dsputil: remove hpel functions (moved to hpeldsp).

This commit is contained in:
Ronald S. Bultje 2013-03-10 14:23:46 -07:00 committed by Michael Niedermayer
parent 94b77678dc
commit f536df99fe
4 changed files with 7 additions and 227 deletions

View File

@ -2869,24 +2869,6 @@ av_cold void ff_dsputil_init(DSPContext* c, AVCodecContext *avctx)
c->add_pixels8 = add_pixels8_c;
#define hpel_funcs(prefix, idx, num) \
c->prefix ## _pixels_tab idx [0] = prefix ## _pixels ## num ## _8_c; \
c->prefix ## _pixels_tab idx [1] = prefix ## _pixels ## num ## _x2_8_c; \
c->prefix ## _pixels_tab idx [2] = prefix ## _pixels ## num ## _y2_8_c; \
c->prefix ## _pixels_tab idx [3] = prefix ## _pixels ## num ## _xy2_8_c
hpel_funcs(put, [0], 16);
hpel_funcs(put, [1], 8);
hpel_funcs(put, [2], 4);
hpel_funcs(put, [3], 2);
hpel_funcs(put_no_rnd, [0], 16);
hpel_funcs(put_no_rnd, [1], 8);
hpel_funcs(avg, [0], 16);
hpel_funcs(avg, [1], 8);
hpel_funcs(avg, [2], 4);
hpel_funcs(avg, [3], 2);
hpel_funcs(avg_no_rnd,, 16);
#undef FUNC
#undef FUNCC
#define FUNC(f, depth) f ## _ ## depth

View File

@ -80,7 +80,6 @@ could be reached easily ...
/* add and put pixel (decoding) */
// blocksizes for op_pixels_func are 8x4,8x8 16x8 16x16
//h for op_pixels_func is limited to {width/2, width} but never larger than 16 and never smaller than 4
typedef void (*op_pixels_func)(uint8_t *block/*align width (8 or 16)*/, const uint8_t *pixels/*align 1*/, ptrdiff_t line_size, int h);
typedef void (*tpel_mc_func)(uint8_t *block/*align width (8 or 16)*/, const uint8_t *pixels/*align 1*/, int line_size, int w, int h);
typedef void (*qpel_mc_func)(uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);
@ -179,54 +178,6 @@ typedef struct DSPContext {
int (*ssd_int8_vs_int16)(const int8_t *pix1, const int16_t *pix2,
int size);
/**
* Halfpel motion compensation with rounding (a+b+1)>>1.
* this is an array[4][4] of motion compensation functions for 4
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
* @param block destination where the result is stored
* @param pixels source
* @param line_size number of bytes in a horizontal line of block
* @param h height
*/
op_pixels_func put_pixels_tab[4][4];
/**
* Halfpel motion compensation with rounding (a+b+1)>>1.
* This is an array[4][4] of motion compensation functions for 4
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
* @param block destination into which the result is averaged (a+b+1)>>1
* @param pixels source
* @param line_size number of bytes in a horizontal line of block
* @param h height
*/
op_pixels_func avg_pixels_tab[4][4];
/**
* Halfpel motion compensation with no rounding (a+b)>>1.
* this is an array[2][4] of motion compensation functions for 2
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
* @param block destination where the result is stored
* @param pixels source
* @param line_size number of bytes in a horizontal line of block
* @param h height
*/
op_pixels_func put_no_rnd_pixels_tab[4][4];
/**
* Halfpel motion compensation with no rounding (a+b)>>1.
* this is an array[4] of motion compensation functions for 1
* horizontal blocksize (16) and the 4 halfpel positions<br>
* *pixels_tab[0][ xhalfpel + 2*yhalfpel ]
* @param block destination into which the result is averaged (a+b)>>1
* @param pixels source
* @param line_size number of bytes in a horizontal line of block
* @param h height
*/
op_pixels_func avg_no_rnd_pixels_tab[4];
/**
* Thirdpel motion compensation with rounding (a+b+1)>>1.
* this is an array[12] of motion compensation functions for the 9 thirdpe

View File

@ -132,22 +132,6 @@ static inline void FUNC(OPNAME ## _no_rnd_pixels16_l2)(uint8_t *dst, const uint8
FUNC(OPNAME ## _no_rnd_pixels8_l2)(dst+8*sizeof(pixel), src1+8*sizeof(pixel), src2+8*sizeof(pixel), dst_stride, src_stride1, src_stride2, h);\
}\
\
static inline void FUNCC(OPNAME ## _no_rnd_pixels8_x2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){\
FUNC(OPNAME ## _no_rnd_pixels8_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\
}\
\
static inline void FUNCC(OPNAME ## _pixels8_x2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){\
FUNC(OPNAME ## _pixels8_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\
}\
\
static inline void FUNCC(OPNAME ## _no_rnd_pixels8_y2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){\
FUNC(OPNAME ## _no_rnd_pixels8_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
}\
\
static inline void FUNCC(OPNAME ## _pixels8_y2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){\
FUNC(OPNAME ## _pixels8_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
}\
\
static inline void FUNC(OPNAME ## _pixels8_l4)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, const uint8_t *src3, const uint8_t *src4,\
int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
/* FIXME HIGH BIT DEPTH */\
@ -185,22 +169,6 @@ static inline void FUNC(OPNAME ## _pixels8_l4)(uint8_t *dst, const uint8_t *src1
}\
}\
\
static inline void FUNCC(OPNAME ## _pixels4_x2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){\
FUNC(OPNAME ## _pixels4_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\
}\
\
static inline void FUNCC(OPNAME ## _pixels4_y2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){\
FUNC(OPNAME ## _pixels4_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
}\
\
static inline void FUNCC(OPNAME ## _pixels2_x2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){\
FUNC(OPNAME ## _pixels2_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\
}\
\
static inline void FUNCC(OPNAME ## _pixels2_y2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){\
FUNC(OPNAME ## _pixels2_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
}\
\
static inline void FUNC(OPNAME ## _no_rnd_pixels8_l4)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, const uint8_t *src3, const uint8_t *src4,\
int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
/* FIXME HIGH BIT DEPTH*/\
@ -248,79 +216,6 @@ static inline void FUNC(OPNAME ## _no_rnd_pixels16_l4)(uint8_t *dst, const uint8
FUNC(OPNAME ## _no_rnd_pixels8_l4)(dst+8*sizeof(pixel), src1+8*sizeof(pixel), src2+8*sizeof(pixel), src3+8*sizeof(pixel), src4+8*sizeof(pixel), dst_stride, src_stride1, src_stride2, src_stride3, src_stride4, h);\
}\
\
static inline void FUNCC(OPNAME ## _pixels2_xy2)(uint8_t *_block, const uint8_t *_pixels, ptrdiff_t line_size, int h)\
{\
int i, a0, b0, a1, b1;\
pixel *block = (pixel*)_block;\
const pixel *pixels = (const pixel*)_pixels;\
line_size >>= sizeof(pixel)-1;\
a0= pixels[0];\
b0= pixels[1] + 2;\
a0 += b0;\
b0 += pixels[2];\
\
pixels+=line_size;\
for(i=0; i<h; i+=2){\
a1= pixels[0];\
b1= pixels[1];\
a1 += b1;\
b1 += pixels[2];\
\
block[0]= (a1+a0)>>2; /* FIXME non put */\
block[1]= (b1+b0)>>2;\
\
pixels+=line_size;\
block +=line_size;\
\
a0= pixels[0];\
b0= pixels[1] + 2;\
a0 += b0;\
b0 += pixels[2];\
\
block[0]= (a1+a0)>>2;\
block[1]= (b1+b0)>>2;\
pixels+=line_size;\
block +=line_size;\
}\
}\
\
static inline void FUNCC(OPNAME ## _pixels4_xy2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)\
{\
/* FIXME HIGH BIT DEPTH */\
int i;\
const uint32_t a= AV_RN32(pixels );\
const uint32_t b= AV_RN32(pixels+1);\
uint32_t l0= (a&0x03030303UL)\
+ (b&0x03030303UL)\
+ 0x02020202UL;\
uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
+ ((b&0xFCFCFCFCUL)>>2);\
uint32_t l1,h1;\
\
pixels+=line_size;\
for(i=0; i<h; i+=2){\
uint32_t a= AV_RN32(pixels );\
uint32_t b= AV_RN32(pixels+1);\
l1= (a&0x03030303UL)\
+ (b&0x03030303UL);\
h1= ((a&0xFCFCFCFCUL)>>2)\
+ ((b&0xFCFCFCFCUL)>>2);\
OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
pixels+=line_size;\
block +=line_size;\
a= AV_RN32(pixels );\
b= AV_RN32(pixels+1);\
l0= (a&0x03030303UL)\
+ (b&0x03030303UL)\
+ 0x02020202UL;\
h0= ((a&0xFCFCFCFCUL)>>2)\
+ ((b&0xFCFCFCFCUL)>>2);\
OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
pixels+=line_size;\
block +=line_size;\
}\
}\
\
static inline void FUNCC(OPNAME ## _pixels8_xy2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)\
{\
/* FIXME HIGH BIT DEPTH */\
@ -363,55 +258,7 @@ static inline void FUNCC(OPNAME ## _pixels8_xy2)(uint8_t *block, const uint8_t *
}\
}\
\
static inline void FUNCC(OPNAME ## _no_rnd_pixels8_xy2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)\
{\
/* FIXME HIGH BIT DEPTH */\
int j;\
for(j=0; j<2; j++){\
int i;\
const uint32_t a= AV_RN32(pixels );\
const uint32_t b= AV_RN32(pixels+1);\
uint32_t l0= (a&0x03030303UL)\
+ (b&0x03030303UL)\
+ 0x01010101UL;\
uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
+ ((b&0xFCFCFCFCUL)>>2);\
uint32_t l1,h1;\
\
pixels+=line_size;\
for(i=0; i<h; i+=2){\
uint32_t a= AV_RN32(pixels );\
uint32_t b= AV_RN32(pixels+1);\
l1= (a&0x03030303UL)\
+ (b&0x03030303UL);\
h1= ((a&0xFCFCFCFCUL)>>2)\
+ ((b&0xFCFCFCFCUL)>>2);\
OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
pixels+=line_size;\
block +=line_size;\
a= AV_RN32(pixels );\
b= AV_RN32(pixels+1);\
l0= (a&0x03030303UL)\
+ (b&0x03030303UL)\
+ 0x01010101UL;\
h0= ((a&0xFCFCFCFCUL)>>2)\
+ ((b&0xFCFCFCFCUL)>>2);\
OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
pixels+=line_size;\
block +=line_size;\
}\
pixels+=4-line_size*(h+1);\
block +=4-line_size*h;\
}\
}\
\
CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16_x2) , FUNCC(OPNAME ## _pixels8_x2) , 8*sizeof(pixel))\
CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16_y2) , FUNCC(OPNAME ## _pixels8_y2) , 8*sizeof(pixel))\
CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16_xy2), FUNCC(OPNAME ## _pixels8_xy2), 8*sizeof(pixel))\
av_unused CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16) , FUNCC(OPNAME ## _pixels8) , 8*sizeof(pixel))\
CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_x2) , FUNCC(OPNAME ## _no_rnd_pixels8_x2) , 8*sizeof(pixel))\
CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_y2) , FUNCC(OPNAME ## _no_rnd_pixels8_y2) , 8*sizeof(pixel))\
CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_xy2), FUNCC(OPNAME ## _no_rnd_pixels8_xy2), 8*sizeof(pixel))\
#define op_avg(a, b) a = rnd_avg_pixel4(a, b)
#define op_put(a, b) a = b

View File

@ -35,9 +35,9 @@
// blocksizes for op_pixels_func are 8x4,8x8 16x8 16x16
// h for op_pixels_func is limited to {width/2, width} but never larger
// than 16 and never smaller than 4
typedef void (*hpel_pixels_func)(uint8_t *block /*align width (8 or 16)*/,
const uint8_t *pixels /*align 1*/,
ptrdiff_t line_size, int h);
typedef void (*op_pixels_func)(uint8_t *block /*align width (8 or 16)*/,
const uint8_t *pixels /*align 1*/,
ptrdiff_t line_size, int h);
/**
* Half-pel DSP context.
@ -53,7 +53,7 @@ typedef struct HpelDSPContext {
* @param line_size number of bytes in a horizontal line of block
* @param h height
*/
hpel_pixels_func put_pixels_tab[4][4];
op_pixels_func put_pixels_tab[4][4];
/**
* Halfpel motion compensation with rounding (a+b+1)>>1.
@ -65,7 +65,7 @@ typedef struct HpelDSPContext {
* @param line_size number of bytes in a horizontal line of block
* @param h height
*/
hpel_pixels_func avg_pixels_tab[4][4];
op_pixels_func avg_pixels_tab[4][4];
/**
* Halfpel motion compensation with no rounding (a+b)>>1.
@ -77,7 +77,7 @@ typedef struct HpelDSPContext {
* @param line_size number of bytes in a horizontal line of block
* @param h height
*/
hpel_pixels_func put_no_rnd_pixels_tab[2][4];
op_pixels_func put_no_rnd_pixels_tab[2][4];
/**
* Halfpel motion compensation with no rounding (a+b)>>1.
@ -89,7 +89,7 @@ typedef struct HpelDSPContext {
* @param line_size number of bytes in a horizontal line of block
* @param h height
*/
hpel_pixels_func avg_no_rnd_pixels_tab[4];
op_pixels_func avg_no_rnd_pixels_tab[4];
} HpelDSPContext;
void ff_hpeldsp_init(HpelDSPContext* p, int flags);