RepoMirrors
/
ffmpeg
mirror of https://git.ffmpeg.org/ffmpeg.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
ffmpeg/libavcodec/hpeldsp.c

369 lines
23 KiB
C
Raw Normal View History

hpeldsp: add half-pel functions (currently copies of dsputil).
2013-03-09 23:14:35 +00:00
/*
* Half-pel DSP functions.
* Copyright (c) 2000, 2001 Fabrice Bellard
* Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
*
* gmc & q-pel & 32/64 bit based MC by Michael Niedermayer <michaelni@gmx.at>
*
* 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
*/
/**
* @file
* Half-pel DSP functions.
*/
#include "libavutil/attributes.h"
hpeldsp: Add half-pel functions (currently copies of dsputil) Signed-off-by: Martin Storsjö <martin@martin.st>
2013-03-09 23:14:35 +00:00
#include "libavutil/intreadwrite.h"
hpeldsp: add half-pel functions (currently copies of dsputil).
2013-03-09 23:14:35 +00:00
#include "hpeldsp.h"
#define BIT_DEPTH 8
hpeldsp_template: Detemplatize the code The indirection makes no sense without multiple instantiation.
2014-01-10 13:57:59 +00:00
#include "hpel_template.c"
Rename tpel_template.c ---> pel_template.c The new name more accurately describes what the file is about.
2014-06-12 14:05:49 +00:00
#include "pel_template.c"
hpeldsp_template: Move content to hpeldsp There is no point in having this separate; it is not used as a template.
2014-01-10 14:01:04 +00:00
#define PIXOP2(OPNAME, OP) \
static inline void OPNAME ## _no_rnd_pixels8_l2_8(uint8_t *dst, \
const uint8_t *src1, \
const uint8_t *src2, \
int dst_stride, \
int src_stride1, \
int src_stride2, \
int h) \
{ \
int i; \
\
for (i = 0; i < h; i++) { \
uint32_t a, b; \
a = AV_RN32(&src1[i * src_stride1]); \
b = AV_RN32(&src2[i * src_stride2]); \
OP(*((uint32_t *) &dst[i * dst_stride]), \
no_rnd_avg32(a, b)); \
a = AV_RN32(&src1[i * src_stride1 + 4]); \
b = AV_RN32(&src2[i * src_stride2 + 4]); \
OP(*((uint32_t *) &dst[i * dst_stride + 4]), \
no_rnd_avg32(a, b)); \
} \
} \
\
static inline void OPNAME ## _no_rnd_pixels8_x2_8_c(uint8_t *block, \
const uint8_t *pixels, \
ptrdiff_t line_size, \
int h) \
{ \
OPNAME ## _no_rnd_pixels8_l2_8(block, pixels, pixels + 1, \
line_size, line_size, line_size, h); \
} \
\
static inline void OPNAME ## _pixels8_x2_8_c(uint8_t *block, \
const uint8_t *pixels, \
ptrdiff_t line_size, \
int h) \
{ \
OPNAME ## _pixels8_l2_8(block, pixels, pixels + 1, \
line_size, line_size, line_size, h); \
} \
\
static inline void OPNAME ## _no_rnd_pixels8_y2_8_c(uint8_t *block, \
const uint8_t *pixels, \
ptrdiff_t line_size, \
int h) \
{ \
OPNAME ## _no_rnd_pixels8_l2_8(block, pixels, pixels + line_size, \
line_size, line_size, line_size, h); \
} \
\
static inline void OPNAME ## _pixels8_y2_8_c(uint8_t *block, \
const uint8_t *pixels, \
ptrdiff_t line_size, \
int h) \
{ \
OPNAME ## _pixels8_l2_8(block, pixels, pixels + line_size, \
line_size, line_size, line_size, h); \
} \
\
static inline void OPNAME ## _pixels4_x2_8_c(uint8_t *block, \
const uint8_t *pixels, \
ptrdiff_t line_size, \
int h) \
{ \
OPNAME ## _pixels4_l2_8(block, pixels, pixels + 1, \
line_size, line_size, line_size, h); \
} \
\
static inline void OPNAME ## _pixels4_y2_8_c(uint8_t *block, \
const uint8_t *pixels, \
ptrdiff_t line_size, \
int h) \
{ \
OPNAME ## _pixels4_l2_8(block, pixels, pixels + line_size, \
line_size, line_size, line_size, h); \
} \
\
static inline void OPNAME ## _pixels2_x2_8_c(uint8_t *block, \
const uint8_t *pixels, \
ptrdiff_t line_size, \
int h) \
{ \
OPNAME ## _pixels2_l2_8(block, pixels, pixels + 1, \
line_size, line_size, line_size, h); \
} \
\
static inline void OPNAME ## _pixels2_y2_8_c(uint8_t *block, \
const uint8_t *pixels, \
ptrdiff_t line_size, \
int h) \
{ \
OPNAME ## _pixels2_l2_8(block, pixels, pixels + line_size, \
line_size, line_size, line_size, h); \
} \
\
static inline void OPNAME ## _pixels2_xy2_8_c(uint8_t *block, \
const uint8_t *pixels, \
ptrdiff_t line_size, \
int h) \
{ \
int i, a1, b1; \
int a0 = pixels[0]; \
int 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 OPNAME ## _pixels4_xy2_8_c(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 OPNAME ## _pixels8_xy2_8_c(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) + \
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; \
} \
pixels += 4 - line_size * (h + 1); \
block += 4 - line_size * h; \
} \
} \
\
static inline void OPNAME ## _no_rnd_pixels8_xy2_8_c(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(OPNAME ## _pixels16_x2_8_c, \
OPNAME ## _pixels8_x2_8_c, \
8) \
CALL_2X_PIXELS(OPNAME ## _pixels16_y2_8_c, \
OPNAME ## _pixels8_y2_8_c, \
8) \
CALL_2X_PIXELS(OPNAME ## _pixels16_xy2_8_c, \
OPNAME ## _pixels8_xy2_8_c, \
8) \
CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_8_c, \
OPNAME ## _pixels8_8_c, \
8) \
CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_x2_8_c, \
OPNAME ## _no_rnd_pixels8_x2_8_c, \
8) \
CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_y2_8_c, \
OPNAME ## _no_rnd_pixels8_y2_8_c, \
8) \
CALL_2X_PIXELS(OPNAME ## _no_rnd_pixels16_xy2_8_c, \
OPNAME ## _no_rnd_pixels8_xy2_8_c, \
8) \
#define op_avg(a, b) a = rnd_avg32(a, b)
#define op_put(a, b) a = b
#define put_no_rnd_pixels8_8_c put_pixels8_8_c
PIXOP2(avg, op_avg)
PIXOP2(put, op_put)
#undef op_avg
#undef op_put
hpeldsp: add half-pel functions (currently copies of dsputil).
2013-03-09 23:14:35 +00:00
hpeldsp: Add half-pel functions (currently copies of dsputil) Signed-off-by: Martin Storsjö <martin@martin.st>
2013-03-09 23:14:35 +00:00
av_cold void ff_hpeldsp_init(HpelDSPContext *c, int flags)
hpeldsp: add half-pel functions (currently copies of dsputil).
2013-03-09 23:14:35 +00:00
{
#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);
aarch64: hpeldsp NEON optimizations Ported from ARMv7 NEON.
2013-12-20 19:03:58 +00:00
if (ARCH_AARCH64)
ff_hpeldsp_init_aarch64(c, flags);
Merge commit 'bfb41b5039e36b7f873d6ea7d24b31bf3e1a8075' * commit 'bfb41b5039e36b7f873d6ea7d24b31bf3e1a8075': bfin: hpeldsp: Move half-pel assembly from dsputil to hpeldsp Conflicts: libavcodec/bfin/Makefile libavcodec/bfin/hpel_pixels_bfin.S libavcodec/bfin/hpeldsp_bfin.c libavcodec/bfin/hpeldsp_bfin.h libavcodec/hpeldsp.h Merged-by: Michael Niedermayer <michaelni@gmx.at>
2013-04-20 12:29:16 +00:00
if (ARCH_ALPHA)
ff_hpeldsp_init_alpha(c, flags);
arm: hpeldsp: Move half-pel assembly from dsputil to hpeldsp Signed-off-by: Martin Storsjö <martin@martin.st>
2013-03-10 23:16:45 +00:00
if (ARCH_ARM)
ff_hpeldsp_init_arm(c, flags);
ppc: hpeldsp: Move half-pel assembly from dsputil to hpeldsp Signed-off-by: Martin Storsjö <martin@martin.st>
2013-03-10 22:50:53 +00:00
if (ARCH_PPC)
ff_hpeldsp_init_ppc(c, flags);
x86: hpeldsp: Move half-pel assembly from dsputil to hpeldsp Signed-off-by: Martin Storsjö <martin@martin.st>
2013-03-10 22:37:59 +00:00
if (ARCH_X86)
ff_hpeldsp_init_x86(c, flags);
hpeldsp: add half-pel functions (currently copies of dsputil).
2013-03-09 23:14:35 +00:00
}