Statistics
| Branch: | Revision:

ffmpeg / libavcodec / x86 / fft_sse.c @ 2912e87a

History | View | Annotate | Download (2.93 KB)

1
/*
2
 * FFT/MDCT transform with SSE optimizations
3
 * Copyright (c) 2008 Loren Merritt
4
 *
5
 * This file is part of Libav.
6
 *
7
 * Libav is free software; you can redistribute it and/or
8
 * modify it under the terms of the GNU Lesser General Public
9
 * License as published by the Free Software Foundation; either
10
 * version 2.1 of the License, or (at your option) any later version.
11
 *
12
 * Libav is distributed in the hope that it will be useful,
13
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
15
 * Lesser General Public License for more details.
16
 *
17
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with Libav; if not, write to the Free Software
19
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20
 */
21

    
22
#include "libavutil/x86_cpu.h"
23
#include "libavcodec/dsputil.h"
24
#include "fft.h"
25

    
26
DECLARE_ASM_CONST(16, int, ff_m1m1m1m1)[4] =
27
    { 1 << 31, 1 << 31, 1 << 31, 1 << 31 };
28

    
29
void ff_fft_dispatch_sse(FFTComplex *z, int nbits);
30
void ff_fft_dispatch_interleave_sse(FFTComplex *z, int nbits);
31

    
32
void ff_fft_calc_sse(FFTContext *s, FFTComplex *z)
33
{
34
    int n = 1 << s->nbits;
35

    
36
    ff_fft_dispatch_interleave_sse(z, s->nbits);
37

    
38
    if(n <= 16) {
39
        x86_reg i = -8*n;
40
        __asm__ volatile(
41
            "1: \n"
42
            "movaps     (%0,%1), %%xmm0 \n"
43
            "movaps      %%xmm0, %%xmm1 \n"
44
            "unpcklps 16(%0,%1), %%xmm0 \n"
45
            "unpckhps 16(%0,%1), %%xmm1 \n"
46
            "movaps      %%xmm0,   (%0,%1) \n"
47
            "movaps      %%xmm1, 16(%0,%1) \n"
48
            "add $32, %0 \n"
49
            "jl 1b \n"
50
            :"+r"(i)
51
            :"r"(z+n)
52
            :"memory"
53
        );
54
    }
55
}
56

    
57
void ff_fft_permute_sse(FFTContext *s, FFTComplex *z)
58
{
59
    int n = 1 << s->nbits;
60
    int i;
61
    for(i=0; i<n; i+=2) {
62
        __asm__ volatile(
63
            "movaps %2, %%xmm0 \n"
64
            "movlps %%xmm0, %0 \n"
65
            "movhps %%xmm0, %1 \n"
66
            :"=m"(s->tmp_buf[s->revtab[i]]),
67
             "=m"(s->tmp_buf[s->revtab[i+1]])
68
            :"m"(z[i])
69
        );
70
    }
71
    memcpy(z, s->tmp_buf, n*sizeof(FFTComplex));
72
}
73

    
74
void ff_imdct_calc_sse(FFTContext *s, FFTSample *output, const FFTSample *input)
75
{
76
    x86_reg j, k;
77
    long n = s->mdct_size;
78
    long n4 = n >> 2;
79

    
80
    ff_imdct_half_sse(s, output+n4, input);
81

    
82
    j = -n;
83
    k = n-16;
84
    __asm__ volatile(
85
        "movaps "MANGLE(ff_m1m1m1m1)", %%xmm7 \n"
86
        "1: \n"
87
        "movaps       (%2,%1), %%xmm0 \n"
88
        "movaps       (%3,%0), %%xmm1 \n"
89
        "shufps $0x1b, %%xmm0, %%xmm0 \n"
90
        "shufps $0x1b, %%xmm1, %%xmm1 \n"
91
        "xorps         %%xmm7, %%xmm0 \n"
92
        "movaps        %%xmm1, (%3,%1) \n"
93
        "movaps        %%xmm0, (%2,%0) \n"
94
        "sub $16, %1 \n"
95
        "add $16, %0 \n"
96
        "jl 1b \n"
97
        :"+r"(j), "+r"(k)
98
        :"r"(output+n4), "r"(output+n4*3)
99
        XMM_CLOBBERS_ONLY("%xmm0", "%xmm1", "%xmm7")
100
    );
101
}
102