| 1 | ;****************************************************************************** |
| 2 | ;* optimized audio functions |
| 3 | ;* Copyright (c) 2008 Loren Merritt |
| 4 | ;* |
| 5 | ;* This file is part of FFmpeg. |
| 6 | ;* |
| 7 | ;* FFmpeg 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 | ;* FFmpeg 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 FFmpeg; 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/x86util.asm" |
| 23 | |
| 24 | SECTION_TEXT |
| 25 | |
| 26 | %macro SCALARPRODUCT 0 |
| 27 | ; int ff_scalarproduct_int16(int16_t *v1, int16_t *v2, int order) |
| 28 | cglobal scalarproduct_int16, 3,3,3, v1, v2, order |
| 29 | shl orderq, 1 |
| 30 | add v1q, orderq |
| 31 | add v2q, orderq |
| 32 | neg orderq |
| 33 | pxor m2, m2 |
| 34 | .loop: |
| 35 | movu m0, [v1q + orderq] |
| 36 | movu m1, [v1q + orderq + mmsize] |
| 37 | pmaddwd m0, [v2q + orderq] |
| 38 | pmaddwd m1, [v2q + orderq + mmsize] |
| 39 | paddd m2, m0 |
| 40 | paddd m2, m1 |
| 41 | add orderq, mmsize*2 |
| 42 | jl .loop |
| 43 | HADDD m2, m0 |
| 44 | movd eax, m2 |
| 45 | %if mmsize == 8 |
| 46 | emms |
| 47 | %endif |
| 48 | RET |
| 49 | %endmacro |
| 50 | |
| 51 | INIT_MMX mmxext |
| 52 | SCALARPRODUCT |
| 53 | INIT_XMM sse2 |
| 54 | SCALARPRODUCT |
| 55 | |
| 56 | |
| 57 | ;----------------------------------------------------------------------------- |
| 58 | ; void ff_vector_clip_int32(int32_t *dst, const int32_t *src, int32_t min, |
| 59 | ; int32_t max, unsigned int len) |
| 60 | ;----------------------------------------------------------------------------- |
| 61 | |
| 62 | ; %1 = number of xmm registers used |
| 63 | ; %2 = number of inline load/process/store loops per asm loop |
| 64 | ; %3 = process 4*mmsize (%3=0) or 8*mmsize (%3=1) bytes per loop |
| 65 | ; %4 = CLIPD function takes min/max as float instead of int (CLIPD_SSE2) |
| 66 | ; %5 = suffix |
| 67 | %macro VECTOR_CLIP_INT32 4-5 |
| 68 | cglobal vector_clip_int32%5, 5,5,%1, dst, src, min, max, len |
| 69 | %if %4 |
| 70 | cvtsi2ss m4, minm |
| 71 | cvtsi2ss m5, maxm |
| 72 | %else |
| 73 | movd m4, minm |
| 74 | movd m5, maxm |
| 75 | %endif |
| 76 | SPLATD m4 |
| 77 | SPLATD m5 |
| 78 | .loop: |
| 79 | %assign %%i 0 |
| 80 | %rep %2 |
| 81 | mova m0, [srcq+mmsize*(0+%%i)] |
| 82 | mova m1, [srcq+mmsize*(1+%%i)] |
| 83 | mova m2, [srcq+mmsize*(2+%%i)] |
| 84 | mova m3, [srcq+mmsize*(3+%%i)] |
| 85 | %if %3 |
| 86 | mova m7, [srcq+mmsize*(4+%%i)] |
| 87 | mova m8, [srcq+mmsize*(5+%%i)] |
| 88 | mova m9, [srcq+mmsize*(6+%%i)] |
| 89 | mova m10, [srcq+mmsize*(7+%%i)] |
| 90 | %endif |
| 91 | CLIPD m0, m4, m5, m6 |
| 92 | CLIPD m1, m4, m5, m6 |
| 93 | CLIPD m2, m4, m5, m6 |
| 94 | CLIPD m3, m4, m5, m6 |
| 95 | %if %3 |
| 96 | CLIPD m7, m4, m5, m6 |
| 97 | CLIPD m8, m4, m5, m6 |
| 98 | CLIPD m9, m4, m5, m6 |
| 99 | CLIPD m10, m4, m5, m6 |
| 100 | %endif |
| 101 | mova [dstq+mmsize*(0+%%i)], m0 |
| 102 | mova [dstq+mmsize*(1+%%i)], m1 |
| 103 | mova [dstq+mmsize*(2+%%i)], m2 |
| 104 | mova [dstq+mmsize*(3+%%i)], m3 |
| 105 | %if %3 |
| 106 | mova [dstq+mmsize*(4+%%i)], m7 |
| 107 | mova [dstq+mmsize*(5+%%i)], m8 |
| 108 | mova [dstq+mmsize*(6+%%i)], m9 |
| 109 | mova [dstq+mmsize*(7+%%i)], m10 |
| 110 | %endif |
| 111 | %assign %%i %%i+4*(%3+1) |
| 112 | %endrep |
| 113 | add srcq, mmsize*4*(%2+%3) |
| 114 | add dstq, mmsize*4*(%2+%3) |
| 115 | sub lend, mmsize*(%2+%3) |
| 116 | jg .loop |
| 117 | REP_RET |
| 118 | %endmacro |
| 119 | |
| 120 | INIT_MMX mmx |
| 121 | %define CLIPD CLIPD_MMX |
| 122 | VECTOR_CLIP_INT32 0, 1, 0, 0 |
| 123 | INIT_XMM sse2 |
| 124 | VECTOR_CLIP_INT32 6, 1, 0, 0, _int |
| 125 | %define CLIPD CLIPD_SSE2 |
| 126 | VECTOR_CLIP_INT32 6, 2, 0, 1 |
| 127 | INIT_XMM sse4 |
| 128 | %define CLIPD CLIPD_SSE41 |
| 129 | %ifdef m8 |
| 130 | VECTOR_CLIP_INT32 11, 1, 1, 0 |
| 131 | %else |
| 132 | VECTOR_CLIP_INT32 6, 1, 0, 0 |
| 133 | %endif |
| 134 | |
| 135 | ;----------------------------------------------------- |
| 136 | ;void ff_vector_clipf(float *dst, const float *src, |
| 137 | ; float min, float max, int len) |
| 138 | ;----------------------------------------------------- |
| 139 | INIT_XMM sse |
| 140 | %if UNIX64 |
| 141 | cglobal vector_clipf, 3,3,6, dst, src, len |
| 142 | %else |
| 143 | cglobal vector_clipf, 5,5,6, dst, src, min, max, len |
| 144 | %endif |
| 145 | %if WIN64 |
| 146 | SWAP 0, 2 |
| 147 | SWAP 1, 3 |
| 148 | %elif ARCH_X86_32 |
| 149 | movss m0, minm |
| 150 | movss m1, maxm |
| 151 | %endif |
| 152 | SPLATD m0 |
| 153 | SPLATD m1 |
| 154 | shl lend, 2 |
| 155 | add srcq, lenq |
| 156 | add dstq, lenq |
| 157 | neg lenq |
| 158 | .loop: |
| 159 | mova m2, [srcq+lenq+mmsize*0] |
| 160 | mova m3, [srcq+lenq+mmsize*1] |
| 161 | mova m4, [srcq+lenq+mmsize*2] |
| 162 | mova m5, [srcq+lenq+mmsize*3] |
| 163 | maxps m2, m0 |
| 164 | maxps m3, m0 |
| 165 | maxps m4, m0 |
| 166 | maxps m5, m0 |
| 167 | minps m2, m1 |
| 168 | minps m3, m1 |
| 169 | minps m4, m1 |
| 170 | minps m5, m1 |
| 171 | mova [dstq+lenq+mmsize*0], m2 |
| 172 | mova [dstq+lenq+mmsize*1], m3 |
| 173 | mova [dstq+lenq+mmsize*2], m4 |
| 174 | mova [dstq+lenq+mmsize*3], m5 |
| 175 | add lenq, mmsize*4 |
| 176 | jl .loop |
| 177 | REP_RET |