aboutsummaryrefslogtreecommitdiffstats
path: root/libavcodec/x86/mlpdsp.c
blob: ef73433058e5df1b3f2e412082652e6b6175a426 (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
/*
 * MLP DSP functions x86-optimized
 * Copyright (c) 2009 Ramiro Polla
 *
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * FFmpeg is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with FFmpeg; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include "libavutil/attributes.h"
#include "libavutil/x86/asm.h"
#include "libavcodec/mlpdsp.h"
#include "libavcodec/mlp.h"

#if HAVE_7REGS && HAVE_INLINE_ASM

extern char ff_mlp_firorder_8;
extern char ff_mlp_firorder_7;
extern char ff_mlp_firorder_6;
extern char ff_mlp_firorder_5;
extern char ff_mlp_firorder_4;
extern char ff_mlp_firorder_3;
extern char ff_mlp_firorder_2;
extern char ff_mlp_firorder_1;
extern char ff_mlp_firorder_0;

extern char ff_mlp_iirorder_4;
extern char ff_mlp_iirorder_3;
extern char ff_mlp_iirorder_2;
extern char ff_mlp_iirorder_1;
extern char ff_mlp_iirorder_0;

static const void *firtable[9] = { &ff_mlp_firorder_0, &ff_mlp_firorder_1,
                                   &ff_mlp_firorder_2, &ff_mlp_firorder_3,
                                   &ff_mlp_firorder_4, &ff_mlp_firorder_5,
                                   &ff_mlp_firorder_6, &ff_mlp_firorder_7,
                                   &ff_mlp_firorder_8 };
static const void *iirtable[5] = { &ff_mlp_iirorder_0, &ff_mlp_iirorder_1,
                                   &ff_mlp_iirorder_2, &ff_mlp_iirorder_3,
                                   &ff_mlp_iirorder_4 };

#if ARCH_X86_64

#define MLPMUL(label, offset, offs, offc)   \
    LABEL_MANGLE(label)":             \n\t" \
    "movslq "offset"+"offs"(%0), %%rax\n\t" \
    "movslq "offset"+"offc"(%1), %%rdx\n\t" \
    "imul                 %%rdx, %%rax\n\t" \
    "add                  %%rax, %%rsi\n\t"

#define FIRMULREG(label, offset, firc)\
    LABEL_MANGLE(label)":       \n\t" \
    "movslq "#offset"(%0), %%rax\n\t" \
    "imul        %"#firc", %%rax\n\t" \
    "add            %%rax, %%rsi\n\t"

#define CLEAR_ACCUM                   \
    "xor            %%rsi, %%rsi\n\t"

#define SHIFT_ACCUM                   \
    "shr     %%cl,         %%rsi\n\t"

#define ACCUM    "%%rdx"
#define RESULT   "%%rsi"
#define RESULT32 "%%esi"

#else /* if ARCH_X86_32 */

#define MLPMUL(label, offset, offs, offc)  \
    LABEL_MANGLE(label)":            \n\t" \
    "mov   "offset"+"offs"(%0), %%eax\n\t" \
    "imull "offset"+"offc"(%1)       \n\t" \
    "add                %%eax , %%esi\n\t" \
    "adc                %%edx , %%ecx\n\t"

#define FIRMULREG(label, offset, firc)  \
    MLPMUL(label, #offset, "0", "0")

#define CLEAR_ACCUM                  \
    "xor           %%esi, %%esi\n\t" \
    "xor           %%ecx, %%ecx\n\t"

#define SHIFT_ACCUM                  \
    "mov           %%ecx, %%edx\n\t" \
    "mov           %%esi, %%eax\n\t" \
    "movzbl        %7   , %%ecx\n\t" \
    "shrd    %%cl, %%edx, %%eax\n\t" \

#define ACCUM    "%%edx"
#define RESULT   "%%eax"
#define RESULT32 "%%eax"

#endif /* !ARCH_X86_64 */

#define BINC  AV_STRINGIFY(4* MAX_CHANNELS)
#define IOFFS AV_STRINGIFY(4*(MAX_FIR_ORDER + MAX_BLOCKSIZE))
#define IOFFC AV_STRINGIFY(4* MAX_FIR_ORDER)

#define FIRMUL(label, offset) MLPMUL(label, #offset,   "0",   "0")
#define IIRMUL(label, offset) MLPMUL(label, #offset, IOFFS, IOFFC)

static void mlp_filter_channel_x86(int32_t *state, const int32_t *coeff,
                                   int firorder, int iirorder,
                                   unsigned int filter_shift, int32_t mask,
                                   int blocksize, int32_t *sample_buffer)
{
    const void *firjump = firtable[firorder];
    const void *iirjump = iirtable[iirorder];

    blocksize = -blocksize;

    __asm__ volatile(
        "1:                           \n\t"
        CLEAR_ACCUM
        "jmp  *%5                     \n\t"
        FIRMUL   (ff_mlp_firorder_8, 0x1c   )
        FIRMUL   (ff_mlp_firorder_7, 0x18   )
        FIRMUL   (ff_mlp_firorder_6, 0x14   )
        FIRMUL   (ff_mlp_firorder_5, 0x10   )
        FIRMUL   (ff_mlp_firorder_4, 0x0c   )
        FIRMUL   (ff_mlp_firorder_3, 0x08   )
        FIRMUL   (ff_mlp_firorder_2, 0x04   )
        FIRMULREG(ff_mlp_firorder_1, 0x00, 8)
        LABEL_MANGLE(ff_mlp_firorder_0)":\n\t"
        "jmp  *%6                     \n\t"
        IIRMUL   (ff_mlp_iirorder_4, 0x0c   )
        IIRMUL   (ff_mlp_iirorder_3, 0x08   )
        IIRMUL   (ff_mlp_iirorder_2, 0x04   )
        IIRMUL   (ff_mlp_iirorder_1, 0x00   )
        LABEL_MANGLE(ff_mlp_iirorder_0)":\n\t"
        SHIFT_ACCUM
        "mov  "RESULT"  ,"ACCUM"      \n\t"
        "add  (%2)      ,"RESULT"     \n\t"
        "and   %4       ,"RESULT"     \n\t"
        "sub   $4       ,  %0         \n\t"
        "mov  "RESULT32", (%0)        \n\t"
        "mov  "RESULT32", (%2)        \n\t"
        "add $"BINC"    ,  %2         \n\t"
        "sub  "ACCUM"   ,"RESULT"     \n\t"
        "mov  "RESULT32","IOFFS"(%0)  \n\t"
        "incl              %3         \n\t"
        "js 1b                        \n\t"
        : /* 0*/"+r"(state),
          /* 1*/"+r"(coeff),
          /* 2*/"+r"(sample_buffer),
#if ARCH_X86_64
          /* 3*/"+r"(blocksize)
        : /* 4*/"r"((x86_reg)mask), /* 5*/"r"(firjump),
          /* 6*/"r"(iirjump)      , /* 7*/"c"(filter_shift)
        , /* 8*/"r"((int64_t)coeff[0])
        : "rax", "rdx", "rsi"
#else /* ARCH_X86_32 */
          /* 3*/"+m"(blocksize)
        : /* 4*/"m"(         mask), /* 5*/"m"(firjump),
          /* 6*/"m"(iirjump)      , /* 7*/"m"(filter_shift)
        : "eax", "edx", "esi", "ecx"
#endif /* !ARCH_X86_64 */
    );
}

#endif /* HAVE_7REGS && HAVE_INLINE_ASM */

av_cold void ff_mlpdsp_init_x86(MLPDSPContext *c)
{
#if HAVE_7REGS && HAVE_INLINE_ASM
    c->mlp_filter_channel = mlp_filter_channel_x86;
#endif
}