1 /*
2  * Copyright © 2018, VideoLAN and dav1d authors
3  * Copyright © 2018, Two Orioles, LLC
4  * All rights reserved.
5  *
6  * Redistribution and use in source and binary forms, with or without
7  * modification, are permitted provided that the following conditions are met:
8  *
9  * 1. Redistributions of source code must retain the above copyright notice, this
10  *    list of conditions and the following disclaimer.
11  *
12  * 2. Redistributions in binary form must reproduce the above copyright notice,
13  *    this list of conditions and the following disclaimer in the documentation
14  *    and/or other materials provided with the distribution.
15  *
16  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
17  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
18  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
19  * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
20  * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
21  * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
22  * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
23  * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
24  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
25  * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
26  */
27 
28 #include "src/cpu.h"
29 #include "src/film_grain.h"
30 
31 decl_generate_grain_y_fn(dav1d_generate_grain_y_ssse3);
32 decl_generate_grain_uv_fn(dav1d_generate_grain_uv_420_ssse3);
33 decl_generate_grain_uv_fn(dav1d_generate_grain_uv_422_ssse3);
34 decl_generate_grain_uv_fn(dav1d_generate_grain_uv_444_ssse3);
35 decl_fgy_32x32xn_fn(dav1d_fgy_32x32xn_ssse3);
36 decl_fguv_32x32xn_fn(dav1d_fguv_32x32xn_i420_ssse3);
37 decl_fguv_32x32xn_fn(dav1d_fguv_32x32xn_i422_ssse3);
38 decl_fguv_32x32xn_fn(dav1d_fguv_32x32xn_i444_ssse3);
39 
40 decl_generate_grain_y_fn(dav1d_generate_grain_y_avx2);
41 decl_generate_grain_uv_fn(dav1d_generate_grain_uv_420_avx2);
42 decl_generate_grain_uv_fn(dav1d_generate_grain_uv_422_avx2);
43 decl_generate_grain_uv_fn(dav1d_generate_grain_uv_444_avx2);
44 decl_fgy_32x32xn_fn(dav1d_fgy_32x32xn_avx2);
45 decl_fguv_32x32xn_fn(dav1d_fguv_32x32xn_i420_avx2);
46 decl_fguv_32x32xn_fn(dav1d_fguv_32x32xn_i422_avx2);
47 decl_fguv_32x32xn_fn(dav1d_fguv_32x32xn_i444_avx2);
48 
49 decl_generate_grain_y_fn(dav1d_generate_grain_y_16bpc_avx2);
50 decl_generate_grain_uv_fn(dav1d_generate_grain_uv_420_16bpc_avx2);
51 decl_fgy_32x32xn_fn(dav1d_fgy_32x32xn_16bpc_avx2);
52 decl_fguv_32x32xn_fn(dav1d_fguv_32x32xn_i420_16bpc_avx2);
53 
bitfn(dav1d_film_grain_dsp_init_x86)54 COLD void bitfn(dav1d_film_grain_dsp_init_x86)(Dav1dFilmGrainDSPContext *const c) {
55     const unsigned flags = dav1d_get_cpu_flags();
56 
57     if (!(flags & DAV1D_X86_CPU_FLAG_SSSE3)) return;
58 
59 #if BITDEPTH == 8
60     c->generate_grain_y = dav1d_generate_grain_y_ssse3;
61     c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I420 - 1] = dav1d_generate_grain_uv_420_ssse3;
62     c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I422 - 1] = dav1d_generate_grain_uv_422_ssse3;
63     c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I444 - 1] = dav1d_generate_grain_uv_444_ssse3;
64     c->fgy_32x32xn = dav1d_fgy_32x32xn_ssse3;
65     c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I420 - 1] = dav1d_fguv_32x32xn_i420_ssse3;
66     c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I422 - 1] = dav1d_fguv_32x32xn_i422_ssse3;
67     c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I444 - 1] = dav1d_fguv_32x32xn_i444_ssse3;
68 #endif
69 
70 #if ARCH_X86_64
71     if (!(flags & DAV1D_X86_CPU_FLAG_AVX2)) return;
72 
73 #if BITDEPTH == 8
74     c->generate_grain_y = dav1d_generate_grain_y_avx2;
75     c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I420 - 1] = dav1d_generate_grain_uv_420_avx2;
76     c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I422 - 1] = dav1d_generate_grain_uv_422_avx2;
77     c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I444 - 1] = dav1d_generate_grain_uv_444_avx2;
78     c->fgy_32x32xn = dav1d_fgy_32x32xn_avx2;
79     c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I420 - 1] = dav1d_fguv_32x32xn_i420_avx2;
80     c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I422 - 1] = dav1d_fguv_32x32xn_i422_avx2;
81     c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I444 - 1] = dav1d_fguv_32x32xn_i444_avx2;
82 #else
83     c->generate_grain_y = dav1d_generate_grain_y_16bpc_avx2;
84     c->generate_grain_uv[DAV1D_PIXEL_LAYOUT_I420 - 1] =
85         dav1d_generate_grain_uv_420_16bpc_avx2;
86     c->fgy_32x32xn = dav1d_fgy_32x32xn_16bpc_avx2;
87     c->fguv_32x32xn[DAV1D_PIXEL_LAYOUT_I420 - 1] =
88         dav1d_fguv_32x32xn_i420_16bpc_avx2;
89 #endif
90 #endif
91 }
92