mirror of https://github.com/FFmpeg/FFmpeg.git
Signed-off-by: James Almer <jamrial@gmail.com>pull/247/head
James Almer
8 years ago
parent
30c1f27299
commit
cf9ef83960
16 changed files with 378 additions and 242 deletions
@ -0,0 +1,84 @@ |
||||
/*
|
||||
* 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 |
||||
*/ |
||||
|
||||
#include "config.h" |
||||
#include "libavutil/attributes.h" |
||||
#include "lossless_videoencdsp.h" |
||||
#include "mathops.h" |
||||
|
||||
// 0x7f7f7f7f or 0x7f7f7f7f7f7f7f7f or whatever, depending on the cpu's native arithmetic size
|
||||
#define pb_7f (~0UL / 255 * 0x7f) |
||||
#define pb_80 (~0UL / 255 * 0x80) |
||||
|
||||
static void diff_bytes_c(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, intptr_t w) |
||||
{ |
||||
long i; |
||||
|
||||
#if !HAVE_FAST_UNALIGNED |
||||
if (((long)src1 | (long)src2) & (sizeof(long) - 1)) { |
||||
for (i = 0; i + 7 < w; i += 8) { |
||||
dst[i + 0] = src1[i + 0] - src2[i + 0]; |
||||
dst[i + 1] = src1[i + 1] - src2[i + 1]; |
||||
dst[i + 2] = src1[i + 2] - src2[i + 2]; |
||||
dst[i + 3] = src1[i + 3] - src2[i + 3]; |
||||
dst[i + 4] = src1[i + 4] - src2[i + 4]; |
||||
dst[i + 5] = src1[i + 5] - src2[i + 5]; |
||||
dst[i + 6] = src1[i + 6] - src2[i + 6]; |
||||
dst[i + 7] = src1[i + 7] - src2[i + 7]; |
||||
} |
||||
} else |
||||
#endif |
||||
for (i = 0; i <= w - (int) sizeof(long); i += sizeof(long)) { |
||||
long a = *(long *) (src1 + i); |
||||
long b = *(long *) (src2 + i); |
||||
*(long *) (dst + i) = ((a | pb_80) - (b & pb_7f)) ^ |
||||
((a ^ b ^ pb_80) & pb_80); |
||||
} |
||||
for (; i < w; i++) |
||||
dst[i + 0] = src1[i + 0] - src2[i + 0]; |
||||
} |
||||
|
||||
static void sub_median_pred_c(uint8_t *dst, const uint8_t *src1, |
||||
const uint8_t *src2, intptr_t w, |
||||
int *left, int *left_top) |
||||
{ |
||||
int i; |
||||
uint8_t l, lt; |
||||
|
||||
l = *left; |
||||
lt = *left_top; |
||||
|
||||
for (i = 0; i < w; i++) { |
||||
const int pred = mid_pred(l, src1[i], (l + src1[i] - lt) & 0xFF); |
||||
lt = src1[i]; |
||||
l = src2[i]; |
||||
dst[i] = l - pred; |
||||
} |
||||
|
||||
*left = l; |
||||
*left_top = lt; |
||||
} |
||||
|
||||
av_cold void ff_llvidencdsp_init(LLVidEncDSPContext *c) |
||||
{ |
||||
c->diff_bytes = diff_bytes_c; |
||||
c->sub_median_pred = sub_median_pred_c; |
||||
|
||||
if (ARCH_X86) |
||||
ff_llvidencdsp_init_x86(c); |
||||
} |
||||
@ -0,0 +1,41 @@ |
||||
/*
|
||||
* 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 |
||||
*/ |
||||
|
||||
#ifndef AVCODEC_LOSSLESS_VIDEOENCDSP_H |
||||
#define AVCODEC_LOSSLESS_VIDEOENCDSP_H |
||||
|
||||
#include <stdint.h> |
||||
|
||||
typedef struct LLVidEncDSPContext { |
||||
void (*diff_bytes)(uint8_t *dst /* align 16 */, |
||||
const uint8_t *src1 /* align 16 */, |
||||
const uint8_t *src2 /* align 1 */, |
||||
intptr_t w); |
||||
/**
|
||||
* Subtract HuffYUV's variant of median prediction. |
||||
* Note, this might read from src1[-1], src2[-1]. |
||||
*/ |
||||
void (*sub_median_pred)(uint8_t *dst, const uint8_t *src1, |
||||
const uint8_t *src2, intptr_t w, |
||||
int *left, int *left_top); |
||||
} LLVidEncDSPContext; |
||||
|
||||
void ff_llvidencdsp_init(LLVidEncDSPContext *c); |
||||
void ff_llvidencdsp_init_x86(LLVidEncDSPContext *c); |
||||
|
||||
#endif /* AVCODEC_LOSSLESS_VIDEOENCDSP_H */ |
||||
@ -0,0 +1,54 @@ |
||||
/*
|
||||
* SIMD-optimized HuffYUV encoding functions |
||||
* Copyright (c) 2000, 2001 Fabrice Bellard |
||||
* Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at> |
||||
* |
||||
* MMX optimization by Nick Kurshev <nickols_k@mail.ru> |
||||
* |
||||
* 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 |
||||
*/ |
||||
|
||||
#include "libavutil/attributes.h" |
||||
#include "libavutil/cpu.h" |
||||
#include "libavutil/pixdesc.h" |
||||
#include "libavutil/x86/cpu.h" |
||||
#include "libavcodec/huffyuvencdsp.h" |
||||
|
||||
void ff_diff_int16_mmx (uint16_t *dst, const uint16_t *src1, const uint16_t *src2, |
||||
unsigned mask, int w); |
||||
void ff_diff_int16_sse2(uint16_t *dst, const uint16_t *src1, const uint16_t *src2, |
||||
unsigned mask, int w); |
||||
void ff_sub_hfyu_median_pred_int16_mmxext(uint16_t *dst, const uint16_t *src1, const uint16_t *src2, |
||||
unsigned mask, int w, int *left, int *left_top); |
||||
|
||||
av_cold void ff_huffyuvencdsp_init_x86(HuffYUVEncDSPContext *c, AVCodecContext *avctx) |
||||
{ |
||||
av_unused int cpu_flags = av_get_cpu_flags(); |
||||
const AVPixFmtDescriptor *pix_desc = av_pix_fmt_desc_get(avctx->pix_fmt); |
||||
|
||||
if (ARCH_X86_32 && EXTERNAL_MMX(cpu_flags)) { |
||||
c->diff_int16 = ff_diff_int16_mmx; |
||||
} |
||||
|
||||
if (EXTERNAL_MMXEXT(cpu_flags) && pix_desc && pix_desc->comp[0].depth<16) { |
||||
c->sub_hfyu_median_pred_int16 = ff_sub_hfyu_median_pred_int16_mmxext; |
||||
} |
||||
|
||||
if (EXTERNAL_SSE2(cpu_flags)) { |
||||
c->diff_int16 = ff_diff_int16_sse2; |
||||
} |
||||
} |
||||
@ -0,0 +1,150 @@ |
||||
;************************************************************************ |
||||
;* SIMD-optimized lossless video encoding functions |
||||
;* Copyright (c) 2000, 2001 Fabrice Bellard |
||||
;* Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at> |
||||
;* |
||||
;* MMX optimization by Nick Kurshev <nickols_k@mail.ru> |
||||
;* Conversion to NASM format by Tiancheng "Timothy" Gu <timothygu99@gmail.com> |
||||
;* |
||||
;* 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 |
||||
;* 51, Inc., Foundation Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
||||
;****************************************************************************** |
||||
|
||||
%include "libavutil/x86/x86util.asm" |
||||
|
||||
section .text |
||||
|
||||
; void ff_diff_bytes(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, |
||||
; intptr_t w); |
||||
%macro DIFF_BYTES_PROLOGUE 0 |
||||
%if ARCH_X86_32 |
||||
cglobal diff_bytes, 3,5,2, dst, src1, src2 |
||||
%define wq r4q |
||||
DECLARE_REG_TMP 3 |
||||
mov wq, r3mp |
||||
%else |
||||
cglobal diff_bytes, 4,5,2, dst, src1, src2, w |
||||
DECLARE_REG_TMP 4 |
||||
%endif ; ARCH_X86_32 |
||||
%define i t0q |
||||
%endmacro |
||||
|
||||
; label to jump to if w < regsize |
||||
%macro DIFF_BYTES_LOOP_PREP 1 |
||||
mov i, wq |
||||
and i, -2 * regsize |
||||
jz %1 |
||||
add dstq, i |
||||
add src1q, i |
||||
add src2q, i |
||||
neg i |
||||
%endmacro |
||||
|
||||
; mov type used for src1q, dstq, first reg, second reg |
||||
%macro DIFF_BYTES_LOOP_CORE 4 |
||||
%if mmsize != 16 |
||||
mov%1 %3, [src1q + i] |
||||
mov%1 %4, [src1q + i + regsize] |
||||
psubb %3, [src2q + i] |
||||
psubb %4, [src2q + i + regsize] |
||||
mov%2 [dstq + i], %3 |
||||
mov%2 [regsize + dstq + i], %4 |
||||
%else |
||||
; SSE enforces alignment of psubb operand |
||||
mov%1 %3, [src1q + i] |
||||
movu %4, [src2q + i] |
||||
psubb %3, %4 |
||||
mov%2 [dstq + i], %3 |
||||
mov%1 %3, [src1q + i + regsize] |
||||
movu %4, [src2q + i + regsize] |
||||
psubb %3, %4 |
||||
mov%2 [regsize + dstq + i], %3 |
||||
%endif |
||||
%endmacro |
||||
|
||||
%macro DIFF_BYTES_BODY 2 ; mov type used for src1q, for dstq |
||||
%define regsize mmsize |
||||
.loop_%1%2: |
||||
DIFF_BYTES_LOOP_CORE %1, %2, m0, m1 |
||||
add i, 2 * regsize |
||||
jl .loop_%1%2 |
||||
.skip_main_%1%2: |
||||
and wq, 2 * regsize - 1 |
||||
jz .end_%1%2 |
||||
%if mmsize > 16 |
||||
; fall back to narrower xmm |
||||
%define regsize mmsize / 2 |
||||
DIFF_BYTES_LOOP_PREP .setup_loop_gpr_aa |
||||
.loop2_%1%2: |
||||
DIFF_BYTES_LOOP_CORE %1, %2, xm0, xm1 |
||||
add i, 2 * regsize |
||||
jl .loop2_%1%2 |
||||
.setup_loop_gpr_%1%2: |
||||
and wq, 2 * regsize - 1 |
||||
jz .end_%1%2 |
||||
%endif |
||||
add dstq, wq |
||||
add src1q, wq |
||||
add src2q, wq |
||||
neg wq |
||||
.loop_gpr_%1%2: |
||||
mov t0b, [src1q + wq] |
||||
sub t0b, [src2q + wq] |
||||
mov [dstq + wq], t0b |
||||
inc wq |
||||
jl .loop_gpr_%1%2 |
||||
.end_%1%2: |
||||
REP_RET |
||||
%endmacro |
||||
|
||||
%if ARCH_X86_32 |
||||
INIT_MMX mmx |
||||
DIFF_BYTES_PROLOGUE |
||||
%define regsize mmsize |
||||
DIFF_BYTES_LOOP_PREP .skip_main_aa |
||||
DIFF_BYTES_BODY a, a |
||||
%undef i |
||||
%endif |
||||
|
||||
INIT_XMM sse2 |
||||
DIFF_BYTES_PROLOGUE |
||||
%define regsize mmsize |
||||
DIFF_BYTES_LOOP_PREP .skip_main_aa |
||||
test dstq, regsize - 1 |
||||
jnz .loop_uu |
||||
test src1q, regsize - 1 |
||||
jnz .loop_ua |
||||
DIFF_BYTES_BODY a, a |
||||
DIFF_BYTES_BODY u, a |
||||
DIFF_BYTES_BODY u, u |
||||
%undef i |
||||
|
||||
%if HAVE_AVX2_EXTERNAL |
||||
INIT_YMM avx2 |
||||
DIFF_BYTES_PROLOGUE |
||||
%define regsize mmsize |
||||
; Directly using unaligned SSE2 version is marginally faster than |
||||
; branching based on arguments. |
||||
DIFF_BYTES_LOOP_PREP .skip_main_uu |
||||
test dstq, regsize - 1 |
||||
jnz .loop_uu |
||||
test src1q, regsize - 1 |
||||
jnz .loop_ua |
||||
DIFF_BYTES_BODY a, a |
||||
DIFF_BYTES_BODY u, a |
||||
DIFF_BYTES_BODY u, u |
||||
%undef i |
||||
%endif |
||||
Loading…
Reference in new issue