2 * Copyright (c) 2008 Mans Rullgard <mans@mansr.com>
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
26 function ff_h264_idct_add_neon, export=1
30 vld1.64 {d0-d3}, [r1,:128]
48 vld1.32 {d18[0]}, [r0,:32], r2
51 vld1.32 {d19[1]}, [r0,:32], r2
53 vld1.32 {d18[1]}, [r0,:32], r2
55 vld1.32 {d19[0]}, [r0,:32], r2
57 sub r0, r0, r2, lsl #2
70 vst1.32 {d0[0]}, [r0,:32], r2
71 vst1.32 {d1[1]}, [r0,:32], r2
72 vst1.32 {d0[1]}, [r0,:32], r2
73 vst1.32 {d1[0]}, [r0,:32], r2
78 function ff_h264_idct_dc_add_neon, export=1
79 vld1.16 {d2[],d3[]}, [r1,:16]
81 vld1.32 {d0[0]}, [r0,:32], r2
82 vld1.32 {d0[1]}, [r0,:32], r2
84 vld1.32 {d1[0]}, [r0,:32], r2
85 vld1.32 {d1[1]}, [r0,:32], r2
89 sub r0, r0, r2, lsl #2
90 vst1.32 {d0[0]}, [r0,:32], r2
91 vst1.32 {d0[1]}, [r0,:32], r2
92 vst1.32 {d1[0]}, [r0,:32], r2
93 vst1.32 {d1[1]}, [r0,:32], r2
97 function ff_h264_idct_add16_neon, export=1
115 adrne lr, ff_h264_idct_dc_add_neon
116 adreq lr, ff_h264_idct_add_neon
124 function ff_h264_idct_add16intra_neon, export=1
139 adrne lr, ff_h264_idct_add_neon
140 adreq lr, ff_h264_idct_dc_add_neon
149 function ff_h264_idct_add8_neon, export=1
166 adrne lr, ff_h264_idct_add_neon
167 adreq lr, ff_h264_idct_dc_add_neon
177 scan8: .byte 4+1*8, 5+1*8, 4+2*8, 5+2*8
178 .byte 6+1*8, 7+1*8, 6+2*8, 7+2*8
179 .byte 4+3*8, 5+3*8, 4+4*8, 5+4*8
180 .byte 6+3*8, 7+3*8, 6+4*8, 7+4*8