;* Copyright (c) 2004-2005 Michael Niedermayer, Loren Merritt
;* Copyright (C) 2010 Eli Friedman <eli.friedman@gmail.com>
;*
-;* This file is part of FFmpeg.
+;* This file is part of Libav.
;*
-;* FFmpeg is free software; you can redistribute it and/or
+;* Libav 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,
+;* Libav 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
+;* License along with Libav; if not, write to the Free Software
+;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
;******************************************************************************
-%include "x86inc.asm"
+%include "libavutil/x86/x86util.asm"
SECTION .text
;-----------------------------------------------------------------------------
; biweight pred:
;
-; void h264_biweight_16x16_sse2(uint8_t *dst, uint8_t *src, int stride,
-; int log2_denom, int weightd, int weights,
-; int offset);
+; void h264_biweight_16_sse2(uint8_t *dst, uint8_t *src, int stride,
+; int height, int log2_denom, int weightd,
+; int weights, int offset);
; and
-; void h264_weight_16x16_sse2(uint8_t *dst, int stride,
-; int log2_denom, int weight,
-; int offset);
+; void h264_weight_16_sse2(uint8_t *dst, int stride, int height,
+; int log2_denom, int weight, int offset);
;-----------------------------------------------------------------------------
%macro WEIGHT_SETUP 0
- add r4, r4
- inc r4
- movd m3, r3
- movd m5, r4
- movd m6, r2
+ add r5, r5
+ inc r5
+ movd m3, r4d
+ movd m5, r5d
+ movd m6, r3d
pslld m5, m6
psrld m5, 1
%if mmsize == 16
packuswb m0, m1
%endmacro
-%macro WEIGHT_FUNC_DBL_MM 1
-cglobal h264_weight_16x%1_mmx2, 5, 5, 0
+INIT_MMX mmxext
+cglobal h264_weight_16, 6, 6, 0
WEIGHT_SETUP
- mov r2, %1
-%if %1 == 16
-.nextrow
+.nextrow:
WEIGHT_OP 0, 4
mova [r0 ], m0
WEIGHT_OP 8, 12
mova [r0+8], m0
add r0, r1
- dec r2
+ dec r2d
jnz .nextrow
REP_RET
-%else
- jmp mangle(ff_h264_weight_16x16_mmx2.nextrow)
-%endif
-%endmacro
-INIT_MMX
-WEIGHT_FUNC_DBL_MM 16
-WEIGHT_FUNC_DBL_MM 8
-
-%macro WEIGHT_FUNC_MM 4
-cglobal h264_weight_%1x%2_%4, 7, 7, %3
+%macro WEIGHT_FUNC_MM 2
+cglobal h264_weight_%1, 6, 6, %2
WEIGHT_SETUP
- mov r2, %2
-%if %2 == 16
-.nextrow
+.nextrow:
WEIGHT_OP 0, mmsize/2
mova [r0], m0
add r0, r1
- dec r2
+ dec r2d
jnz .nextrow
REP_RET
-%else
- jmp mangle(ff_h264_weight_%1x16_%4.nextrow)
-%endif
%endmacro
-INIT_MMX
-WEIGHT_FUNC_MM 8, 16, 0, mmx2
-WEIGHT_FUNC_MM 8, 8, 0, mmx2
-WEIGHT_FUNC_MM 8, 4, 0, mmx2
-INIT_XMM
-WEIGHT_FUNC_MM 16, 16, 8, sse2
-WEIGHT_FUNC_MM 16, 8, 8, sse2
+INIT_MMX mmxext
+WEIGHT_FUNC_MM 8, 0
+INIT_XMM sse2
+WEIGHT_FUNC_MM 16, 8
-%macro WEIGHT_FUNC_HALF_MM 5
-cglobal h264_weight_%1x%2_%5, 5, 5, %4
+%macro WEIGHT_FUNC_HALF_MM 2
+cglobal h264_weight_%1, 6, 6, %2
WEIGHT_SETUP
- mov r2, %2/2
+ sar r2d, 1
lea r3, [r1*2]
-%if %2 == mmsize
-.nextrow
+.nextrow:
WEIGHT_OP 0, r1
movh [r0], m0
%if mmsize == 16
movh [r0+r1], m0
%endif
add r0, r3
- dec r2
+ dec r2d
jnz .nextrow
REP_RET
-%else
- jmp mangle(ff_h264_weight_%1x%3_%5.nextrow)
-%endif
%endmacro
-INIT_MMX
-WEIGHT_FUNC_HALF_MM 4, 8, 8, 0, mmx2
-WEIGHT_FUNC_HALF_MM 4, 4, 8, 0, mmx2
-WEIGHT_FUNC_HALF_MM 4, 2, 8, 0, mmx2
-INIT_XMM
-WEIGHT_FUNC_HALF_MM 8, 16, 16, 8, sse2
-WEIGHT_FUNC_HALF_MM 8, 8, 16, 8, sse2
-WEIGHT_FUNC_HALF_MM 8, 4, 16, 8, sse2
+INIT_MMX mmxext
+WEIGHT_FUNC_HALF_MM 4, 0
+INIT_XMM sse2
+WEIGHT_FUNC_HALF_MM 8, 8
%macro BIWEIGHT_SETUP 0
- add r6, 1
- or r6, 1
- add r3, 1
- movd m3, r4
- movd m4, r5
- movd m5, r6
- movd m6, r3
+%if ARCH_X86_64
+%define off_regd r7d
+%else
+%define off_regd r3d
+%endif
+ mov off_regd, r7m
+ add off_regd, 1
+ or off_regd, 1
+ add r4, 1
+%if cpuflag(ssse3)
+ movd m4, r5d
+ movd m0, r6d
+%else
+ movd m3, r5d
+ movd m4, r6d
+%endif
+ movd m5, off_regd
+ movd m6, r4d
pslld m5, m6
psrld m5, 1
+%if cpuflag(ssse3)
+ punpcklbw m4, m0
+ pshuflw m4, m4, 0
+ pshuflw m5, m5, 0
+ punpcklqdq m4, m4
+ punpcklqdq m5, m5
+
+%else
%if mmsize == 16
pshuflw m3, m3, 0
pshuflw m4, m4, 0
pshufw m5, m5, 0
%endif
pxor m7, m7
+%endif
%endmacro
%macro BIWEIGHT_STEPA 3
packuswb m0, m1
%endmacro
-%macro BIWEIGHT_FUNC_DBL_MM 1
-cglobal h264_biweight_16x%1_mmx2, 7, 7, 0
+INIT_MMX mmxext
+cglobal h264_biweight_16, 7, 8, 0
BIWEIGHT_SETUP
- mov r3, %1
-%if %1 == 16
-.nextrow
+ movifnidn r3d, r3m
+.nextrow:
BIWEIGHT_STEPA 0, 1, 0
BIWEIGHT_STEPA 1, 2, 4
BIWEIGHT_STEPB
mova [r0+8], m0
add r0, r2
add r1, r2
- dec r3
+ dec r3d
jnz .nextrow
REP_RET
-%else
- jmp mangle(ff_h264_biweight_16x16_mmx2.nextrow)
-%endif
-%endmacro
-
-INIT_MMX
-BIWEIGHT_FUNC_DBL_MM 16
-BIWEIGHT_FUNC_DBL_MM 8
-%macro BIWEIGHT_FUNC_MM 4
-cglobal h264_biweight_%1x%2_%4, 7, 7, %3
+%macro BIWEIGHT_FUNC_MM 2
+cglobal h264_biweight_%1, 7, 8, %2
BIWEIGHT_SETUP
- mov r3, %2
-%if %2 == 16
-.nextrow
+ movifnidn r3d, r3m
+.nextrow:
BIWEIGHT_STEPA 0, 1, 0
BIWEIGHT_STEPA 1, 2, mmsize/2
BIWEIGHT_STEPB
mova [r0], m0
add r0, r2
add r1, r2
- dec r3
+ dec r3d
jnz .nextrow
REP_RET
-%else
- jmp mangle(ff_h264_biweight_%1x16_%4.nextrow)
-%endif
%endmacro
-INIT_MMX
-BIWEIGHT_FUNC_MM 8, 16, 0, mmx2
-BIWEIGHT_FUNC_MM 8, 8, 0, mmx2
-BIWEIGHT_FUNC_MM 8, 4, 0, mmx2
-INIT_XMM
-BIWEIGHT_FUNC_MM 16, 16, 8, sse2
-BIWEIGHT_FUNC_MM 16, 8, 8, sse2
+INIT_MMX mmxext
+BIWEIGHT_FUNC_MM 8, 0
+INIT_XMM sse2
+BIWEIGHT_FUNC_MM 16, 8
-%macro BIWEIGHT_FUNC_HALF_MM 5
-cglobal h264_biweight_%1x%2_%5, 7, 7, %4
+%macro BIWEIGHT_FUNC_HALF_MM 2
+cglobal h264_biweight_%1, 7, 8, %2
BIWEIGHT_SETUP
- mov r3, %2/2
+ movifnidn r3d, r3m
+ sar r3, 1
lea r4, [r2*2]
-%if %2 == mmsize
-.nextrow
+.nextrow:
BIWEIGHT_STEPA 0, 1, 0
BIWEIGHT_STEPA 1, 2, r2
BIWEIGHT_STEPB
%endif
add r0, r4
add r1, r4
- dec r3
+ dec r3d
jnz .nextrow
REP_RET
-%else
- jmp mangle(ff_h264_biweight_%1x%3_%5.nextrow)
-%endif
%endmacro
-INIT_MMX
-BIWEIGHT_FUNC_HALF_MM 4, 8, 8, 0, mmx2
-BIWEIGHT_FUNC_HALF_MM 4, 4, 8, 0, mmx2
-BIWEIGHT_FUNC_HALF_MM 4, 2, 8, 0, mmx2
-INIT_XMM
-BIWEIGHT_FUNC_HALF_MM 8, 16, 16, 8, sse2
-BIWEIGHT_FUNC_HALF_MM 8, 8, 16, 8, sse2
-BIWEIGHT_FUNC_HALF_MM 8, 4, 16, 8, sse2
-
-%macro BIWEIGHT_SSSE3_SETUP 0
- add r6, 1
- or r6, 1
- add r3, 1
- movd m4, r4
- movd m0, r5
- movd m5, r6
- movd m6, r3
- pslld m5, m6
- psrld m5, 1
- punpcklbw m4, m0
- pshuflw m4, m4, 0
- pshuflw m5, m5, 0
- punpcklqdq m4, m4
- punpcklqdq m5, m5
-%endmacro
+INIT_MMX mmxext
+BIWEIGHT_FUNC_HALF_MM 4, 0
+INIT_XMM sse2
+BIWEIGHT_FUNC_HALF_MM 8, 8
%macro BIWEIGHT_SSSE3_OP 0
pmaddubsw m0, m4
packuswb m0, m2
%endmacro
-%macro BIWEIGHT_SSSE3_16 1
-cglobal h264_biweight_16x%1_ssse3, 7, 7, 8
- BIWEIGHT_SSSE3_SETUP
- mov r3, %1
+INIT_XMM ssse3
+cglobal h264_biweight_16, 7, 8, 8
+ BIWEIGHT_SETUP
+ movifnidn r3d, r3m
-%if %1 == 16
-.nextrow
+.nextrow:
movh m0, [r0]
movh m2, [r0+8]
movh m3, [r1+8]
mova [r0], m0
add r0, r2
add r1, r2
- dec r3
+ dec r3d
jnz .nextrow
REP_RET
-%else
- jmp mangle(ff_h264_biweight_16x16_ssse3.nextrow)
-%endif
-%endmacro
-
-INIT_XMM
-BIWEIGHT_SSSE3_16 16
-BIWEIGHT_SSSE3_16 8
-%macro BIWEIGHT_SSSE3_8 1
-cglobal h264_biweight_8x%1_ssse3, 7, 7, 8
- BIWEIGHT_SSSE3_SETUP
- mov r3, %1/2
+INIT_XMM ssse3
+cglobal h264_biweight_8, 7, 8, 8
+ BIWEIGHT_SETUP
+ movifnidn r3d, r3m
+ sar r3, 1
lea r4, [r2*2]
-%if %1 == 16
-.nextrow
+.nextrow:
movh m0, [r0]
movh m1, [r1]
movh m2, [r0+r2]
movhps [r0+r2], m0
add r0, r4
add r1, r4
- dec r3
+ dec r3d
jnz .nextrow
REP_RET
-%else
- jmp mangle(ff_h264_biweight_8x16_ssse3.nextrow)
-%endif
-%endmacro
-
-INIT_XMM
-BIWEIGHT_SSSE3_8 16
-BIWEIGHT_SSSE3_8 8
-BIWEIGHT_SSSE3_8 4