  329. ret:
    330. 

  330. 	RET
    331. 

  331. 

  332. // func matrixAVX2big(dst []byte, src [][]byte)
    333. 

  333. TEXT ·matrixAVX2big(SB), NOSPLIT, $0
    334. 

  334. 	MOVQ  dst+0(FP), DST
    335. 

  335. 	MOVQ  src+24(FP), SRC
    336. 

  336. 	MOVQ  vec+32(FP), VECT
    337. 

  337. 	MOVQ  len+8(FP), LEN
    338. 

  338. 	TESTQ $127, LEN
    339. 

  339. 	JNZ   not_aligned
    340. 

  340. 

  341. aligned:
    342. 

  342. 	MOVQ $0, POS
    343. 

  343. 

  344. loop128b:
    345. 

  345. 	MOVQ    VECT, TMP1
    346. 

  346. 	SUBQ    $2, TMP1
    347. 

  347. 	MOVQ    $0, TMP2
    348. 

  348. 	MOVQ    (SRC)(TMP2*1), TMP3
    349. 

  349. 	MOVQ    TMP3, TMP4
    350. 

  350. 	VMOVDQU (TMP3)(POS*1), Y0
    351. 

  351. 	VMOVDQU 32(TMP4)(POS*1), Y1
    352. 

  352. 	VMOVDQU 64(TMP3)(POS*1), Y2
    353. 

  353. 	VMOVDQU 96(TMP4)(POS*1), Y3
    354. 

  354. 

  355. next_vect:
    356. 

  356. 	ADDQ    $24, TMP2
    357. 

  357. 	MOVQ    (SRC)(TMP2*1), TMP3
    358. 

  358. 	MOVQ    TMP3, TMP4
    359. 

  359. 	VMOVDQU (TMP3)(POS*1), Y4
    360. 

  360. 	VMOVDQU 32(TMP4)(POS*1), Y5
    361. 

  361. 	VMOVDQU 64(TMP3)(POS*1), Y6
    362. 

  362. 	VMOVDQU 96(TMP4)(POS*1), Y7
    363. 

  363. 	VPXOR   Y4, Y0, Y0
    364. 

  364. 	VPXOR   Y5, Y1, Y1
    365. 

  365. 	VPXOR   Y6, Y2, Y2
    366. 

  366. 	VPXOR   Y7, Y3, Y3
    367. 

  367. 	SUBQ    $1, TMP1
    368. 

  368. 	JGE     next_vect
    369. 

  369. 

  370. 	LONG $0xe77da1c4; WORD $0x0304             // VMOVNTDQ  go1.8 has
    371. 

  371. 	LONG $0xe77da1c4; WORD $0x034c; BYTE $0x20
    372. 

  372. 	LONG $0xe77da1c4; WORD $0x0354; BYTE $0x40
    373. 

  373. 	LONG $0xe77da1c4; WORD $0x035c; BYTE $0x60
    374. 

  374. 

  375. 	ADDQ $128, POS
    376. 

  376. 	CMPQ LEN, POS
    377. 

  377. 	JNE  loop128b
    378. 

  378. 	VZEROUPPER
    379. 

  379. 	RET
    380. 

  380. 

  381. loop_1b:
    382. 

  382. 	MOVQ VECT, TMP1
    383. 

  383. 	MOVQ $0, TMP2
    384. 

  384. 	MOVQ (SRC)(TMP2*1), TMP3
    385. 

  385. 	SUBQ $2, TMP1
    386. 

  386. 	MOVB -1(TMP3)(LEN*1), TMP5
    387. 

  387. 

  388. next_vect_1b:
    389. 

  389. 	ADDQ $24, TMP2
    390. 

  390. 	MOVQ (SRC)(TMP2*1), TMP3
    391. 

  391. 	MOVB -1(TMP3)(LEN*1), TMP6
    392. 

  392. 	XORB TMP6, TMP5
    393. 

  393. 	SUBQ $1, TMP1
    394. 

  394. 	JGE  next_vect_1b
    395. 

  395. 

  396. 	MOVB  TMP5, -1(DST)(LEN*1)
    397. 

  397. 	SUBQ  $1, LEN
    398. 

  398. 	TESTQ $7, LEN
    399. 

  399. 	JNZ   loop_1b
    400. 

  400. 

  401. 	CMPQ  LEN, $0
    402. 

  402. 	JE    ret
    403. 

  403. 	TESTQ $127, LEN
    404. 

  404. 	JZ    aligned
    405. 

  405. 

  406. not_aligned:
    407. 

  407. 	TESTQ $7, LEN
    408. 

  408. 	JNE   loop_1b
    409. 

  409. 	MOVQ  LEN, TMP4
    410. 

  410. 	ANDQ  $127, TMP4
    411. 

  411. 

  412. loop_8b:
    413. 

  413. 	MOVQ VECT, TMP1
    414. 

  414. 	MOVQ $0, TMP2
    415. 

  415. 	MOVQ (SRC)(TMP2*1), TMP3
    416. 

  416. 	SUBQ $2, TMP1
    417. 

  417. 	MOVQ -8(TMP3)(LEN*1), TMP5
    418. 

  418. 

  419. next_vect_8b:
    420. 

  420. 	ADDQ $24, TMP2
    421. 

  421. 	MOVQ (SRC)(TMP2*1), TMP3
    422. 

  422. 	MOVQ -8(TMP3)(LEN*1), TMP6
    423. 

  423. 	XORQ TMP6, TMP5
    424. 

  424. 	SUBQ $1, TMP1
    425. 

  425. 	JGE  next_vect_8b
    426. 

  426. 

  427. 	MOVQ TMP5, -8(DST)(LEN*1)
    428. 

  428. 	SUBQ $8, LEN
    429. 

  429. 	SUBQ $8, TMP4
    430. 

  430. 	JG   loop_8b
    431. 

  431. 

  432. 	CMPQ LEN, $128
    433. 

  433. 	JGE  aligned
    434. 

  434. 	RET
    435. 

  435. 

  436. ret:
    437. 

  437. 	RET
    438. 

  438.