2 * This file is part of FFmpeg.
4 * FFmpeg is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU Lesser General Public
6 * License as published by the Free Software Foundation; either
7 * version 2.1 of the License, or (at your option) any later version.
9 * FFmpeg is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * Lesser General Public License for more details.
14 * You should have received a copy of the GNU Lesser General Public
15 * License along with FFmpeg; if not, write to the Free Software
16 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19 #include "libavutil/attributes.h"
20 #include "libavutil/cpu.h"
21 #include "libavutil/float_dsp.h"
24 void ff_vector_fmul_sse(float *dst
, const float *src0
, const float *src1
,
26 void ff_vector_fmul_avx(float *dst
, const float *src0
, const float *src1
,
29 void ff_vector_dmul_sse2(double *dst
, const double *src0
, const double *src1
,
31 void ff_vector_dmul_avx(double *dst
, const double *src0
, const double *src1
,
34 void ff_vector_fmac_scalar_sse(float *dst
, const float *src
, float mul
,
36 void ff_vector_fmac_scalar_avx(float *dst
, const float *src
, float mul
,
38 void ff_vector_fmac_scalar_fma3(float *dst
, const float *src
, float mul
,
41 void ff_vector_fmul_scalar_sse(float *dst
, const float *src
, float mul
,
44 void ff_vector_dmac_scalar_sse2(double *dst
, const double *src
, double mul
,
46 void ff_vector_dmac_scalar_avx(double *dst
, const double *src
, double mul
,
48 void ff_vector_dmac_scalar_fma3(double *dst
, const double *src
, double mul
,
51 void ff_vector_dmul_scalar_sse2(double *dst
, const double *src
,
53 void ff_vector_dmul_scalar_avx(double *dst
, const double *src
,
56 void ff_vector_fmul_window_sse(float *dst
, const float *src0
,
57 const float *src1
, const float *win
, int len
);
59 void ff_vector_fmul_add_sse(float *dst
, const float *src0
, const float *src1
,
60 const float *src2
, int len
);
61 void ff_vector_fmul_add_avx(float *dst
, const float *src0
, const float *src1
,
62 const float *src2
, int len
);
63 void ff_vector_fmul_add_fma3(float *dst
, const float *src0
, const float *src1
,
64 const float *src2
, int len
);
66 void ff_vector_fmul_reverse_sse(float *dst
, const float *src0
,
67 const float *src1
, int len
);
68 void ff_vector_fmul_reverse_avx(float *dst
, const float *src0
,
69 const float *src1
, int len
);
70 void ff_vector_fmul_reverse_avx2(float *dst
, const float *src0
,
71 const float *src1
, int len
);
73 float ff_scalarproduct_float_sse(const float *v1
, const float *v2
, int order
);
74 float ff_scalarproduct_float_fma3(const float *v1
, const float *v2
, int order
);
76 double ff_scalarproduct_double_sse2(const double *v1
, const double *v2
, size_t order
);
77 double ff_scalarproduct_double_avx(const double *v1
, const double *v2
, size_t order
);
79 void ff_butterflies_float_sse(float *restrict src0
, float *restrict src1
, int len
);
81 av_cold
void ff_float_dsp_init_x86(AVFloatDSPContext
*fdsp
)
83 int cpu_flags
= av_get_cpu_flags();
85 if (EXTERNAL_SSE(cpu_flags
)) {
86 fdsp
->vector_fmul
= ff_vector_fmul_sse
;
87 fdsp
->vector_fmac_scalar
= ff_vector_fmac_scalar_sse
;
88 fdsp
->vector_fmul_scalar
= ff_vector_fmul_scalar_sse
;
89 fdsp
->vector_fmul_window
= ff_vector_fmul_window_sse
;
90 fdsp
->vector_fmul_add
= ff_vector_fmul_add_sse
;
91 fdsp
->vector_fmul_reverse
= ff_vector_fmul_reverse_sse
;
92 fdsp
->scalarproduct_float
= ff_scalarproduct_float_sse
;
93 fdsp
->butterflies_float
= ff_butterflies_float_sse
;
95 if (EXTERNAL_SSE2(cpu_flags
)) {
96 fdsp
->vector_dmul
= ff_vector_dmul_sse2
;
97 fdsp
->vector_dmac_scalar
= ff_vector_dmac_scalar_sse2
;
98 fdsp
->vector_dmul_scalar
= ff_vector_dmul_scalar_sse2
;
99 fdsp
->scalarproduct_double
= ff_scalarproduct_double_sse2
;
101 if (EXTERNAL_AVX_FAST(cpu_flags
)) {
102 fdsp
->vector_fmul
= ff_vector_fmul_avx
;
103 fdsp
->vector_dmul
= ff_vector_dmul_avx
;
104 fdsp
->vector_fmac_scalar
= ff_vector_fmac_scalar_avx
;
105 fdsp
->vector_dmul_scalar
= ff_vector_dmul_scalar_avx
;
106 fdsp
->vector_dmac_scalar
= ff_vector_dmac_scalar_avx
;
107 fdsp
->vector_fmul_add
= ff_vector_fmul_add_avx
;
108 fdsp
->vector_fmul_reverse
= ff_vector_fmul_reverse_avx
;
109 fdsp
->scalarproduct_double
= ff_scalarproduct_double_avx
;
111 if (EXTERNAL_AVX2_FAST(cpu_flags
)) {
112 fdsp
->vector_fmul_reverse
= ff_vector_fmul_reverse_avx2
;
114 if (EXTERNAL_FMA3_FAST(cpu_flags
)) {
115 fdsp
->vector_fmac_scalar
= ff_vector_fmac_scalar_fma3
;
116 fdsp
->vector_fmul_add
= ff_vector_fmul_add_fma3
;
117 fdsp
->vector_dmac_scalar
= ff_vector_dmac_scalar_fma3
;
118 fdsp
->scalarproduct_float
= ff_scalarproduct_float_fma3
;