hpeldsp_neon.S 13.6 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397
/*
 * ARM NEON optimised DSP functions
 * Copyright (c) 2008 Mans Rullgard <mans@mansr.com>
 * Copyright (c) 2013 Janne Grunau <janne-libav@jannau.net>
 *
 * This file is part of Libav.
 *
 * Libav is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * Libav is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with Libav; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include "libavutil/aarch64/asm.S"

.macro  pixels16        rnd=1, avg=0
  .if \avg
        mov             x12, x0
  .endif
1:      ld1             {v0.16B},  [x1], x2
        ld1             {v1.16B},  [x1], x2
        ld1             {v2.16B},  [x1], x2
        ld1             {v3.16B},  [x1], x2
  .if \avg
        ld1             {v4.16B},  [x12], x2
        urhadd          v0.16B,  v0.16B,  v4.16B
        ld1             {v5.16B},  [x12], x2
        urhadd          v1.16B,  v1.16B,  v5.16B
        ld1             {v6.16B},  [x12], x2
        urhadd          v2.16B,  v2.16B,  v6.16B
        ld1             {v7.16B},  [x12], x2
        urhadd          v3.16B,  v3.16B,  v7.16B
  .endif
        subs            w3,  w3,  #4
        st1             {v0.16B},  [x0], x2
        st1             {v1.16B},  [x0], x2
        st1             {v2.16B},  [x0], x2
        st1             {v3.16B},  [x0], x2
        b.ne            1b
        ret
.endm

.macro  pixels16_x2     rnd=1, avg=0
1:      ld1             {v0.16B, v1.16B}, [x1], x2
        ld1             {v2.16B, v3.16B}, [x1], x2
        subs            w3,  w3,  #2
        ext             v1.16B,  v0.16B,  v1.16B,  #1
        avg             v0.16B,  v0.16B,  v1.16B
        ext             v3.16B,  v2.16B,  v3.16B,  #1
        avg             v2.16B,  v2.16B,  v3.16B
  .if \avg
        ld1             {v1.16B}, [x0], x2
        ld1             {v3.16B}, [x0]
        urhadd          v0.16B,  v0.16B,  v1.16B
        urhadd          v2.16B,  v2.16B,  v3.16B
        sub             x0,  x0,  x2
  .endif
        st1             {v0.16B}, [x0], x2
        st1             {v2.16B}, [x0], x2
        b.ne            1b
        ret
.endm

.macro  pixels16_y2     rnd=1, avg=0
        sub             w3,  w3,  #2
        ld1             {v0.16B}, [x1], x2
        ld1             {v1.16B}, [x1], x2
1:      subs            w3,  w3,  #2
        avg             v2.16B,  v0.16B,  v1.16B
        ld1             {v0.16B}, [x1], x2
        avg             v3.16B,  v0.16B,  v1.16B
        ld1             {v1.16B}, [x1], x2
  .if \avg
        ld1             {v4.16B}, [x0], x2
        ld1             {v5.16B}, [x0]
        urhadd          v2.16B,  v2.16B,  v4.16B
        urhadd          v3.16B,  v3.16B,  v5.16B
        sub             x0,  x0,  x2
  .endif
        st1             {v2.16B}, [x0], x2
        st1             {v3.16B}, [x0], x2
        b.ne            1b

        avg             v2.16B,  v0.16B,  v1.16B
        ld1             {v0.16B}, [x1], x2
        avg             v3.16B,  v0.16B,  v1.16B
  .if \avg
        ld1             {v4.16B}, [x0], x2
        ld1             {v5.16B}, [x0]
        urhadd          v2.16B,  v2.16B,  v4.16B
        urhadd          v3.16B,  v3.16B,  v5.16B
        sub             x0,  x0,  x2
  .endif
        st1             {v2.16B},     [x0], x2
        st1             {v3.16B},     [x0], x2

        ret
.endm

.macro  pixels16_xy2    rnd=1, avg=0
        sub             w3,  w3,  #2
        ld1             {v0.16B, v1.16B}, [x1], x2
        ld1             {v4.16B, v5.16B}, [x1], x2
NRND    movi            v26.8H, #1
        ext             v1.16B,  v0.16B,  v1.16B,  #1
        ext             v5.16B,  v4.16B,  v5.16B,  #1
        uaddl           v16.8H,  v0.8B,   v1.8B
        uaddl2          v20.8H,  v0.16B,  v1.16B
        uaddl           v18.8H,  v4.8B,   v5.8B
        uaddl2          v22.8H,  v4.16B,  v5.16B
1:      subs            w3,  w3,  #2
        ld1             {v0.16B, v1.16B}, [x1], x2
        add             v24.8H,  v16.8H,  v18.8H
NRND    add             v24.8H,  v24.8H,  v26.8H
        ext             v30.16B, v0.16B,  v1.16B,  #1
        add             v1.8H,   v20.8H,  v22.8H
        mshrn           v28.8B,  v24.8H,  #2
NRND    add             v1.8H,   v1.8H,   v26.8H
        mshrn2          v28.16B, v1.8H,   #2
  .if \avg
        ld1             {v16.16B},        [x0]
        urhadd          v28.16B, v28.16B, v16.16B
  .endif
        uaddl           v16.8H,  v0.8B,   v30.8B
        ld1             {v2.16B, v3.16B}, [x1], x2
        uaddl2          v20.8H,  v0.16B,  v30.16B
        st1             {v28.16B},        [x0], x2
        add             v24.8H,  v16.8H,  v18.8H
NRND    add             v24.8H,  v24.8H,  v26.8H
        ext             v3.16B,  v2.16B,  v3.16B,  #1
        add             v0.8H,   v20.8H,  v22.8H
        mshrn           v30.8B,  v24.8H,  #2
NRND    add             v0.8H,   v0.8H,   v26.8H
        mshrn2          v30.16B, v0.8H,   #2
  .if \avg
        ld1             {v18.16B},        [x0]
        urhadd          v30.16B, v30.16B, v18.16B
  .endif
        uaddl           v18.8H,   v2.8B,  v3.8B
        uaddl2          v22.8H,   v2.16B, v3.16B
        st1             {v30.16B},        [x0], x2
        b.gt            1b

        ld1             {v0.16B, v1.16B}, [x1], x2
        add             v24.8H,  v16.8H,  v18.8H
NRND    add             v24.8H,  v24.8H,  v26.8H
        ext             v30.16B, v0.16B,  v1.16B,  #1
        add             v1.8H,   v20.8H,  v22.8H
        mshrn           v28.8B,  v24.8H,  #2
NRND    add             v1.8H,   v1.8H,   v26.8H
        mshrn2          v28.16B, v1.8H,   #2
  .if \avg
        ld1             {v16.16B},        [x0]
        urhadd          v28.16B, v28.16B, v16.16B
  .endif
        uaddl           v16.8H,  v0.8B,   v30.8B
        uaddl2          v20.8H,  v0.16B,  v30.16B
        st1             {v28.16B},        [x0], x2
        add             v24.8H,  v16.8H,  v18.8H
NRND    add             v24.8H,  v24.8H,  v26.8H
        add             v0.8H,   v20.8H,  v22.8H
        mshrn           v30.8B,  v24.8H,  #2
NRND    add             v0.8H,   v0.8H,   v26.8H
        mshrn2          v30.16B, v0.8H,   #2
  .if \avg
        ld1             {v18.16B},        [x0]
        urhadd          v30.16B, v30.16B, v18.16B
  .endif
        st1             {v30.16B},        [x0], x2

        ret
.endm

.macro  pixels8         rnd=1, avg=0
1:      ld1             {v0.8B}, [x1], x2
        ld1             {v1.8B}, [x1], x2
        ld1             {v2.8B}, [x1], x2
        ld1             {v3.8B}, [x1], x2
  .if \avg
        ld1             {v4.8B}, [x0], x2
        urhadd          v0.8B,  v0.8B,  v4.8B
        ld1             {v5.8B}, [x0], x2
        urhadd          v1.8B,  v1.8B,  v5.8B
        ld1             {v6.8B}, [x0], x2
        urhadd          v2.8B,  v2.8B,  v6.8B
        ld1             {v7.8B}, [x0], x2
        urhadd          v3.8B,  v3.8B,  v7.8B
        sub             x0,  x0,  x2,  lsl #2
  .endif
        subs            w3,  w3,  #4
        st1             {v0.8B}, [x0], x2
        st1             {v1.8B}, [x0], x2
        st1             {v2.8B}, [x0], x2
        st1             {v3.8B}, [x0], x2
        b.ne            1b
        ret
.endm

.macro  pixels8_x2      rnd=1, avg=0
1:      ld1             {v0.8B, v1.8B}, [x1], x2
        ext             v1.8B,  v0.8B,  v1.8B,  #1
        ld1             {v2.8B, v3.8B}, [x1], x2
        ext             v3.8B,  v2.8B,  v3.8B,  #1
        subs            w3,  w3,  #2
        avg             v0.8B,   v0.8B,   v1.8B
        avg             v2.8B,   v2.8B,   v3.8B
  .if \avg
        ld1             {v4.8B},     [x0], x2
        ld1             {v5.8B},     [x0]
        urhadd          v0.8B,   v0.8B,   v4.8B
        urhadd          v2.8B,   v2.8B,   v5.8B
        sub             x0,  x0,  x2
  .endif
        st1             {v0.8B}, [x0], x2
        st1             {v2.8B}, [x0], x2
        b.ne            1b
        ret
.endm

.macro  pixels8_y2      rnd=1, avg=0
        sub             w3,  w3,  #2
        ld1             {v0.8B},  [x1], x2
        ld1             {v1.8B},  [x1], x2
1:      subs            w3,  w3,  #2
        avg             v4.8B,  v0.8B,  v1.8B
        ld1             {v0.8B},  [x1], x2
        avg             v5.8B,  v0.8B,  v1.8B
        ld1             {v1.8B},  [x1], x2
  .if \avg
        ld1             {v2.8B},     [x0], x2
        ld1             {v3.8B},     [x0]
        urhadd          v4.8B,  v4.8B,  v2.8B
        urhadd          v5.8B,  v5.8B,  v3.8B
        sub             x0,  x0,  x2
  .endif
        st1             {v4.8B},     [x0], x2
        st1             {v5.8B},     [x0], x2
        b.ne            1b

        avg             v4.8B,  v0.8B,  v1.8B
        ld1             {v0.8B},  [x1], x2
        avg             v5.8B,  v0.8B,  v1.8B
  .if \avg
        ld1             {v2.8B},     [x0], x2
        ld1             {v3.8B},     [x0]
        urhadd          v4.8B,  v4.8B,  v2.8B
        urhadd          v5.8B,  v5.8B,  v3.8B
        sub             x0,  x0,  x2
  .endif
        st1             {v4.8B},     [x0], x2
        st1             {v5.8B},     [x0], x2

        ret
.endm

.macro  pixels8_xy2     rnd=1, avg=0
        sub             w3,  w3,  #2
        ld1             {v0.16B},     [x1], x2
        ld1             {v1.16B},     [x1], x2
NRND    movi            v19.8H, #1
        ext             v4.16B,  v0.16B,  v4.16B,  #1
        ext             v6.16B,  v1.16B,  v6.16B,  #1
        uaddl           v16.8H,  v0.8B,  v4.8B
        uaddl           v17.8H,  v1.8B,  v6.8B
1:      subs            w3,  w3,  #2
        ld1             {v0.16B},     [x1], x2
        add             v18.8H, v16.8H,  v17.8H
        ext             v4.16B,  v0.16B,  v4.16B,  #1
NRND    add             v18.8H, v18.8H, v19.8H
        uaddl           v16.8H,  v0.8B,  v4.8B
        mshrn           v5.8B,  v18.8H, #2
        ld1             {v1.16B},     [x1], x2
        add             v18.8H, v16.8H,  v17.8H
  .if \avg
        ld1             {v7.8B},     [x0]
        urhadd          v5.8B,  v5.8B,  v7.8B
  .endif
NRND    add             v18.8H, v18.8H, v19.8H
        st1             {v5.8B},     [x0], x2
        mshrn           v7.8B,  v18.8H, #2
  .if \avg
        ld1             {v5.8B},     [x0]
        urhadd          v7.8B,  v7.8B,  v5.8B
  .endif
        ext             v6.16B,  v1.16B,  v6.16B,  #1
        uaddl           v17.8H,  v1.8B,   v6.8B
        st1             {v7.8B},     [x0], x2
        b.gt            1b

        ld1             {v0.16B},     [x1], x2
        add             v18.8H, v16.8H, v17.8H
        ext             v4.16B, v0.16B, v4.16B,  #1
NRND    add             v18.8H, v18.8H, v19.8H
        uaddl           v16.8H,  v0.8B, v4.8B
        mshrn           v5.8B,  v18.8H, #2
        add             v18.8H, v16.8H, v17.8H
  .if \avg
        ld1             {v7.8B},     [x0]
        urhadd          v5.8B,  v5.8B,  v7.8B
  .endif
NRND    add             v18.8H, v18.8H, v19.8H
        st1             {v5.8B},     [x0], x2
        mshrn           v7.8B,  v18.8H, #2
  .if \avg
        ld1             {v5.8B},     [x0]
        urhadd          v7.8B,  v7.8B,  v5.8B
  .endif
        st1             {v7.8B},     [x0], x2

        ret
.endm

.macro  pixfunc         pfx, name, suf, rnd=1, avg=0
  .if \rnd
    .macro avg  rd, rn, rm
        urhadd          \rd, \rn, \rm
    .endm
    .macro mshrn rd, rn, rm
        rshrn           \rd, \rn, \rm
    .endm
    .macro mshrn2 rd, rn, rm
        rshrn2          \rd, \rn, \rm
    .endm
    .macro NRND insn:vararg
    .endm
  .else
    .macro avg  rd, rn, rm
        uhadd           \rd, \rn, \rm
    .endm
    .macro mshrn rd, rn, rm
        shrn            \rd, \rn, \rm
    .endm
    .macro mshrn2 rd, rn, rm
        shrn2           \rd, \rn, \rm
    .endm
    .macro NRND insn:vararg
        \insn
    .endm
  .endif
function ff_\pfx\name\suf\()_neon, export=1
        \name           \rnd, \avg
endfunc
        .purgem         avg
        .purgem         mshrn
        .purgem         mshrn2
        .purgem         NRND
.endm

.macro  pixfunc2        pfx, name, avg=0
        pixfunc         \pfx, \name,          rnd=1, avg=\avg
        pixfunc         \pfx, \name, _no_rnd, rnd=0, avg=\avg
.endm

function ff_put_h264_qpel16_mc00_neon, export=1
        mov             w3,  #16
endfunc

        pixfunc         put_, pixels16,     avg=0
        pixfunc2        put_, pixels16_x2,  avg=0
        pixfunc2        put_, pixels16_y2,  avg=0
        pixfunc2        put_, pixels16_xy2, avg=0

function ff_avg_h264_qpel16_mc00_neon, export=1
        mov             w3,  #16
endfunc

        pixfunc         avg_, pixels16,     avg=1
        pixfunc2        avg_, pixels16_x2,  avg=1
        pixfunc2        avg_, pixels16_y2,  avg=1
        pixfunc2        avg_, pixels16_xy2, avg=1

function ff_put_h264_qpel8_mc00_neon, export=1
        mov             w3,  #8
endfunc

        pixfunc         put_, pixels8,     avg=0
        pixfunc2        put_, pixels8_x2,  avg=0
        pixfunc2        put_, pixels8_y2,  avg=0
        pixfunc2        put_, pixels8_xy2, avg=0

function ff_avg_h264_qpel8_mc00_neon, export=1
        mov             w3,  #8
endfunc

        pixfunc         avg_, pixels8,     avg=1
        pixfunc         avg_, pixels8_x2,  avg=1
        pixfunc         avg_, pixels8_y2,  avg=1
        pixfunc         avg_, pixels8_xy2, avg=1