falkTX
/
FFmpeg
mirror of https://github.com/falkTX/FFmpeg.git
1
0
Fork 0
Code Issues 0 Releases 338 Wiki Activity
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
38308 Commits
32 Branches
1.0GB
Tree: ce1e8045e0
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from 'ce1e8045e0'
${ noResults }
FFmpeg/libavcodec/arm/h264qpel_neon.S

956 lines
32KB
Raw Blame History 

  1. /*

  2.  * Copyright (c) 2008 Mans Rullgard <mans@mansr.com>

  3.  *

  4.  * This file is part of Libav.

  5.  *

  6.  * Libav is free software; you can redistribute it and/or

  7.  * modify it under the terms of the GNU Lesser General Public

  8.  * License as published by the Free Software Foundation; either

  9.  * version 2.1 of the License, or (at your option) any later version.

  10.  *

  11.  * Libav is distributed in the hope that it will be useful,

  12.  * but WITHOUT ANY WARRANTY; without even the implied warranty of

  13.  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU

  14.  * Lesser General Public License for more details.

  15.  *

  16.  * You should have received a copy of the GNU Lesser General Public

  17.  * License along with Libav; if not, write to the Free Software

  18.  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA

  19.  */

  20. 

  21. #include "libavutil/arm/asm.S"

  22. #include "neon.S"

  23. 

  24.         /* H.264 qpel MC */

  25. 

  26. .macro  lowpass_const   r

  27.         movw            \r,  #5

  28.         movt            \r,  #20

  29.         vmov.32         d6[0], \r

  30. .endm

  31. 

  32. .macro  lowpass_8       r0,  r1,  r2,  r3,  d0,  d1,  narrow=1

  33.   .if \narrow

  34.         t0 .req q0

  35.         t1 .req q8

  36.   .else

  37.         t0 .req \d0

  38.         t1 .req \d1

  39.   .endif

  40.         vext.8          d2,  \r0, \r1, #2

  41.         vext.8          d3,  \r0, \r1, #3

  42.         vaddl.u8        q1,  d2,  d3

  43.         vext.8          d4,  \r0, \r1, #1

  44.         vext.8          d5,  \r0, \r1, #4

  45.         vaddl.u8        q2,  d4,  d5

  46.         vext.8          d30, \r0, \r1, #5

  47.         vaddl.u8        t0,  \r0, d30

  48.         vext.8          d18, \r2, \r3, #2

  49.         vmla.i16        t0,  q1,  d6[1]

  50.         vext.8          d19, \r2, \r3, #3

  51.         vaddl.u8        q9,  d18, d19

  52.         vext.8          d20, \r2, \r3, #1

  53.         vmls.i16        t0,  q2,  d6[0]

  54.         vext.8          d21, \r2, \r3, #4

  55.         vaddl.u8        q10, d20, d21

  56.         vext.8          d31, \r2, \r3, #5

  57.         vaddl.u8        t1,  \r2, d31

  58.         vmla.i16        t1,  q9,  d6[1]

  59.         vmls.i16        t1,  q10, d6[0]

  60.   .if \narrow

  61.         vqrshrun.s16    \d0, t0,  #5

  62.         vqrshrun.s16    \d1, t1,  #5

  63.   .endif

  64.         .unreq  t0

  65.         .unreq  t1

  66. .endm

  67. 

  68. .macro  lowpass_8_1     r0,  r1,  d0,  narrow=1

  69.   .if \narrow

  70.         t0 .req q0

  71.   .else

  72.         t0 .req \d0

  73.   .endif

  74.         vext.8          d2,  \r0, \r1, #2

  75.         vext.8          d3,  \r0, \r1, #3

  76.         vaddl.u8        q1,  d2,  d3

  77.         vext.8          d4,  \r0, \r1, #1

  78.         vext.8          d5,  \r0, \r1, #4

  79.         vaddl.u8        q2,  d4,  d5

  80.         vext.8          d30, \r0, \r1, #5

  81.         vaddl.u8        t0,  \r0, d30

  82.         vmla.i16        t0,  q1,  d6[1]

  83.         vmls.i16        t0,  q2,  d6[0]

  84.   .if \narrow

  85.         vqrshrun.s16    \d0, t0,  #5

  86.   .endif

  87.         .unreq  t0

  88. .endm

  89. 

  90. .macro  lowpass_8.16    r0,  r1,  l0,  h0,  l1,  h1,  d

  91.         vext.16         q1,  \r0, \r1, #2

  92.         vext.16         q0,  \r0, \r1, #3

  93.         vaddl.s16       q9,  d2,  d0

  94.         vext.16         q2,  \r0, \r1, #1

  95.         vaddl.s16       q1,  d3,  d1

  96.         vext.16         q3,  \r0, \r1, #4

  97.         vaddl.s16       q10, d4,  d6

  98.         vext.16         \r1, \r0, \r1, #5

  99.         vaddl.s16       q2,  d5,  d7

  100.         vaddl.s16       q0,  \h0, \h1

  101.         vaddl.s16       q8,  \l0, \l1

  102. 

  103.         vshl.i32        q3,  q9,  #4

  104.         vshl.i32        q9,  q9,  #2

  105.         vshl.i32        q15, q10, #2

  106.         vadd.i32        q9,  q9,  q3

  107.         vadd.i32        q10, q10, q15

  108. 

  109.         vshl.i32        q3,  q1,  #4

  110.         vshl.i32        q1,  q1,  #2

  111.         vshl.i32        q15, q2,  #2

  112.         vadd.i32        q1,  q1,  q3

  113.         vadd.i32        q2,  q2,  q15

  114. 

  115.         vadd.i32        q9,  q9,  q8

  116.         vsub.i32        q9,  q9,  q10

  117. 

  118.         vadd.i32        q1,  q1,  q0

  119.         vsub.i32        q1,  q1,  q2

  120. 

  121.         vrshrn.s32      d18, q9,  #10

  122.         vrshrn.s32      d19, q1,  #10

  123. 

  124.         vqmovun.s16     \d,  q9

  125. .endm

  126. 

  127. function put_h264_qpel16_h_lowpass_neon_packed

  128.         mov             r4,  lr

  129.         mov             r12, #16

  130.         mov             r3,  #8

  131.         bl              put_h264_qpel8_h_lowpass_neon

  132.         sub             r1,  r1,  r2, lsl #4

  133.         add             r1,  r1,  #8

  134.         mov             r12, #16

  135.         mov             lr,  r4

  136.         b               put_h264_qpel8_h_lowpass_neon

  137. endfunc

  138. 

  139. .macro  h264_qpel_h_lowpass type

  140. function \type\()_h264_qpel16_h_lowpass_neon

  141.         push            {lr}

  142.         mov             r12, #16

  143.         bl              \type\()_h264_qpel8_h_lowpass_neon

  144.         sub             r0,  r0,  r3, lsl #4

  145.         sub             r1,  r1,  r2, lsl #4

  146.         add             r0,  r0,  #8

  147.         add             r1,  r1,  #8

  148.         mov             r12, #16

  149.         pop             {lr}

  150. endfunc

  151. 

  152. function \type\()_h264_qpel8_h_lowpass_neon

  153. 1:      vld1.8          {d0, d1},  [r1], r2

  154.         vld1.8          {d16,d17}, [r1], r2

  155.         subs            r12, r12, #2

  156.         lowpass_8       d0,  d1,  d16, d17, d0,  d16

  157.   .ifc \type,avg

  158.         vld1.8          {d2},     [r0,:64], r3

  159.         vrhadd.u8       d0,  d0,  d2

  160.         vld1.8          {d3},     [r0,:64]

  161.         vrhadd.u8       d16, d16, d3

  162.         sub             r0,  r0,  r3

  163.   .endif

  164.         vst1.8          {d0},     [r0,:64], r3

  165.         vst1.8          {d16},    [r0,:64], r3

  166.         bne             1b

  167.         bx              lr

  168. endfunc

  169. .endm

  170. 

  171.         h264_qpel_h_lowpass put

  172.         h264_qpel_h_lowpass avg

  173. 

  174. .macro  h264_qpel_h_lowpass_l2 type

  175. function \type\()_h264_qpel16_h_lowpass_l2_neon

  176.         push            {lr}

  177.         mov             r12, #16

  178.         bl              \type\()_h264_qpel8_h_lowpass_l2_neon

  179.         sub             r0,  r0,  r2, lsl #4

  180.         sub             r1,  r1,  r2, lsl #4

  181.         sub             r3,  r3,  r2, lsl #4

  182.         add             r0,  r0,  #8

  183.         add             r1,  r1,  #8

  184.         add             r3,  r3,  #8

  185.         mov             r12, #16

  186.         pop             {lr}

  187. endfunc

  188. 

  189. function \type\()_h264_qpel8_h_lowpass_l2_neon

  190. 1:      vld1.8          {d0, d1},  [r1], r2

  191.         vld1.8          {d16,d17}, [r1], r2

  192.         vld1.8          {d28},     [r3], r2

  193.         vld1.8          {d29},     [r3], r2

  194.         subs            r12, r12, #2

  195.         lowpass_8       d0,  d1,  d16, d17, d0,  d1

  196.         vrhadd.u8       q0,  q0,  q14

  197.   .ifc \type,avg

  198.         vld1.8          {d2},      [r0,:64], r2

  199.         vrhadd.u8       d0,  d0,  d2

  200.         vld1.8          {d3},      [r0,:64]

  201.         vrhadd.u8       d1,  d1,  d3

  202.         sub             r0,  r0,  r2

  203.   .endif

  204.         vst1.8          {d0},      [r0,:64], r2

  205.         vst1.8          {d1},      [r0,:64], r2

  206.         bne             1b

  207.         bx              lr

  208. endfunc

  209. .endm

  210. 

  211.         h264_qpel_h_lowpass_l2 put

  212.         h264_qpel_h_lowpass_l2 avg

  213. 

  214. function put_h264_qpel16_v_lowpass_neon_packed

  215.         mov             r4,  lr

  216.         mov             r2,  #8

  217.         bl              put_h264_qpel8_v_lowpass_neon

  218.         sub             r1,  r1,  r3, lsl #2

  219.         bl              put_h264_qpel8_v_lowpass_neon

  220.         sub             r1,  r1,  r3, lsl #4

  221.         sub             r1,  r1,  r3, lsl #2

  222.         add             r1,  r1,  #8

  223.         bl              put_h264_qpel8_v_lowpass_neon

  224.         sub             r1,  r1,  r3, lsl #2

  225.         mov             lr,  r4

  226.         b               put_h264_qpel8_v_lowpass_neon

  227. endfunc

  228. 

  229. .macro  h264_qpel_v_lowpass type

  230. function \type\()_h264_qpel16_v_lowpass_neon

  231.         mov             r4,  lr

  232.         bl              \type\()_h264_qpel8_v_lowpass_neon

  233.         sub             r1,  r1,  r3, lsl #2

  234.         bl              \type\()_h264_qpel8_v_lowpass_neon

  235.         sub             r0,  r0,  r2, lsl #4

  236.         add             r0,  r0,  #8

  237.         sub             r1,  r1,  r3, lsl #4

  238.         sub             r1,  r1,  r3, lsl #2

  239.         add             r1,  r1,  #8

  240.         bl              \type\()_h264_qpel8_v_lowpass_neon

  241.         sub             r1,  r1,  r3, lsl #2

  242.         mov             lr,  r4

  243. endfunc

  244. 

  245. function \type\()_h264_qpel8_v_lowpass_neon

  246.         vld1.8          {d8},  [r1], r3

  247.         vld1.8          {d10}, [r1], r3

  248.         vld1.8          {d12}, [r1], r3

  249.         vld1.8          {d14}, [r1], r3

  250.         vld1.8          {d22}, [r1], r3

  251.         vld1.8          {d24}, [r1], r3

  252.         vld1.8          {d26}, [r1], r3

  253.         vld1.8          {d28}, [r1], r3

  254.         vld1.8          {d9},  [r1], r3

  255.         vld1.8          {d11}, [r1], r3

  256.         vld1.8          {d13}, [r1], r3

  257.         vld1.8          {d15}, [r1], r3

  258.         vld1.8          {d23}, [r1]

  259. 

  260.         transpose_8x8   q4,  q5,  q6,  q7,  q11, q12, q13, q14

  261.         lowpass_8       d8,  d9,  d10, d11, d8,  d10

  262.         lowpass_8       d12, d13, d14, d15, d12, d14

  263.         lowpass_8       d22, d23, d24, d25, d22, d24

  264.         lowpass_8       d26, d27, d28, d29, d26, d28

  265.         transpose_8x8   d8,  d10, d12, d14, d22, d24, d26, d28

  266. 

  267.   .ifc \type,avg

  268.         vld1.8          {d9},  [r0,:64], r2

  269.         vrhadd.u8       d8,  d8,  d9

  270.         vld1.8          {d11}, [r0,:64], r2

  271.         vrhadd.u8       d10, d10, d11

  272.         vld1.8          {d13}, [r0,:64], r2

  273.         vrhadd.u8       d12, d12, d13

  274.         vld1.8          {d15}, [r0,:64], r2

  275.         vrhadd.u8       d14, d14, d15

  276.         vld1.8          {d23}, [r0,:64], r2

  277.         vrhadd.u8       d22, d22, d23

  278.         vld1.8          {d25}, [r0,:64], r2

  279.         vrhadd.u8       d24, d24, d25

  280.         vld1.8          {d27}, [r0,:64], r2

  281.         vrhadd.u8       d26, d26, d27

  282.         vld1.8          {d29}, [r0,:64], r2

  283.         vrhadd.u8       d28, d28, d29

  284.         sub             r0,  r0,  r2,  lsl #3

  285.   .endif

  286. 

  287.         vst1.8          {d8},  [r0,:64], r2

  288.         vst1.8          {d10}, [r0,:64], r2

  289.         vst1.8          {d12}, [r0,:64], r2

  290.         vst1.8          {d14}, [r0,:64], r2

  291.         vst1.8          {d22}, [r0,:64], r2

  292.         vst1.8          {d24}, [r0,:64], r2

  293.         vst1.8          {d26}, [r0,:64], r2

  294.         vst1.8          {d28}, [r0,:64], r2

  295. 

  296.         bx              lr

  297. endfunc

  298. .endm

  299. 

  300.         h264_qpel_v_lowpass put

  301.         h264_qpel_v_lowpass avg

  302. 

  303. .macro  h264_qpel_v_lowpass_l2 type

  304. function \type\()_h264_qpel16_v_lowpass_l2_neon

  305.         mov             r4,  lr

  306.         bl              \type\()_h264_qpel8_v_lowpass_l2_neon

  307.         sub             r1,  r1,  r3, lsl #2

  308.         bl              \type\()_h264_qpel8_v_lowpass_l2_neon

  309.         sub             r0,  r0,  r3, lsl #4

  310.         sub             r12, r12, r2, lsl #4

  311.         add             r0,  r0,  #8

  312.         add             r12, r12, #8

  313.         sub             r1,  r1,  r3, lsl #4

  314.         sub             r1,  r1,  r3, lsl #2

  315.         add             r1,  r1,  #8

  316.         bl              \type\()_h264_qpel8_v_lowpass_l2_neon

  317.         sub             r1,  r1,  r3, lsl #2

  318.         mov             lr,  r4

  319. endfunc

  320. 

  321. function \type\()_h264_qpel8_v_lowpass_l2_neon

  322.         vld1.8          {d8},  [r1], r3

  323.         vld1.8          {d10}, [r1], r3

  324.         vld1.8          {d12}, [r1], r3

  325.         vld1.8          {d14}, [r1], r3

  326.         vld1.8          {d22}, [r1], r3

  327.         vld1.8          {d24}, [r1], r3

  328.         vld1.8          {d26}, [r1], r3

  329.         vld1.8          {d28}, [r1], r3

  330.         vld1.8          {d9},  [r1], r3

  331.         vld1.8          {d11}, [r1], r3

  332.         vld1.8          {d13}, [r1], r3

  333.         vld1.8          {d15}, [r1], r3

  334.         vld1.8          {d23}, [r1]

  335. 

  336.         transpose_8x8   q4,  q5,  q6,  q7,  q11, q12, q13, q14

  337.         lowpass_8       d8,  d9,  d10, d11, d8,  d9

  338.         lowpass_8       d12, d13, d14, d15, d12, d13

  339.         lowpass_8       d22, d23, d24, d25, d22, d23

  340.         lowpass_8       d26, d27, d28, d29, d26, d27

  341.         transpose_8x8   d8,  d9,  d12, d13, d22, d23, d26, d27

  342. 

  343.         vld1.8          {d0},  [r12], r2

  344.         vld1.8          {d1},  [r12], r2

  345.         vld1.8          {d2},  [r12], r2

  346.         vld1.8          {d3},  [r12], r2

  347.         vld1.8          {d4},  [r12], r2

  348.         vrhadd.u8       q0,  q0,  q4

  349.         vld1.8          {d5},  [r12], r2

  350.         vrhadd.u8       q1,  q1,  q6

  351.         vld1.8          {d10}, [r12], r2

  352.         vrhadd.u8       q2,  q2,  q11

  353.         vld1.8          {d11}, [r12], r2

  354.         vrhadd.u8       q5,  q5,  q13

  355. 

  356.   .ifc \type,avg

  357.         vld1.8          {d16}, [r0,:64], r3

  358.         vrhadd.u8       d0,  d0,  d16

  359.         vld1.8          {d17}, [r0,:64], r3

  360.         vrhadd.u8       d1,  d1,  d17

  361.         vld1.8          {d16}, [r0,:64], r3

  362.         vrhadd.u8       d2,  d2,  d16

  363.         vld1.8          {d17}, [r0,:64], r3

  364.         vrhadd.u8       d3,  d3,  d17

  365.         vld1.8          {d16}, [r0,:64], r3

  366.         vrhadd.u8       d4,  d4,  d16

  367.         vld1.8          {d17}, [r0,:64], r3

  368.         vrhadd.u8       d5,  d5,  d17

  369.         vld1.8          {d16}, [r0,:64], r3

  370.         vrhadd.u8       d10, d10, d16

  371.         vld1.8          {d17}, [r0,:64], r3

  372.         vrhadd.u8       d11, d11, d17

  373.         sub             r0,  r0,  r3,  lsl #3

  374.   .endif

  375. 

  376.         vst1.8          {d0},  [r0,:64], r3

  377.         vst1.8          {d1},  [r0,:64], r3

  378.         vst1.8          {d2},  [r0,:64], r3

  379.         vst1.8          {d3},  [r0,:64], r3

  380.         vst1.8          {d4},  [r0,:64], r3

  381.         vst1.8          {d5},  [r0,:64], r3

  382.         vst1.8          {d10}, [r0,:64], r3

  383.         vst1.8          {d11}, [r0,:64], r3

  384. 

  385.         bx              lr

  386. endfunc

  387. .endm

  388. 

  389.         h264_qpel_v_lowpass_l2 put

  390.         h264_qpel_v_lowpass_l2 avg

  391. 

  392. function put_h264_qpel8_hv_lowpass_neon_top

  393.         lowpass_const   r12

  394.         mov             r12, #12

  395. 1:      vld1.8          {d0, d1},  [r1], r3

  396.         vld1.8          {d16,d17}, [r1], r3

  397.         subs            r12, r12, #2

  398.         lowpass_8       d0,  d1,  d16, d17, q11, q12, narrow=0

  399.         vst1.8          {d22-d25}, [r4,:128]!

  400.         bne             1b

  401. 

  402.         vld1.8          {d0, d1},  [r1]

  403.         lowpass_8_1     d0,  d1,  q12, narrow=0

  404. 

  405.         mov             r12, #-16

  406.         add             r4,  r4,  r12

  407.         vld1.8          {d30,d31}, [r4,:128], r12

  408.         vld1.8          {d20,d21}, [r4,:128], r12

  409.         vld1.8          {d18,d19}, [r4,:128], r12

  410.         vld1.8          {d16,d17}, [r4,:128], r12

  411.         vld1.8          {d14,d15}, [r4,:128], r12

  412.         vld1.8          {d12,d13}, [r4,:128], r12

  413.         vld1.8          {d10,d11}, [r4,:128], r12

  414.         vld1.8          {d8, d9},  [r4,:128], r12

  415.         vld1.8          {d6, d7},  [r4,:128], r12

  416.         vld1.8          {d4, d5},  [r4,:128], r12

  417.         vld1.8          {d2, d3},  [r4,:128], r12

  418.         vld1.8          {d0, d1},  [r4,:128]

  419. 

  420.         swap4           d1,  d3,  d5,  d7,  d8,  d10, d12, d14

  421.         transpose16_4x4 q0,  q1,  q2,  q3,  q4,  q5,  q6,  q7

  422. 

  423.         swap4           d17, d19, d21, d31, d24, d26, d28, d22

  424.         transpose16_4x4 q8,  q9,  q10, q15, q12, q13, q14, q11

  425. 

  426.         vst1.8          {d30,d31}, [r4,:128]!

  427.         vst1.8          {d6, d7},  [r4,:128]!

  428.         vst1.8          {d20,d21}, [r4,:128]!

  429.         vst1.8          {d4, d5},  [r4,:128]!

  430.         vst1.8          {d18,d19}, [r4,:128]!

  431.         vst1.8          {d2, d3},  [r4,:128]!

  432.         vst1.8          {d16,d17}, [r4,:128]!

  433.         vst1.8          {d0, d1},  [r4,:128]

  434. 

  435.         lowpass_8.16    q4,  q12, d8,  d9,  d24, d25, d8

  436.         lowpass_8.16    q5,  q13, d10, d11, d26, d27, d9

  437.         lowpass_8.16    q6,  q14, d12, d13, d28, d29, d10

  438.         lowpass_8.16    q7,  q11, d14, d15, d22, d23, d11

  439. 

  440.         vld1.8          {d16,d17}, [r4,:128], r12

  441.         vld1.8          {d30,d31}, [r4,:128], r12

  442.         lowpass_8.16    q8,  q15, d16, d17, d30, d31, d12

  443.         vld1.8          {d16,d17}, [r4,:128], r12

  444.         vld1.8          {d30,d31}, [r4,:128], r12

  445.         lowpass_8.16    q8,  q15, d16, d17, d30, d31, d13

  446.         vld1.8          {d16,d17}, [r4,:128], r12

  447.         vld1.8          {d30,d31}, [r4,:128], r12

  448.         lowpass_8.16    q8,  q15, d16, d17, d30, d31, d14

  449.         vld1.8          {d16,d17}, [r4,:128], r12

  450.         vld1.8          {d30,d31}, [r4,:128]

  451.         lowpass_8.16    q8,  q15, d16, d17, d30, d31, d15

  452. 

  453.         transpose_8x8   d12, d13, d14, d15, d8,  d9,  d10, d11

  454. 

  455.         bx              lr

  456. endfunc

  457. 

  458. .macro  h264_qpel8_hv_lowpass type

  459. function \type\()_h264_qpel8_hv_lowpass_neon

  460.         mov             r10, lr

  461.         bl              put_h264_qpel8_hv_lowpass_neon_top

  462.   .ifc \type,avg

  463.         vld1.8          {d0},      [r0,:64], r2

  464.         vrhadd.u8       d12, d12, d0

  465.         vld1.8          {d1},      [r0,:64], r2

  466.         vrhadd.u8       d13, d13, d1

  467.         vld1.8          {d2},      [r0,:64], r2

  468.         vrhadd.u8       d14, d14, d2

  469.         vld1.8          {d3},      [r0,:64], r2

  470.         vrhadd.u8       d15, d15, d3

  471.         vld1.8          {d4},      [r0,:64], r2

  472.         vrhadd.u8       d8,  d8,  d4

  473.         vld1.8          {d5},      [r0,:64], r2

  474.         vrhadd.u8       d9,  d9,  d5

  475.         vld1.8          {d6},      [r0,:64], r2

  476.         vrhadd.u8       d10, d10, d6

  477.         vld1.8          {d7},      [r0,:64], r2

  478.         vrhadd.u8       d11, d11, d7

  479.         sub             r0,  r0,  r2,  lsl #3

  480.   .endif

  481. 

  482.         vst1.8          {d12},     [r0,:64], r2

  483.         vst1.8          {d13},     [r0,:64], r2

  484.         vst1.8          {d14},     [r0,:64], r2

  485.         vst1.8          {d15},     [r0,:64], r2

  486.         vst1.8          {d8},      [r0,:64], r2

  487.         vst1.8          {d9},      [r0,:64], r2

  488.         vst1.8          {d10},     [r0,:64], r2

  489.         vst1.8          {d11},     [r0,:64], r2

  490. 

  491.         mov             lr,  r10

  492.         bx              lr

  493. endfunc

  494. .endm

  495. 

  496.         h264_qpel8_hv_lowpass put

  497.         h264_qpel8_hv_lowpass avg

  498. 

  499. .macro  h264_qpel8_hv_lowpass_l2 type

  500. function \type\()_h264_qpel8_hv_lowpass_l2_neon

  501.         mov             r10, lr

  502.         bl              put_h264_qpel8_hv_lowpass_neon_top

  503. 

  504.         vld1.8          {d0, d1},  [r2,:128]!

  505.         vld1.8          {d2, d3},  [r2,:128]!

  506.         vrhadd.u8       q0,  q0,  q6

  507.         vld1.8          {d4, d5},  [r2,:128]!

  508.         vrhadd.u8       q1,  q1,  q7

  509.         vld1.8          {d6, d7},  [r2,:128]!

  510.         vrhadd.u8       q2,  q2,  q4

  511.         vrhadd.u8       q3,  q3,  q5

  512.   .ifc \type,avg

  513.         vld1.8          {d16},     [r0,:64], r3

  514.         vrhadd.u8       d0,  d0,  d16

  515.         vld1.8          {d17},     [r0,:64], r3

  516.         vrhadd.u8       d1,  d1,  d17

  517.         vld1.8          {d18},     [r0,:64], r3

  518.         vrhadd.u8       d2,  d2,  d18

  519.         vld1.8          {d19},     [r0,:64], r3

  520.         vrhadd.u8       d3,  d3,  d19

  521.         vld1.8          {d20},     [r0,:64], r3

  522.         vrhadd.u8       d4,  d4,  d20

  523.         vld1.8          {d21},     [r0,:64], r3

  524.         vrhadd.u8       d5,  d5,  d21

  525.         vld1.8          {d22},     [r0,:64], r3

  526.         vrhadd.u8       d6,  d6,  d22

  527.         vld1.8          {d23},     [r0,:64], r3

  528.         vrhadd.u8       d7,  d7,  d23

  529.         sub             r0,  r0,  r3,  lsl #3

  530.   .endif

  531.         vst1.8          {d0},      [r0,:64], r3

  532.         vst1.8          {d1},      [r0,:64], r3

  533.         vst1.8          {d2},      [r0,:64], r3

  534.         vst1.8          {d3},      [r0,:64], r3

  535.         vst1.8          {d4},      [r0,:64], r3

  536.         vst1.8          {d5},      [r0,:64], r3

  537.         vst1.8          {d6},      [r0,:64], r3

  538.         vst1.8          {d7},      [r0,:64], r3

  539. 

  540.         mov             lr,  r10

  541.         bx              lr

  542. endfunc

  543. .endm

  544. 

  545.         h264_qpel8_hv_lowpass_l2 put

  546.         h264_qpel8_hv_lowpass_l2 avg

  547. 

  548. .macro  h264_qpel16_hv  type

  549. function \type\()_h264_qpel16_hv_lowpass_neon

  550.         mov             r9,  lr

  551.         bl              \type\()_h264_qpel8_hv_lowpass_neon

  552.         sub             r1,  r1,  r3, lsl #2

  553.         bl              \type\()_h264_qpel8_hv_lowpass_neon

  554.         sub             r1,  r1,  r3, lsl #4

  555.         sub             r1,  r1,  r3, lsl #2

  556.         add             r1,  r1,  #8

  557.         sub             r0,  r0,  r2, lsl #4

  558.         add             r0,  r0,  #8

  559.         bl              \type\()_h264_qpel8_hv_lowpass_neon

  560.         sub             r1,  r1,  r3, lsl #2

  561.         mov             lr,  r9

  562.         b               \type\()_h264_qpel8_hv_lowpass_neon

  563. endfunc

  564. 

  565. function \type\()_h264_qpel16_hv_lowpass_l2_neon

  566.         mov             r9,  lr

  567.         sub             r2,  r4,  #256

  568.         bl              \type\()_h264_qpel8_hv_lowpass_l2_neon

  569.         sub             r1,  r1,  r3, lsl #2

  570.         bl              \type\()_h264_qpel8_hv_lowpass_l2_neon

  571.         sub             r1,  r1,  r3, lsl #4

  572.         sub             r1,  r1,  r3, lsl #2

  573.         add             r1,  r1,  #8

  574.         sub             r0,  r0,  r3, lsl #4

  575.         add             r0,  r0,  #8

  576.         bl              \type\()_h264_qpel8_hv_lowpass_l2_neon

  577.         sub             r1,  r1,  r3, lsl #2

  578.         mov             lr,  r9

  579.         b               \type\()_h264_qpel8_hv_lowpass_l2_neon

  580. endfunc

  581. .endm

  582. 

  583.         h264_qpel16_hv put

  584.         h264_qpel16_hv avg

  585. 

  586. .macro  h264_qpel8      type

  587. function ff_\type\()_h264_qpel8_mc10_neon, export=1

  588.         lowpass_const   r3

  589.         mov             r3,  r1

  590.         sub             r1,  r1,  #2

  591.         mov             r12, #8

  592.         b               \type\()_h264_qpel8_h_lowpass_l2_neon

  593. endfunc

  594. 

  595. function ff_\type\()_h264_qpel8_mc20_neon, export=1

  596.         lowpass_const   r3

  597.         sub             r1,  r1,  #2

  598.         mov             r3,  r2

  599.         mov             r12, #8

  600.         b               \type\()_h264_qpel8_h_lowpass_neon

  601. endfunc

  602. 

  603. function ff_\type\()_h264_qpel8_mc30_neon, export=1

  604.         lowpass_const   r3

  605.         add             r3,  r1,  #1

  606.         sub             r1,  r1,  #2

  607.         mov             r12, #8

  608.         b               \type\()_h264_qpel8_h_lowpass_l2_neon

  609. endfunc

  610. 

  611. function ff_\type\()_h264_qpel8_mc01_neon, export=1

  612.         push            {lr}

  613.         mov             r12, r1

  614. \type\()_h264_qpel8_mc01:

  615.         lowpass_const   r3

  616.         mov             r3,  r2

  617.         sub             r1,  r1,  r2, lsl #1

  618.         vpush           {d8-d15}

  619.         bl              \type\()_h264_qpel8_v_lowpass_l2_neon

  620.         vpop            {d8-d15}

  621.         pop             {pc}

  622. endfunc

  623. 

  624. function ff_\type\()_h264_qpel8_mc11_neon, export=1

  625.         push            {r0, r1, r11, lr}

  626. \type\()_h264_qpel8_mc11:

  627.         lowpass_const   r3

  628.         mov             r11, sp

  629. A       bic             sp,  sp,  #15

  630. T       bic             r0,  r11, #15

  631. T       mov             sp,  r0

  632.         sub             sp,  sp,  #64

  633.         mov             r0,  sp

  634.         sub             r1,  r1,  #2

  635.         mov             r3,  #8

  636.         mov             r12, #8

  637.         vpush           {d8-d15}

  638.         bl              put_h264_qpel8_h_lowpass_neon

  639.         ldrd            r0,  r1,  [r11], #8

  640.         mov             r3,  r2

  641.         add             r12, sp,  #64

  642.         sub             r1,  r1,  r2, lsl #1

  643.         mov             r2,  #8

  644.         bl              \type\()_h264_qpel8_v_lowpass_l2_neon

  645.         vpop            {d8-d15}

  646.         mov             sp,  r11

  647.         pop             {r11, pc}

  648. endfunc

  649. 

  650. function ff_\type\()_h264_qpel8_mc21_neon, export=1

  651.         push            {r0, r1, r4, r10, r11, lr}

  652. \type\()_h264_qpel8_mc21:

  653.         lowpass_const   r3

  654.         mov             r11, sp

  655. A       bic             sp,  sp,  #15

  656. T       bic             r0,  r11, #15

  657. T       mov             sp,  r0

  658.         sub             sp,  sp,  #(8*8+16*12)

  659.         sub             r1,  r1,  #2

  660.         mov             r3,  #8

  661.         mov             r0,  sp

  662.         mov             r12, #8

  663.         vpush           {d8-d15}

  664.         bl              put_h264_qpel8_h_lowpass_neon

  665.         mov             r4,  r0

  666.         ldrd            r0,  r1,  [r11], #8

  667.         sub             r1,  r1,  r2, lsl #1

  668.         sub             r1,  r1,  #2

  669.         mov             r3,  r2

  670.         sub             r2,  r4,  #64

  671.         bl              \type\()_h264_qpel8_hv_lowpass_l2_neon

  672.         vpop            {d8-d15}

  673.         mov             sp,  r11

  674.         pop             {r4, r10, r11, pc}

  675. endfunc

  676. 

  677. function ff_\type\()_h264_qpel8_mc31_neon, export=1

  678.         add             r1,  r1,  #1

  679.         push            {r0, r1, r11, lr}

  680.         sub             r1,  r1,  #1

  681.         b               \type\()_h264_qpel8_mc11

  682. endfunc

  683. 

  684. function ff_\type\()_h264_qpel8_mc02_neon, export=1

  685.         push            {lr}

  686.         lowpass_const   r3

  687.         sub             r1,  r1,  r2, lsl #1

  688.         mov             r3,  r2

  689.         vpush           {d8-d15}

  690.         bl              \type\()_h264_qpel8_v_lowpass_neon

  691.         vpop            {d8-d15}

  692.         pop             {pc}

  693. endfunc

  694. 

  695. function ff_\type\()_h264_qpel8_mc12_neon, export=1

  696.         push            {r0, r1, r4, r10, r11, lr}

  697. \type\()_h264_qpel8_mc12:

  698.         lowpass_const   r3

  699.         mov             r11, sp

  700. A       bic             sp,  sp,  #15

  701. T       bic             r0,  r11, #15

  702. T       mov             sp,  r0

  703.         sub             sp,  sp,  #(8*8+16*12)

  704.         sub             r1,  r1,  r2, lsl #1

  705.         mov             r3,  r2

  706.         mov             r2,  #8

  707.         mov             r0,  sp

  708.         vpush           {d8-d15}

  709.         bl              put_h264_qpel8_v_lowpass_neon

  710.         mov             r4,  r0

  711.         ldrd            r0,  r1,  [r11], #8

  712.         sub             r1,  r1,  r3, lsl #1

  713.         sub             r1,  r1,  #2

  714.         sub             r2,  r4,  #64

  715.         bl              \type\()_h264_qpel8_hv_lowpass_l2_neon

  716.         vpop            {d8-d15}

  717.         mov             sp,  r11

  718.         pop             {r4, r10, r11, pc}

  719. endfunc

  720. 

  721. function ff_\type\()_h264_qpel8_mc22_neon, export=1

  722.         push            {r4, r10, r11, lr}

  723.         mov             r11, sp

  724. A       bic             sp,  sp,  #15

  725. T       bic             r4,  r11, #15

  726. T       mov             sp,  r4

  727.         sub             r1,  r1,  r2, lsl #1

  728.         sub             r1,  r1,  #2

  729.         mov             r3,  r2

  730.         sub             sp,  sp,  #(16*12)

  731.         mov             r4,  sp

  732.         vpush           {d8-d15}

  733.         bl              \type\()_h264_qpel8_hv_lowpass_neon

  734.         vpop            {d8-d15}

  735.         mov             sp,  r11

  736.         pop             {r4, r10, r11, pc}

  737. endfunc

  738. 

  739. function ff_\type\()_h264_qpel8_mc32_neon, export=1

  740.         push            {r0, r1, r4, r10, r11, lr}

  741.         add             r1,  r1,  #1

  742.         b               \type\()_h264_qpel8_mc12

  743. endfunc

  744. 

  745. function ff_\type\()_h264_qpel8_mc03_neon, export=1

  746.         push            {lr}

  747.         add             r12, r1,  r2

  748.         b               \type\()_h264_qpel8_mc01

  749. endfunc

  750. 

  751. function ff_\type\()_h264_qpel8_mc13_neon, export=1

  752.         push            {r0, r1, r11, lr}

  753.         add             r1,  r1,  r2

  754.         b               \type\()_h264_qpel8_mc11

  755. endfunc

  756. 

  757. function ff_\type\()_h264_qpel8_mc23_neon, export=1

  758.         push            {r0, r1, r4, r10, r11, lr}

  759.         add             r1,  r1,  r2

  760.         b               \type\()_h264_qpel8_mc21

  761. endfunc

  762. 

  763. function ff_\type\()_h264_qpel8_mc33_neon, export=1

  764.         add             r1,  r1,  #1

  765.         push            {r0, r1, r11, lr}

  766.         add             r1,  r1,  r2

  767.         sub             r1,  r1,  #1

  768.         b               \type\()_h264_qpel8_mc11

  769. endfunc

  770. .endm

  771. 

  772.         h264_qpel8 put

  773.         h264_qpel8 avg

  774. 

  775. .macro  h264_qpel16     type

  776. function ff_\type\()_h264_qpel16_mc10_neon, export=1

  777.         lowpass_const   r3

  778.         mov             r3,  r1

  779.         sub             r1,  r1,  #2

  780.         b               \type\()_h264_qpel16_h_lowpass_l2_neon

  781. endfunc

  782. 

  783. function ff_\type\()_h264_qpel16_mc20_neon, export=1

  784.         lowpass_const   r3

  785.         sub             r1,  r1,  #2

  786.         mov             r3,  r2

  787.         b               \type\()_h264_qpel16_h_lowpass_neon

  788. endfunc

  789. 

  790. function ff_\type\()_h264_qpel16_mc30_neon, export=1

  791.         lowpass_const   r3

  792.         add             r3,  r1,  #1

  793.         sub             r1,  r1,  #2

  794.         b               \type\()_h264_qpel16_h_lowpass_l2_neon

  795. endfunc

  796. 

  797. function ff_\type\()_h264_qpel16_mc01_neon, export=1

  798.         push            {r4, lr}

  799.         mov             r12, r1

  800. \type\()_h264_qpel16_mc01:

  801.         lowpass_const   r3

  802.         mov             r3,  r2

  803.         sub             r1,  r1,  r2, lsl #1

  804.         vpush           {d8-d15}

  805.         bl              \type\()_h264_qpel16_v_lowpass_l2_neon

  806.         vpop            {d8-d15}

  807.         pop             {r4, pc}

  808. endfunc

  809. 

  810. function ff_\type\()_h264_qpel16_mc11_neon, export=1

  811.         push            {r0, r1, r4, r11, lr}

  812. \type\()_h264_qpel16_mc11:

  813.         lowpass_const   r3

  814.         mov             r11, sp

  815. A       bic             sp,  sp,  #15

  816. T       bic             r0,  r11, #15

  817. T       mov             sp,  r0

  818.         sub             sp,  sp,  #256

  819.         mov             r0,  sp

  820.         sub             r1,  r1,  #2

  821.         mov             r3,  #16

  822.         vpush           {d8-d15}

  823.         bl              put_h264_qpel16_h_lowpass_neon

  824.         ldrd            r0,  r1,  [r11], #8

  825.         mov             r3,  r2

  826.         add             r12, sp,  #64

  827.         sub             r1,  r1,  r2, lsl #1

  828.         mov             r2,  #16

  829.         bl              \type\()_h264_qpel16_v_lowpass_l2_neon

  830.         vpop            {d8-d15}

  831.         mov             sp,  r11

  832.         pop             {r4, r11, pc}

  833. endfunc

  834. 

  835. function ff_\type\()_h264_qpel16_mc21_neon, export=1

  836.         push            {r0, r1, r4-r5, r9-r11, lr}

  837. \type\()_h264_qpel16_mc21:

  838.         lowpass_const   r3

  839.         mov             r11, sp

  840. A       bic             sp,  sp,  #15

  841. T       bic             r0,  r11, #15

  842. T       mov             sp,  r0

  843.         sub             sp,  sp,  #(16*16+16*12)

  844.         sub             r1,  r1,  #2

  845.         mov             r0,  sp

  846.         vpush           {d8-d15}

  847.         bl              put_h264_qpel16_h_lowpass_neon_packed

  848.         mov             r4,  r0

  849.         ldrd            r0,  r1,  [r11], #8

  850.         sub             r1,  r1,  r2, lsl #1

  851.         sub             r1,  r1,  #2

  852.         mov             r3,  r2

  853.         bl              \type\()_h264_qpel16_hv_lowpass_l2_neon

  854.         vpop            {d8-d15}

  855.         mov             sp,  r11

  856.         pop             {r4-r5, r9-r11, pc}

  857. endfunc

  858. 

  859. function ff_\type\()_h264_qpel16_mc31_neon, export=1

  860.         add             r1,  r1,  #1

  861.         push            {r0, r1, r4, r11, lr}

  862.         sub             r1,  r1,  #1

  863.         b               \type\()_h264_qpel16_mc11

  864. endfunc

  865. 

  866. function ff_\type\()_h264_qpel16_mc02_neon, export=1

  867.         push            {r4, lr}

  868.         lowpass_const   r3

  869.         sub             r1,  r1,  r2, lsl #1

  870.         mov             r3,  r2

  871.         vpush           {d8-d15}

  872.         bl              \type\()_h264_qpel16_v_lowpass_neon

  873.         vpop            {d8-d15}

  874.         pop             {r4, pc}

  875. endfunc

  876. 

  877. function ff_\type\()_h264_qpel16_mc12_neon, export=1

  878.         push            {r0, r1, r4-r5, r9-r11, lr}

  879. \type\()_h264_qpel16_mc12:

  880.         lowpass_const   r3

  881.         mov             r11, sp

  882. A       bic             sp,  sp,  #15

  883. T       bic             r0,  r11, #15

  884. T       mov             sp,  r0

  885.         sub             sp,  sp,  #(16*16+16*12)

  886.         sub             r1,  r1,  r2, lsl #1

  887.         mov             r0,  sp

  888.         mov             r3,  r2

  889.         vpush           {d8-d15}

  890.         bl              put_h264_qpel16_v_lowpass_neon_packed

  891.         mov             r4,  r0

  892.         ldrd            r0,  r1,  [r11], #8

  893.         sub             r1,  r1,  r3, lsl #1

  894.         sub             r1,  r1,  #2

  895.         mov             r2,  r3

  896.         bl              \type\()_h264_qpel16_hv_lowpass_l2_neon

  897.         vpop            {d8-d15}

  898.         mov             sp,  r11

  899.         pop             {r4-r5, r9-r11, pc}

  900. endfunc

  901. 

  902. function ff_\type\()_h264_qpel16_mc22_neon, export=1

  903.         push            {r4, r9-r11, lr}

  904.         lowpass_const   r3

  905.         mov             r11, sp

  906. A       bic             sp,  sp,  #15

  907. T       bic             r4,  r11, #15

  908. T       mov             sp,  r4

  909.         sub             r1,  r1,  r2, lsl #1

  910.         sub             r1,  r1,  #2

  911.         mov             r3,  r2

  912.         sub             sp,  sp,  #(16*12)

  913.         mov             r4,  sp

  914.         vpush           {d8-d15}

  915.         bl              \type\()_h264_qpel16_hv_lowpass_neon

  916.         vpop            {d8-d15}

  917.         mov             sp,  r11

  918.         pop             {r4, r9-r11, pc}

  919. endfunc

  920. 

  921. function ff_\type\()_h264_qpel16_mc32_neon, export=1

  922.         push            {r0, r1, r4-r5, r9-r11, lr}

  923.         add             r1,  r1,  #1

  924.         b               \type\()_h264_qpel16_mc12

  925. endfunc

  926. 

  927. function ff_\type\()_h264_qpel16_mc03_neon, export=1

  928.         push            {r4, lr}

  929.         add             r12, r1,  r2

  930.         b               \type\()_h264_qpel16_mc01

  931. endfunc

  932. 

  933. function ff_\type\()_h264_qpel16_mc13_neon, export=1

  934.         push            {r0, r1, r4, r11, lr}

  935.         add             r1,  r1,  r2

  936.         b               \type\()_h264_qpel16_mc11

  937. endfunc

  938. 

  939. function ff_\type\()_h264_qpel16_mc23_neon, export=1

  940.         push            {r0, r1, r4-r5, r9-r11, lr}

  941.         add             r1,  r1,  r2

  942.         b               \type\()_h264_qpel16_mc21

  943. endfunc

  944. 

  945. function ff_\type\()_h264_qpel16_mc33_neon, export=1

  946.         add             r1,  r1,  #1

  947.         push            {r0, r1, r4, r11, lr}

  948.         add             r1,  r1,  r2

  949.         sub             r1,  r1,  #1

  950.         b               \type\()_h264_qpel16_mc11

  951. endfunc

  952. .endm

  953. 

  954.         h264_qpel16 put

  955.         h264_qpel16 avg