1 ;******************************************************************************
2 ;* TTA Encoder DSP SIMD optimizations
4 ;* Copyright (C) 2014-2016 James Almer
6 ;* This file is part of FFmpeg.
8 ;* FFmpeg is free software; you can redistribute it and/or
9 ;* modify it under the terms of the GNU Lesser General Public
10 ;* License as published by the Free Software Foundation; either
11 ;* 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 ;* but WITHOUT ANY WARRANTY; without even the implied warranty of
15 ;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 ;* Lesser General Public License for more details.
18 ;* 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 ;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 ;******************************************************************************
23 %include "libavutil/x86/x86util.asm"
27 pd_n0113: dd ~0, ~1, ~1, ~3
28 pd_1224: dd 1, 2, 2, 4
32 %macro TTAENC_FILTER 2
34 cglobal ttaenc_filter_process, 5,5,%2, qm, dx, dl, error, in, shift, round
40 movd m6, [errorq] ; if (filter->error < 0) {
41 SPLATD m6 ; for (int i = 0; i < 8; i++)
42 psignd m0, m4, m6 ; filter->qm[i] -= filter->dx[i];
43 psignd m1, m5, m6 ; } else if (filter->error > 0) {
44 paddd m2, m0 ; for (int i = 0; i < 8; i++)
45 paddd m3, m1 ; filter->qm[i] += filter->dx[i];
47 mova [qmq + 0x10], m3 ;
72 ; Using horizontal add (phaddd) seems to be slower than shuffling stuff around
73 paddd m2, m3 ; int sum = filter->round +
74 ; filter->dl[0] * filter->qm[0] +
75 pshufd m3, m2, 0xe ; filter->dl[1] * filter->qm[1] +
76 paddd m2, m3 ; filter->dl[2] * filter->qm[2] +
77 ; filter->dl[3] * filter->qm[3] +
78 movd m6, roundm ; filter->dl[4] * filter->qm[4] +
79 paddd m6, m2 ; filter->dl[5] * filter->qm[5] +
80 pshufd m2, m2, 0x1 ; filter->dl[6] * filter->qm[6] +
81 paddd m6, m2 ; filter->dl[7] * filter->qm[7];
83 palignr m5, m4, 4 ; filter->dx[0] = filter->dx[1]; filter->dx[1] = filter->dx[2];
84 ; filter->dx[2] = filter->dx[3]; filter->dx[3] = filter->dx[4];
86 palignr m2, m1, m0, 4 ; filter->dl[0] = filter->dl[1]; filter->dl[1] = filter->dl[2];
87 ; filter->dl[2] = filter->dl[3]; filter->dl[3] = filter->dl[4];
89 psrad m4, m1, 30 ; filter->dx[4] = ((filter->dl[4] >> 30) | 1);
90 por m4, [pd_1224 ] ; filter->dx[5] = ((filter->dl[5] >> 30) | 2) & ~1;
91 pand m4, [pd_n0113] ; filter->dx[6] = ((filter->dl[6] >> 30) | 2) & ~1;
92 ; filter->dx[7] = ((filter->dl[7] >> 30) | 4) & ~3;
102 movd [inq], m3 ; *in -= (sum >> filter->shift);
103 movd [errorq], m3 ; filter->error = *in;
106 pslldq m0, 12 ; filter->dl[4] = -filter->dl[5];
107 pshufd m0, m0, 0xf0 ; filter->dl[5] = -filter->dl[6];
108 psubd m0, m1 ; filter->dl[6] = *in - filter->dl[7];
109 psrldq m1, m0, 4 ; filter->dl[7] = *in;
110 pshufd m1, m1, 0xf4 ; filter->dl[5] += filter->dl[6];
111 paddd m0, m1 ; filter->dl[4] += filter->dl[5];
114 mova [dlq + 0x10], m0 ;
118 TTAENC_FILTER ssse3, 8
119 TTAENC_FILTER sse4, 7