ffmpeg/libavutil/aarch64/float_dsp_neon.S

203 lines
7.7 KiB
ArmAsm

/*
* ARM NEON optimised Float DSP functions
* Copyright (c) 2008 Mans Rullgard <mans@mansr.com>
* Copyright (c) 2014 Janne Grunau <janne-libav@jannau.net>
*
* This file is part of FFmpeg.
*
* FFmpeg 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.
*
* FFmpeg 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 FFmpeg; if not, write to the Free Software
* 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