sbrdsp_neon.S 13.9 KB
Newer Older
1 2 3
/*
 * Copyright (c) 2012 Mans Rullgard
 *
4
 * This file is part of FFmpeg.
5
 *
6
 * FFmpeg is free software; you can redistribute it and/or
7 8 9 10
 * 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.
 *
11
 * FFmpeg is distributed in the hope that it will be useful,
12 13 14 15 16
 * 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
17
 * License along with FFmpeg; if not, write to the Free Software
18 19 20
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

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

function ff_sbr_sum64x5_neon, export=1
        push            {lr}
        add             r1,  r0,  # 64*4
        add             r2,  r0,  #128*4
        add             r3,  r0,  #192*4
        add             lr,  r0,  #256*4
        mov             r12, #64
1:
        vld1.32         {q0},     [r0,:128]
        vld1.32         {q1},     [r1,:128]!
        vadd.f32        q0,  q0,  q1
        vld1.32         {q2},     [r2,:128]!
        vadd.f32        q0,  q0,  q2
        vld1.32         {q3},     [r3,:128]!
        vadd.f32        q0,  q0,  q3
        vld1.32         {q8},     [lr,:128]!
        vadd.f32        q0,  q0,  q8
        vst1.32         {q0},     [r0,:128]!
        subs            r12, #4
        bgt             1b
        pop             {pc}
endfunc

function ff_sbr_sum_square_neon, export=1
        vmov.f32        q0,  #0.0
1:
        vld1.32         {q1},     [r0,:128]!
        vmla.f32        q0,  q1,  q1
        subs            r1,  r1,  #2
        bgt             1b
        vadd.f32        d0,  d0,  d1
        vpadd.f32       d0,  d0,  d0
NOVFP   vmov.32         r0,  d0[0]
        bx              lr
endfunc

function ff_sbr_neg_odd_64_neon, export=1
        mov             r1,  r0
        vmov.i32        q8,  #1<<31
        vld2.32         {q0,q1},  [r0,:128]!
        veor            q1,  q1,  q8
        vld2.32         {q2,q3},  [r0,:128]!
    .rept 3
        vst2.32         {q0,q1},  [r1,:128]!
        veor            q3,  q3,  q8
        vld2.32         {q0,q1},  [r0,:128]!
        vst2.32         {q2,q3},  [r1,:128]!
        veor            q1,  q1,  q8
        vld2.32         {q2,q3},  [r0,:128]!
    .endr
        veor            q3,  q3,  q8
        vst2.32         {q0,q1},  [r1,:128]!
        vst2.32         {q2,q3},  [r1,:128]!
        bx              lr
endfunc

function ff_sbr_qmf_pre_shuffle_neon, export=1
        add             r1,  r0,  #60*4
        add             r2,  r0,  #64*4
        vld1.32         {d0},     [r0,:64]!
        vst1.32         {d0},     [r2,:64]!
        mov             r3,  #-16
        mov             r12, #24
        vmov.i32        q8,  #1<<31
        vld1.32         {q0},     [r1,:128], r3
        vld1.32         {d2},     [r0,:64]!
1:
        vld1.32         {d3,d4},  [r0,:128]!
        vrev64.32       q0,  q0
        vld1.32         {q9},     [r1,:128], r3
        veor            q0,  q0,  q8
        vld1.32         {d5,d6},  [r0,:128]!
        vswp            d0,  d1
        vrev64.32       q9,  q9
        vst2.32         {q0,q1},  [r2,:64]!
        vmov            q10, q2
        veor            q9,  q9,  q8
        vmov            d2,  d6
        vswp            d18, d19
        vld1.32         {q0},     [r1,:128], r3
        vst2.32         {q9,q10}, [r2,:64]!
        subs            r12, r12, #8
        bgt             1b
        vld1.32         {d3,d4},  [r0,:128]!
        vrev64.32       q0,  q0
        vld1.32         {q9},     [r1,:128], r3
        veor            q0,  q0,  q8
        vld1.32         {d5},     [r0,:64]!
        vswp            d0,  d1
        vrev64.32       q9,  q9
        vst2.32         {q0,q1},  [r2,:64]!
        vswp            d4,  d5
        veor            q1,  q9,  q8
        vst2.32         {d3,d5},  [r2,:64]!
        vst2.32         {d2[0],d4[0]}, [r2,:64]!
        bx              lr
endfunc

function ff_sbr_qmf_post_shuffle_neon, export=1
        add             r2,  r1,  #60*4
        mov             r3,  #-16
        mov             r12, #32
        vmov.i32        q8,  #1<<31
        vld1.32         {q0},     [r2,:128], r3
        vld1.32         {q1},     [r1,:128]!
1:
        pld             [r2, #-32]
        vrev64.32       q0,  q0
        vswp            d2,  d3
        veor            q0,  q0,  q8
        vld1.32         {q2},     [r2,:128], r3
        vld1.32         {q3},     [r1,:128]!
        vst2.32         {d1,d3},  [r0,:128]!
        vst2.32         {d0,d2},  [r0,:128]!
        pld             [r2, #-32]
        vrev64.32       q2,  q2
        vswp            d6,  d7
        veor            q2,  q2,  q8
        vld1.32         {q0},     [r2,:128], r3
        vld1.32         {q1},     [r1,:128]!
        vst2.32         {d5,d7},  [r0,:128]!
        vst2.32         {d4,d6},  [r0,:128]!
        subs            r12, r12, #8
        bgt             1b
        bx              lr
endfunc

function ff_sbr_qmf_deint_neg_neon, export=1
        add             r1,  r1,  #60*4
        add             r2,  r0,  #62*4
        mov             r3,  #-16
        mov             r12, #32
        vmov.i32        d2,  #1<<31
1:
        vld2.32         {d0,d1},  [r1,:128], r3
        veor            d0,  d0,  d2
        vrev64.32       d1,  d1
        vst1.32         {d0},     [r2,:64]
        vst1.32         {d1},     [r0,:64]!
        sub             r2,  r2,  #8
        subs            r12, r12, #2
        bgt             1b
        bx              lr
endfunc

function ff_sbr_qmf_deint_bfly_neon, export=1
        push            {lr}
        add             r2,  r2,  #60*4
        add             r3,  r0,  #124*4
        mov             r12, #64
        mov             lr,  #-16
1:
        vld1.32         {q0},     [r1,:128]!
        vld1.32         {q1},     [r2,:128], lr
        vrev64.32       q2,  q0
        vrev64.32       q3,  q1
        vadd.f32        d3,  d4,  d3
        vadd.f32        d2,  d5,  d2
        vsub.f32        d0,  d0,  d7
        vsub.f32        d1,  d1,  d6
        vst1.32         {q1},     [r3,:128], lr
        vst1.32         {q0},     [r0,:128]!
        subs            r12, r12, #4
        bgt             1b
        pop             {pc}
endfunc

function ff_sbr_hf_g_filt_neon, export=1
        ldr             r12, [sp]
        add             r1,  r1,  r12, lsl #3
        mov             r12, #40*2*4
        sub             r3,  r3,  #1
        vld2.32         {d2[],d3[]},[r2,:64]!
        vld1.32         {d0},     [r1,:64], r12
1:
        vld1.32         {d1},     [r1,:64], r12
        vmul.f32        q3,  q0,  q1
        vld2.32         {d2[],d3[]},[r2,:64]!
        vld1.32         {d0},     [r1,:64], r12
        vst1.32         {q3},     [r0,:64]!
        subs            r3,  r3,  #2
        bgt             1b
        it              lt
        bxlt            lr
        vmul.f32        d0,  d0,  d2
        vst1.32         {d0},     [r0,:64]!
        bx              lr
endfunc

function ff_sbr_hf_gen_neon, export=1
NOVFP   vld1.32         {d1[]},   [sp,:32]
VFP     vdup.32         d1,  d0[0]
        vmul.f32        d0,  d1,  d1
        vld1.32         {d3},     [r2,:64]
        vld1.32         {d2},     [r3,:64]
        vmul.f32        q0,  q0,  q1
        ldrd            r2,  r3,  [sp, #4*!HAVE_VFP_ARGS]
        vtrn.32         d0,  d1
        vneg.f32        d18, d1
        vtrn.32         d18, d1
        add             r0,  r0,  r2,  lsl #3
        add             r1,  r1,  r2,  lsl #3
        sub             r1,  r1,  #2*8
        sub             r3,  r3,  r2
        vld1.32         {q1},     [r1,:128]!
1:
        vld1.32         {q3},     [r1,:128]!
        vrev64.32       q2,  q1
        vmov            q8,  q3
        vrev64.32       d20, d3
        vrev64.32       d21, d6
        vmla.f32        q3,  q1,  d0[0]
        vmla.f32        d6,  d4,  d18
        vmla.f32        d7,  d20, d18
        vmla.f32        d6,  d3,  d0[1]
        vmla.f32        d7,  d16, d0[1]
        vmla.f32        d6,  d5,  d1
        vmla.f32        d7,  d21, d1
        vmov            q1,  q8
        vst1.32         {q3},     [r0,:128]!
        subs            r3,  r3,  #2
        bgt             1b
        bx              lr
endfunc

function ff_sbr_autocorrelate_neon, export=1
        vld1.32         {q0},     [r0,:128]!
        vmov.f32        q1,  #0.0
        vmov.f32        q3,  #0.0
        vmov.f32        d20, #0.0
        vmul.f32        d21, d1,  d1
        vmov            q8,  q0
        vmov            q11, q0
        mov             r12, #36
1:
        vld1.32         {q2},     [r0,:128]!
        vrev64.32       q12, q2
        vmla.f32        q10, q2,  q2
        vmla.f32        d2,  d1,  d4
        vmla.f32        d3,  d1,  d24
        vmla.f32        d6,  d0,  d4
        vmla.f32        d7,  d0,  d24
        vmla.f32        d2,  d4,  d5
        vmla.f32        d3,  d4,  d25
        vmla.f32        d6,  d1,  d5
        vmla.f32        d7,  d1,  d25
        vmov            q0,  q2
        subs            r12, r12, #2
        bgt             1b
        vld1.32         {q2},     [r0,:128]!
        vrev64.32       q12, q2
        vmla.f32        d2,  d1,  d4
        vmla.f32        d3,  d1,  d24
        vmla.f32        d6,  d0,  d4
        vmla.f32        d7,  d0,  d24
        vadd.f32        d20, d20, d21
        vrev64.32       d18, d17
        vmla.f32        d6,  d1,  d5
        vmla.f32        d7,  d1,  d25
        vmov            q0,  q1
        vmla.f32        d0,  d16, d17
        vmla.f32        d1,  d16, d18
        vmla.f32        d2,  d4,  d5
        vmla.f32        d3,  d4,  d25
        vneg.f32        s15, s15
        vmov            d21, d20
        vpadd.f32       d0,  d0,  d2
        vpadd.f32       d7,  d6,  d7
        vtrn.32         d1,  d3
        vsub.f32        d6,  d1,  d3
        vmla.f32        d20, d22, d22
        vmla.f32        d21, d4,  d4
        vtrn.32         d0,  d6
        vpadd.f32       d20, d20, d21
        vst1.32         {q3},     [r1,:128]!
        vst1.32         {d20[1]}, [r1,:32]
        add             r1,  r1,  #2*4
        vst1.32         {d0},     [r1,:64]
        add             r1,  r1,  #4*4
        vst1.32         {d20[0]}, [r1,:32]
        bx              lr
endfunc

function ff_sbr_hf_apply_noise_0_neon, export=1
        vmov.i32        d3,  #0
.Lhf_apply_noise_0:
        push            {r4,lr}
310
        movrelx         r4,  X(ff_sbr_noise_table)
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
        ldr             r12, [sp, #12]
        add             r3,  r3,  #1
        bfc             r3,  #9,  #23
        sub             r12, r12, #1
1:
        add             lr,  r4,  r3,  lsl #3
        vld2.32         {q0},     [r0,:64]
        vld2.32         {q3},     [lr,:64]
        vld1.32         {d2},     [r1,:64]!
        vld1.32         {d18},    [r2,:64]!
        vceq.f32        d16, d2,  #0
        veor            d2,  d2,  d3
        vmov            q2,  q0
        vmla.f32        d0,  d6,  d18
        vmla.f32        d1,  d7,  d18
        vadd.f32        d4,  d4,  d2
        add             r3,  r3,  #2
        bfc             r3,  #9,  #23
        vbif            d0,  d4,  d16
        vbif            d1,  d5,  d16
        vst2.32         {q0},     [r0,:64]!
        subs            r12, r12, #2
        bgt             1b
        blt             2f
        add             lr,  r4,  r3,  lsl #3
        vld1.32         {d0},     [r0,:64]
        vld1.32         {d6},     [lr,:64]
        vld1.32         {d2[]},   [r1,:32]!
        vld1.32         {d3[]},   [r2,:32]!
        vceq.f32        d4,  d2,  #0
        veor            d2,  d2,  d3
        vmov            d1,  d0
        vmla.f32        d0,  d6,  d3
        vadd.f32        s2,  s2,  s4
        vbif            d0,  d1,  d4
        vst1.32         {d0},     [r0,:64]!
2:
        pop             {r4,pc}
endfunc

function ff_sbr_hf_apply_noise_1_neon, export=1
        ldr             r12, [sp]
        push            {r4,lr}
        lsl             r12, r12, #31
        eor             lr,  r12, #1<<31
        vmov            d3,  r12, lr
.Lhf_apply_noise_1:
358
        movrelx         r4,  X(ff_sbr_noise_table)
359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411
        ldr             r12, [sp, #12]
        add             r3,  r3,  #1
        bfc             r3,  #9,  #23
        sub             r12, r12, #1
1:
        add             lr,  r4,  r3,  lsl #3
        vld2.32         {q0},     [r0,:64]
        vld2.32         {q3},     [lr,:64]
        vld1.32         {d2},     [r1,:64]!
        vld1.32         {d18},    [r2,:64]!
        vceq.f32        d16, d2,  #0
        veor            d2,  d2,  d3
        vmov            q2,  q0
        vmla.f32        d0,  d6,  d18
        vmla.f32        d1,  d7,  d18
        vadd.f32        d5,  d5,  d2
        add             r3,  r3,  #2
        bfc             r3,  #9,  #23
        vbif            d0,  d4,  d16
        vbif            d1,  d5,  d16
        vst2.32         {q0},     [r0,:64]!
        subs            r12, r12, #2
        bgt             1b
        blt             2f
        add             lr,  r4,  r3,  lsl #3
        vld1.32         {d0},     [r0,:64]
        vld1.32         {d6},     [lr,:64]
        vld1.32         {d2[]},   [r1,:32]!
        vld1.32         {d18[]},  [r2,:32]!
        vceq.f32        d4,  d2,  #0
        veor            d2,  d2,  d3
        vmov            d1,  d0
        vmla.f32        d0,  d6,  d18
        vadd.f32        s3,  s3,  s5
        vbif            d0,  d1,  d4
        vst1.32         {d0},     [r0,:64]!
2:
        pop             {r4,pc}
endfunc

function ff_sbr_hf_apply_noise_2_neon, export=1
        vmov.i32        d3,  #1<<31
        b               .Lhf_apply_noise_0
endfunc

function ff_sbr_hf_apply_noise_3_neon, export=1
        ldr             r12, [sp]
        push            {r4,lr}
        lsl             r12, r12, #31
        eor             lr,  r12, #1<<31
        vmov            d3,  lr, r12
        b               .Lhf_apply_noise_1
endfunc