2 * Copyright (c) 2014 Janne Grunau <janne-libav@jannau.net>
4 * This file is part of FFmpeg.
6 * FFmpeg 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.
11 * FFmpeg 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.
16 * 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 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 #include "libavutil/aarch64/asm.S"
23 #define FRAC_BITS 23 // fractional bits for sb_samples and dct
24 #define WFRAC_BITS 16 // fractional bits for window
25 #define OUT_SHIFT (WFRAC_BITS + FRAC_BITS - 15)
27 const tbl_rev128.s, align=4
34 .macro apply_window type, st
35 function ff_mpadsp_apply_window_\type\()_neon, export=1
39 ld1 {v0.4s,v1.4s,v2.4s,v3.4s}, [x7], #64
40 ld1 {v4.4s,v5.4s,v6.4s,v7.4s}, [x7], #64
41 st1 {v0.4s,v1.4s,v2.4s,v3.4s}, [x8], #64
42 st1 {v4.4s,v5.4s,v6.4s,v7.4s}, [x8], #64
43 movrel x15, tbl_rev128.s
53 add x5, x3, x4, lsl #5
54 sub x5, x5, x4 // samples2
58 ld1r {v16.2s}, [x2] // dither_state
61 movi v30.2d, #(1<<OUT_SHIFT)-1
62 trn1 v31.2d, v29.2d, v30.2d
63 trn2 v30.2d, v30.2d, v29.2d
64 trn1 v16.2d, v16.2d, v29.2d
73 sub x6, x1, x14, lsl #4
74 add x7, x7, x14, lsl #4
75 add x11, x6, #(32)<<2 // w + 32
76 add x12, x7, #(32)<<2 // w2 + 32
84 ld1 {v1.4s}, [x10], x9
87 tbl v6.16b, {v0.16b}, v27.16b
88 tbl v7.16b, {v1.16b}, v27.16b
89 ld1 {v4.4s}, [x11], x9
90 ld1 {v5.4s}, [x12], x9
102 sub x10, x10, #64<<5 // 64 * 8 * sizeof(int32_t)
105 and v28.16b, v16.16b, v30.16b
106 ext v28.16b, v29.16b, v28.16b, #8
109 round_sample v19, 1, 1
111 round_sample v16, 1, 0
112 shrn v16.2s, v16.2d, #OUT_SHIFT
113 round_sample v19, 0, 0
114 shrn v19.2s, v19.2d, #OUT_SHIFT
115 round_sample v17, 0, 1
116 round_sample v18, 1, 1
117 round_sample v17, 1, 0
118 shrn2 v16.4s, v17.2d, #OUT_SHIFT
119 round_sample v18, 0, 0
120 shrn2 v19.4s, v18.2d, #OUT_SHIFT
124 ext v18.16b, v18.16b, v18.16b, #8
127 st1 {v16.\st\()}[0], [x3], x4
129 st1 {v18.\st\()}[1], [x5], x13
131 st1 {v16.\st\()}[1], [x3], x4
132 st1 {v18.\st\()}[0], [x5], x13
133 st1 {v16.\st\()}[2], [x3], x4
134 st1 {v18.\st\()}[3], [x5], x13
135 st1 {v16.\st\()}[3], [x3], x4
136 st1 {v18.\st\()}[2], [x5], x13
145 // comuting samples[16]
147 ld1 {v0.2s}, [x6], x9
148 ld1 {v1.2s}, [x0], x9
150 ld1 {v2.2s}, [x6], x9
151 ld1 {v3.2s}, [x0], x9
153 ld1 {v0.2s}, [x6], x9
154 ld1 {v1.2s}, [x0], x9
157 ld1 {v2.2s}, [x6], x9
158 ld1 {v3.2s}, [x0], x9
163 and v28.16b, v16.16b, v30.16b
164 shrn v20.2s, v16.2d, #OUT_SHIFT
167 st1 {v28.s}[0], [x2] // save dither_state
183 .macro round_sample r, idx, next
184 add \r\().2d, \r\().2d, v28.2d
186 and v28.16b, \r\().16b, v30.16b
188 and v28.16b, \r\().16b, v31.16b
192 ext v28.16b, v28.16b, v29.16b, #8
194 ext v28.16b, v29.16b, v28.16b, #8
199 smlal \d\().2d, \s1\().2s, \s2\().2s
201 .macro MLA2 d, s1, s2
202 smlal2 \d\().2d, \s1\().4s, \s2\().4s
205 smlsl \d\().2d, \s1\().2s, \s2\().2s
207 .macro MLS2 d, s1, s2
208 smlsl2 \d\().2d, \s1\().4s, \s2\().4s
210 apply_window fixed, h
213 // nothing to do for round_sample and ML{A,S}2
214 .macro round_sample r, idx, next
216 .macro MLA2 d, s1, s2
218 .macro MLS2 d, s1, s2
221 fmla \d\().4s, \s1\().4s, \s2\().4s
224 fmls \d\().4s, \s1\().4s, \s2\().4s
226 apply_window float, s