aac.h 5.65 KB
Newer Older
1 2 3
/*
 * Copyright (c) 2010 Mans Rullgard <mans@mansr.com>
 *
4
 * This file is part of Libav.
5
 *
6
 * Libav is free software; you can redistribute it and/or
7 8 9 10
 * 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.
 *
11
 * Libav is distributed in the hope that it will be useful,
12 13 14 15 16
 * 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
17
 * License along with Libav; if not, write to the Free Software
18 19 20
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

21 22 23 24 25
#ifndef AVCODEC_ARM_AAC_H
#define AVCODEC_ARM_AAC_H

#include "config.h"

26 27 28 29 30 31 32
#if HAVE_NEON && HAVE_INLINE_ASM

#define VMUL2 VMUL2
static inline float *VMUL2(float *dst, const float *v, unsigned idx,
                           const float *scale)
{
    unsigned v0, v1;
33 34 35 36 37 38 39 40 41 42 43
    __asm__ ("ubfx     %0,  %6,  #0, #4      \n\t"
             "ubfx     %1,  %6,  #4, #4      \n\t"
             "ldr      %0,  [%5, %0, lsl #2] \n\t"
             "ldr      %1,  [%5, %1, lsl #2] \n\t"
             "vld1.32  {d1[]},   [%7,:32]    \n\t"
             "vmov     d0,  %0,  %1          \n\t"
             "vmul.f32 d0,  d0,  d1          \n\t"
             "vst1.32  {d0},     [%2,:64]!   \n\t"
             : "=&r"(v0), "=&r"(v1), "+r"(dst), "=m"(dst[0]), "=m"(dst[1])
             : "r"(v), "r"(idx), "r"(scale)
             : "d0", "d1");
44 45 46 47 48 49 50 51
    return dst;
}

#define VMUL4 VMUL4
static inline float *VMUL4(float *dst, const float *v, unsigned idx,
                           const float *scale)
{
    unsigned v0, v1, v2, v3;
52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68
    __asm__ ("ubfx     %0,  %10, #0, #2      \n\t"
             "ubfx     %1,  %10, #2, #2      \n\t"
             "ldr      %0,  [%9, %0, lsl #2] \n\t"
             "ubfx     %2,  %10, #4, #2      \n\t"
             "ldr      %1,  [%9, %1, lsl #2] \n\t"
             "ubfx     %3,  %10, #6, #2      \n\t"
             "ldr      %2,  [%9, %2, lsl #2] \n\t"
             "vmov     d0,  %0,  %1          \n\t"
             "ldr      %3,  [%9, %3, lsl #2] \n\t"
             "vld1.32  {d2[],d3[]},[%11,:32] \n\t"
             "vmov     d1,  %2,  %3          \n\t"
             "vmul.f32 q0,  q0,  q1          \n\t"
             "vst1.32  {q0},     [%4,:128]!  \n\t"
             : "=&r"(v0), "=&r"(v1), "=&r"(v2), "=&r"(v3), "+r"(dst),
               "=m"(dst[0]), "=m"(dst[1]), "=m"(dst[2]), "=m"(dst[3])
             : "r"(v), "r"(idx), "r"(scale)
             : "d0", "d1", "d2", "d3");
69 70 71 72 73 74 75 76
    return dst;
}

#define VMUL2S VMUL2S
static inline float *VMUL2S(float *dst, const float *v, unsigned idx,
                            unsigned sign, const float *scale)
{
    unsigned v0, v1, v2, v3;
77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93
    __asm__ ("ubfx     %0,  %8,  #0, #4      \n\t"
             "ubfx     %1,  %8,  #4, #4      \n\t"
             "ldr      %0,  [%7, %0, lsl #2] \n\t"
             "lsl      %2,  %10, #30         \n\t"
             "ldr      %1,  [%7, %1, lsl #2] \n\t"
             "lsl      %3,  %10, #31         \n\t"
             "vmov     d0,  %0,  %1          \n\t"
             "bic      %2,  %2,  #1<<30      \n\t"
             "vld1.32  {d1[]},   [%9,:32]    \n\t"
             "vmov     d2,  %2,  %3          \n\t"
             "veor     d0,  d0,  d2          \n\t"
             "vmul.f32 d0,  d0,  d1          \n\t"
             "vst1.32  {d0},     [%4,:64]!   \n\t"
             : "=&r"(v0), "=&r"(v1), "=&r"(v2), "=&r"(v3), "+r"(dst),
               "=m"(dst[0]), "=m"(dst[1])
             : "r"(v), "r"(idx), "r"(scale), "r"(sign)
             : "d0", "d1", "d2");
94 95 96 97 98 99 100 101
    return dst;
}

#define VMUL4S VMUL4S
static inline float *VMUL4S(float *dst, const float *v, unsigned idx,
                            unsigned sign, const float *scale)
{
    unsigned v0, v1, v2, v3, nz;
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
    __asm__ ("vld1.32  {d2[],d3[]},[%13,:32] \n\t"
             "ubfx     %0,  %12, #0, #2      \n\t"
             "ubfx     %1,  %12, #2, #2      \n\t"
             "ldr      %0,  [%11,%0, lsl #2] \n\t"
             "ubfx     %2,  %12, #4, #2      \n\t"
             "ldr      %1,  [%11,%1, lsl #2] \n\t"
             "ubfx     %3,  %12, #6, #2      \n\t"
             "ldr      %2,  [%11,%2, lsl #2] \n\t"
             "vmov     d0,  %0,  %1          \n\t"
             "ldr      %3,  [%11,%3, lsl #2] \n\t"
             "lsr      %6,  %12, #12         \n\t"
             "rbit     %6,  %6               \n\t"
             "vmov     d1,  %2,  %3          \n\t"
             "lsls     %6,  %6,  #1          \n\t"
             "and      %0,  %5,  #1<<31      \n\t"
             "lslcs    %5,  %5,  #1          \n\t"
             "lsls     %6,  %6,  #1          \n\t"
             "and      %1,  %5,  #1<<31      \n\t"
             "lslcs    %5,  %5,  #1          \n\t"
             "lsls     %6,  %6,  #1          \n\t"
             "and      %2,  %5,  #1<<31      \n\t"
             "lslcs    %5,  %5,  #1          \n\t"
             "vmov     d4,  %0,  %1          \n\t"
             "and      %3,  %5,  #1<<31      \n\t"
             "vmov     d5,  %2,  %3          \n\t"
             "veor     q0,  q0,  q2          \n\t"
             "vmul.f32 q0,  q0,  q1          \n\t"
             "vst1.32  {q0},     [%4,:128]!  \n\t"
             : "=&r"(v0), "=&r"(v1), "=&r"(v2), "=&r"(v3), "+r"(dst),
               "+r"(sign), "=r"(nz),
               "=m"(dst[0]), "=m"(dst[1]), "=m"(dst[2]), "=m"(dst[3])
             : "r"(v), "r"(idx), "r"(scale)
             : "cc", "d0", "d1", "d2", "d3", "d4", "d5");
135 136 137 138
    return dst;
}

#endif /* HAVE_NEON && HAVE_INLINE_ASM */
139 140

#endif /* AVCODEC_ARM_AAC_H */