float_dsp_neon.S 7.75 KB
Newer Older
1 2 3 4 5
/*
 * ARM NEON optimised Float DSP functions
 * Copyright (c) 2008 Mans Rullgard <mans@mansr.com>
 * Copyright (c) 2014 Janne Grunau <janne-libav@jannau.net>
 *
6
 * This file is part of FFmpeg.
7
 *
8
 * FFmpeg is free software; you can redistribute it and/or
9 10 11 12
 * 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.
 *
13
 * FFmpeg is distributed in the hope that it will be useful,
14 15 16 17 18
 * 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
19
 * License along with FFmpeg; if not, write to the Free Software
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
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include "config.h"
#include "asm.S"

function ff_vector_fmul_neon, export=1
1:      subs            w3,  w3,  #16
        ld1             {v0.4S, v1.4S}, [x1], #32
        ld1             {v2.4S, v3.4S}, [x1], #32
        ld1             {v4.4S, v5.4S}, [x2], #32
        ld1             {v6.4S, v7.4S}, [x2], #32
        fmul            v16.4S, v0.4S,  v4.4S
        fmul            v17.4S, v1.4S,  v5.4S
        fmul            v18.4S, v2.4S,  v6.4S
        fmul            v19.4S, v3.4S,  v7.4S
        st1             {v16.4S, v17.4S}, [x0], #32
        st1             {v18.4S, v19.4S}, [x0], #32
        b.ne            1b
        ret
endfunc

function ff_vector_fmac_scalar_neon, export=1
        mov             x3,  #-32
1:      subs            w2,  w2,  #16
        ld1             {v16.4S, v17.4S}, [x0], #32
        ld1             {v18.4S, v19.4S}, [x0], x3
        ld1             {v4.4S,  v5.4S},  [x1], #32
        ld1             {v6.4S,  v7.4S},  [x1], #32
        fmla            v16.4S, v4.4S,  v0.S[0]
        fmla            v17.4S, v5.4S,  v0.S[0]
        fmla            v18.4S, v6.4S,  v0.S[0]
        fmla            v19.4S, v7.4S,  v0.S[0]
        st1             {v16.4S, v17.4S}, [x0], #32
        st1             {v18.4S, v19.4S}, [x0], #32
        b.ne            1b
        ret
endfunc

function ff_vector_fmul_scalar_neon, export=1
        mov             w4,  #15
        bics            w3,  w2,  w4
        dup             v16.4S, v0.S[0]
        b.eq            3f
        ld1             {v0.4S, v1.4S}, [x1], #32
1:      subs            w3,  w3,  #16
        fmul            v0.4S,  v0.4S,  v16.4S
        ld1             {v2.4S, v3.4S}, [x1], #32
        fmul            v1.4S,  v1.4S,  v16.4S
        fmul            v2.4S,  v2.4S,  v16.4S
        st1             {v0.4S, v1.4S}, [x0], #32
        fmul            v3.4S,  v3.4S,  v16.4S
        b.eq            2f
        ld1             {v0.4S, v1.4S}, [x1], #32
        st1             {v2.4S, v3.4S}, [x0], #32
        b               1b
2:      ands            w2,  w2,  #15
        st1             {v2.4S, v3.4S}, [x0], #32
        b.eq            4f
3:      ld1             {v0.4S}, [x1], #16
        fmul            v0.4S,  v0.4S,  v16.4S
        st1             {v0.4S}, [x0], #16
        subs            w2,  w2,  #4
        b.gt            3b
4:      ret
endfunc

function ff_vector_dmul_scalar_neon, export=1
        dup             v16.2D, v0.D[0]
        ld1             {v0.2D, v1.2D}, [x1], #32
1:      subs            w2,  w2,  #8
        fmul            v0.2D,  v0.2D,  v16.2D
        ld1             {v2.2D, v3.2D}, [x1], #32
        fmul            v1.2D,  v1.2D,  v16.2D
        fmul            v2.2D,  v2.2D,  v16.2D
        st1             {v0.2D, v1.2D}, [x0], #32
        fmul            v3.2D,  v3.2D,  v16.2D
        ld1             {v0.2D, v1.2D}, [x1], #32
        st1             {v2.2D, v3.2D}, [x0], #32
        b.gt            1b
        ret
endfunc

function ff_vector_fmul_window_neon, export=1
        sxtw            x4,  w4                 // len
        sub             x2,  x2,  #8
        sub             x5,  x4,  #2
        add             x2,  x2,  x5, lsl #2    // src1 + 4 * (len - 4)
        add             x6,  x3,  x5, lsl #3    // win  + 8 * (len - 2)
        add             x5,  x0,  x5, lsl #3    // dst  + 8 * (len - 2)
        mov             x7,  #-16
        ld1             {v0.4S},  [x1], #16     // s0
        ld1             {v2.4S},  [x3], #16     // wi
        ld1             {v1.4S},  [x2], x7      // s1
1:      ld1             {v3.4S},  [x6], x7      // wj
        subs            x4,  x4,  #4
        fmul            v17.4S, v0.4S,  v2.4S   // s0 * wi
        rev64           v4.4S,  v1.4S
        rev64           v5.4S,  v3.4S
        rev64           v17.4S, v17.4S
        ext             v4.16B,  v4.16B,  v4.16B,  #8 // s1_r
        ext             v5.16B,  v5.16B,  v5.16B,  #8 // wj_r
        ext             v17.16B, v17.16B, v17.16B, #8 // (s0 * wi)_rev
        fmul            v16.4S, v0.4S,  v5.4S  // s0 * wj_r
        fmla            v17.4S, v1.4S,  v3.4S  // (s0 * wi)_rev + s1 * wj
        b.eq            2f
        ld1             {v0.4S},  [x1], #16
        fmls            v16.4S, v4.4S,  v2.4S  // s0 * wj_r - s1_r * wi
        st1             {v17.4S}, [x5], x7
        ld1             {v2.4S},  [x3], #16
        ld1             {v1.4S},  [x2], x7
        st1             {v16.4S}, [x0], #16
        b               1b
2:
        fmls            v16.4S, v4.4S,  v2.4S  // s0 * wj_r - s1_r * wi
        st1             {v17.4S}, [x5], x7
        st1             {v16.4S}, [x0], #16
        ret
endfunc

function ff_vector_fmul_add_neon, export=1
        ld1             {v0.4S, v1.4S},  [x1], #32
        ld1             {v2.4S, v3.4S},  [x2], #32
        ld1             {v4.4S, v5.4S},  [x3], #32
1:      subs            w4,  w4,  #8
        fmla            v4.4S,  v0.4S,  v2.4S
        fmla            v5.4S,  v1.4S,  v3.4S
        b.eq            2f
        ld1             {v0.4S, v1.4S},  [x1], #32
        ld1             {v2.4S, v3.4S},  [x2], #32
        st1             {v4.4S, v5.4S},  [x0], #32
        ld1             {v4.4S, v5.4S},  [x3], #32
        b               1b
2:      st1             {v4.4S, v5.4S},  [x0], #32
        ret
endfunc

function ff_vector_fmul_reverse_neon, export=1
        sxtw            x3,  w3
        add             x2,  x2,  x3,  lsl #2
        sub             x2,  x2,  #32
        mov             x4, #-32
        ld1             {v2.4S, v3.4S},  [x2], x4
        ld1             {v0.4S, v1.4S},  [x1], #32
1:      subs            x3,  x3,  #8
        rev64           v3.4S,  v3.4S
        rev64           v2.4S,  v2.4S
        ext             v3.16B, v3.16B, v3.16B,  #8
        ext             v2.16B, v2.16B, v2.16B,  #8
        fmul            v16.4S, v0.4S,  v3.4S
        fmul            v17.4S, v1.4S,  v2.4S
        b.eq            2f
        ld1             {v2.4S, v3.4S},  [x2], x4
        ld1             {v0.4S, v1.4S},  [x1], #32
        st1             {v16.4S, v17.4S},  [x0], #32
        b               1b
2:      st1             {v16.4S, v17.4S},  [x0], #32
        ret
endfunc

function ff_butterflies_float_neon, export=1
1:      ld1             {v0.4S}, [x0]
        ld1             {v1.4S}, [x1]
        subs            w2,  w2,  #4
        fsub            v2.4S,   v0.4S,  v1.4S
        fadd            v3.4S,   v0.4S,  v1.4S
        st1             {v2.4S}, [x1],   #16
        st1             {v3.4S}, [x0],   #16
        b.gt            1b
        ret
endfunc

function ff_scalarproduct_float_neon, export=1
        movi            v2.4S,  #0
1:      ld1             {v0.4S}, [x0],   #16
        ld1             {v1.4S}, [x1],   #16
        subs            w2,      w2,     #4
        fmla            v2.4S,   v0.4S,  v1.4S
        b.gt            1b
        faddp           v0.4S,   v2.4S,  v2.4S
        faddp           s0,      v0.2S
        ret
endfunc