📄 dsputil_mmx_avg.h
字号:
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r" ((long)line_size)
:"%"REG_a, "memory");
}
static void DEF(put_pixels8_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{
__asm __volatile(
"lea (%3, %3), %%"REG_a" \n\t"
"movq (%1), %%mm0 \n\t"
"sub %3, %2 \n\t"
"1: \n\t"
"movq (%1, %3), %%mm1 \n\t"
"movq (%1, %%"REG_a"), %%mm2 \n\t"
"add %%"REG_a", %1 \n\t"
PAVGB" %%mm1, %%mm0 \n\t"
PAVGB" %%mm2, %%mm1 \n\t"
"movq %%mm0, (%2, %3) \n\t"
"movq %%mm1, (%2, %%"REG_a") \n\t"
"movq (%1, %3), %%mm1 \n\t"
"movq (%1, %%"REG_a"), %%mm0 \n\t"
"add %%"REG_a", %2 \n\t"
"add %%"REG_a", %1 \n\t"
PAVGB" %%mm1, %%mm2 \n\t"
PAVGB" %%mm0, %%mm1 \n\t"
"movq %%mm2, (%2, %3) \n\t"
"movq %%mm1, (%2, %%"REG_a") \n\t"
"add %%"REG_a", %2 \n\t"
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D" (block)
:"r" ((long)line_size)
:"%"REG_a, "memory");
}
/* GL: this function does incorrect rounding if overflow */
static void DEF(put_no_rnd_pixels8_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{
MOVQ_BONE(mm6);
__asm __volatile(
"lea (%3, %3), %%"REG_a" \n\t"
"movq (%1), %%mm0 \n\t"
"sub %3, %2 \n\t"
"1: \n\t"
"movq (%1, %3), %%mm1 \n\t"
"movq (%1, %%"REG_a"), %%mm2 \n\t"
"add %%"REG_a", %1 \n\t"
"psubusb %%mm6, %%mm1 \n\t"
PAVGB" %%mm1, %%mm0 \n\t"
PAVGB" %%mm2, %%mm1 \n\t"
"movq %%mm0, (%2, %3) \n\t"
"movq %%mm1, (%2, %%"REG_a") \n\t"
"movq (%1, %3), %%mm1 \n\t"
"movq (%1, %%"REG_a"), %%mm0 \n\t"
"add %%"REG_a", %2 \n\t"
"add %%"REG_a", %1 \n\t"
"psubusb %%mm6, %%mm1 \n\t"
PAVGB" %%mm1, %%mm2 \n\t"
PAVGB" %%mm0, %%mm1 \n\t"
"movq %%mm2, (%2, %3) \n\t"
"movq %%mm1, (%2, %%"REG_a") \n\t"
"add %%"REG_a", %2 \n\t"
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D" (block)
:"r" ((long)line_size)
:"%"REG_a, "memory");
}
static void DEF(avg_pixels8)(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{
__asm __volatile(
"lea (%3, %3), %%"REG_a" \n\t"
"1: \n\t"
"movq (%2), %%mm0 \n\t"
"movq (%2, %3), %%mm1 \n\t"
PAVGB" (%1), %%mm0 \n\t"
PAVGB" (%1, %3), %%mm1 \n\t"
"movq %%mm0, (%2) \n\t"
"movq %%mm1, (%2, %3) \n\t"
"add %%"REG_a", %1 \n\t"
"add %%"REG_a", %2 \n\t"
"movq (%2), %%mm0 \n\t"
"movq (%2, %3), %%mm1 \n\t"
PAVGB" (%1), %%mm0 \n\t"
PAVGB" (%1, %3), %%mm1 \n\t"
"add %%"REG_a", %1 \n\t"
"movq %%mm0, (%2) \n\t"
"movq %%mm1, (%2, %3) \n\t"
"add %%"REG_a", %2 \n\t"
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r" ((long)line_size)
:"%"REG_a, "memory");
}
static void DEF(avg_pixels8_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{
__asm __volatile(
"lea (%3, %3), %%"REG_a" \n\t"
"1: \n\t"
"movq (%1), %%mm0 \n\t"
"movq (%1, %3), %%mm2 \n\t"
PAVGB" 1(%1), %%mm0 \n\t"
PAVGB" 1(%1, %3), %%mm2 \n\t"
PAVGB" (%2), %%mm0 \n\t"
PAVGB" (%2, %3), %%mm2 \n\t"
"add %%"REG_a", %1 \n\t"
"movq %%mm0, (%2) \n\t"
"movq %%mm2, (%2, %3) \n\t"
"movq (%1), %%mm0 \n\t"
"movq (%1, %3), %%mm2 \n\t"
PAVGB" 1(%1), %%mm0 \n\t"
PAVGB" 1(%1, %3), %%mm2 \n\t"
"add %%"REG_a", %2 \n\t"
"add %%"REG_a", %1 \n\t"
PAVGB" (%2), %%mm0 \n\t"
PAVGB" (%2, %3), %%mm2 \n\t"
"movq %%mm0, (%2) \n\t"
"movq %%mm2, (%2, %3) \n\t"
"add %%"REG_a", %2 \n\t"
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r" ((long)line_size)
:"%"REG_a, "memory");
}
static void DEF(avg_pixels8_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{
__asm __volatile(
"lea (%3, %3), %%"REG_a" \n\t"
"movq (%1), %%mm0 \n\t"
"sub %3, %2 \n\t"
"1: \n\t"
"movq (%1, %3), %%mm1 \n\t"
"movq (%1, %%"REG_a"), %%mm2 \n\t"
"add %%"REG_a", %1 \n\t"
PAVGB" %%mm1, %%mm0 \n\t"
PAVGB" %%mm2, %%mm1 \n\t"
"movq (%2, %3), %%mm3 \n\t"
"movq (%2, %%"REG_a"), %%mm4 \n\t"
PAVGB" %%mm3, %%mm0 \n\t"
PAVGB" %%mm4, %%mm1 \n\t"
"movq %%mm0, (%2, %3) \n\t"
"movq %%mm1, (%2, %%"REG_a") \n\t"
"movq (%1, %3), %%mm1 \n\t"
"movq (%1, %%"REG_a"), %%mm0 \n\t"
PAVGB" %%mm1, %%mm2 \n\t"
PAVGB" %%mm0, %%mm1 \n\t"
"add %%"REG_a", %2 \n\t"
"add %%"REG_a", %1 \n\t"
"movq (%2, %3), %%mm3 \n\t"
"movq (%2, %%"REG_a"), %%mm4 \n\t"
PAVGB" %%mm3, %%mm2 \n\t"
PAVGB" %%mm4, %%mm1 \n\t"
"movq %%mm2, (%2, %3) \n\t"
"movq %%mm1, (%2, %%"REG_a") \n\t"
"add %%"REG_a", %2 \n\t"
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r" ((long)line_size)
:"%"REG_a, "memory");
}
/* Note this is not correctly rounded, but this function is only
* used for B-frames so it does not matter. */
static void DEF(avg_pixels8_xy2)(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{
MOVQ_BONE(mm6);
__asm __volatile(
"lea (%3, %3), %%"REG_a" \n\t"
"movq (%1), %%mm0 \n\t"
PAVGB" 1(%1), %%mm0 \n\t"
ASMALIGN(3)
"1: \n\t"
"movq (%1, %%"REG_a"), %%mm2 \n\t"
"movq (%1, %3), %%mm1 \n\t"
"psubusb %%mm6, %%mm2 \n\t"
PAVGB" 1(%1, %3), %%mm1 \n\t"
PAVGB" 1(%1, %%"REG_a"), %%mm2 \n\t"
"add %%"REG_a", %1 \n\t"
PAVGB" %%mm1, %%mm0 \n\t"
PAVGB" %%mm2, %%mm1 \n\t"
PAVGB" (%2), %%mm0 \n\t"
PAVGB" (%2, %3), %%mm1 \n\t"
"movq %%mm0, (%2) \n\t"
"movq %%mm1, (%2, %3) \n\t"
"movq (%1, %3), %%mm1 \n\t"
"movq (%1, %%"REG_a"), %%mm0 \n\t"
PAVGB" 1(%1, %3), %%mm1 \n\t"
PAVGB" 1(%1, %%"REG_a"), %%mm0 \n\t"
"add %%"REG_a", %2 \n\t"
"add %%"REG_a", %1 \n\t"
PAVGB" %%mm1, %%mm2 \n\t"
PAVGB" %%mm0, %%mm1 \n\t"
PAVGB" (%2), %%mm2 \n\t"
PAVGB" (%2, %3), %%mm1 \n\t"
"movq %%mm2, (%2) \n\t"
"movq %%mm1, (%2, %3) \n\t"
"add %%"REG_a", %2 \n\t"
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r" ((long)line_size)
:"%"REG_a, "memory");
}
//FIXME the following could be optimized too ...
static void DEF(put_no_rnd_pixels16_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){
DEF(put_no_rnd_pixels8_x2)(block , pixels , line_size, h);
DEF(put_no_rnd_pixels8_x2)(block+8, pixels+8, line_size, h);
}
static void DEF(put_pixels16_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){
DEF(put_pixels8_y2)(block , pixels , line_size, h);
DEF(put_pixels8_y2)(block+8, pixels+8, line_size, h);
}
static void DEF(put_no_rnd_pixels16_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){
DEF(put_no_rnd_pixels8_y2)(block , pixels , line_size, h);
DEF(put_no_rnd_pixels8_y2)(block+8, pixels+8, line_size, h);
}
static void DEF(avg_pixels16)(uint8_t *block, const uint8_t *pixels, int line_size, int h){
DEF(avg_pixels8)(block , pixels , line_size, h);
DEF(avg_pixels8)(block+8, pixels+8, line_size, h);
}
static void DEF(avg_pixels16_x2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){
DEF(avg_pixels8_x2)(block , pixels , line_size, h);
DEF(avg_pixels8_x2)(block+8, pixels+8, line_size, h);
}
static void DEF(avg_pixels16_y2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){
DEF(avg_pixels8_y2)(block , pixels , line_size, h);
DEF(avg_pixels8_y2)(block+8, pixels+8, line_size, h);
}
static void DEF(avg_pixels16_xy2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){
DEF(avg_pixels8_xy2)(block , pixels , line_size, h);
DEF(avg_pixels8_xy2)(block+8, pixels+8, line_size, h);
}
#define QPEL_2TAP_L3(OPNAME) \
static void DEF(OPNAME ## 2tap_qpel16_l3)(uint8_t *dst, uint8_t *src, int stride, int h, int off1, int off2){\
asm volatile(\
"1: \n\t"\
"movq (%1,%2), %%mm0 \n\t"\
"movq 8(%1,%2), %%mm1 \n\t"\
PAVGB" (%1,%3), %%mm0 \n\t"\
PAVGB" 8(%1,%3), %%mm1 \n\t"\
PAVGB" (%1), %%mm0 \n\t"\
PAVGB" 8(%1), %%mm1 \n\t"\
STORE_OP( (%1,%4),%%mm0)\
STORE_OP(8(%1,%4),%%mm1)\
"movq %%mm0, (%1,%4) \n\t"\
"movq %%mm1, 8(%1,%4) \n\t"\
"add %5, %1 \n\t"\
"decl %0 \n\t"\
"jnz 1b \n\t"\
:"+g"(h), "+r"(src)\
:"r"((long)off1), "r"((long)off2),\
"r"((long)(dst-src)), "r"((long)stride)\
:"memory"\
);\
}\
static void DEF(OPNAME ## 2tap_qpel8_l3)(uint8_t *dst, uint8_t *src, int stride, int h, int off1, int off2){\
asm volatile(\
"1: \n\t"\
"movq (%1,%2), %%mm0 \n\t"\
PAVGB" (%1,%3), %%mm0 \n\t"\
PAVGB" (%1), %%mm0 \n\t"\
STORE_OP((%1,%4),%%mm0)\
"movq %%mm0, (%1,%4) \n\t"\
"add %5, %1 \n\t"\
"decl %0 \n\t"\
"jnz 1b \n\t"\
:"+g"(h), "+r"(src)\
:"r"((long)off1), "r"((long)off2),\
"r"((long)(dst-src)), "r"((long)stride)\
:"memory"\
);\
}
#define STORE_OP(a,b) PAVGB" "#a","#b" \n\t"
QPEL_2TAP_L3(avg_)
#undef STORE_OP
#define STORE_OP(a,b)
QPEL_2TAP_L3(put_)
#undef STORE_OP
#undef QPEL_2TAP_L3
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -