diff options
author | Jason Garrett-Glaser <darkshikari@gmail.com> | 2010-06-29 00:40:12 +0000 |
---|---|---|
committer | Jason Garrett-Glaser <darkshikari@gmail.com> | 2010-06-29 00:40:12 +0000 |
commit | 37355fe823170537d97d9858877a928c645984f4 (patch) | |
tree | c6ebf00c9084c352fa3dd56fd6922c8acf8ad206 /libavcodec/x86/x86util.asm | |
parent | bc14f04b2f04e00573fa3d0e3778dadd6fa4f466 (diff) | |
download | ffmpeg-37355fe823170537d97d9858877a928c645984f4.tar.gz |
Make x86util.asm LGPL so we can use it in LGPL asm
Strip out most x264-specific stuff (not used anywhere in ffmpeg).
Originally committed as revision 23877 to svn://svn.ffmpeg.org/ffmpeg/trunk
Diffstat (limited to 'libavcodec/x86/x86util.asm')
-rw-r--r-- | libavcodec/x86/x86util.asm | 229 |
1 files changed, 16 insertions, 213 deletions
diff --git a/libavcodec/x86/x86util.asm b/libavcodec/x86/x86util.asm index a2c3b949c4..674ca1aab1 100644 --- a/libavcodec/x86/x86util.asm +++ b/libavcodec/x86/x86util.asm @@ -1,28 +1,27 @@ ;***************************************************************************** ;* x86util.asm ;***************************************************************************** -;* Copyright (C) 2008 x264 project +;* Copyright (C) 2008-2010 x264 project ;* -;* Authors: Holger Lubitz <holger@lubitz.org> -;* Loren Merritt <lorenm@u.washington.edu> +;* Authors: Loren Merritt <lorenm@u.washington.edu> +;* Holger Lubitz <holger@lubitz.org> ;* -;* This program is free software; you can redistribute it and/or modify -;* it under the terms of the GNU General Public License as published by -;* the Free Software Foundation; either version 2 of the License, or -;* (at your option) any later version. +;* This file is part of FFmpeg. ;* -;* This program is distributed in the hope that it will be useful, +;* 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 General Public License for more details. +;* 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 General Public License -;* along with this program; if not, write to the Free Software -;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA. -;***************************************************************************** - -%assign FENC_STRIDE 16 -%assign FDEC_STRIDE 32 +;* 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 +;****************************************************************************** %macro SBUTTERFLY 4 mova m%4, m%2 @@ -244,165 +243,6 @@ %endif %endmacro -%macro HADAMARD4_V 4+ - SUMSUB_BADC %1, %2, %3, %4 - SUMSUB_BADC %1, %3, %2, %4 -%endmacro - -%macro HADAMARD8_V 8+ - SUMSUB_BADC %1, %2, %3, %4 - SUMSUB_BADC %5, %6, %7, %8 - SUMSUB_BADC %1, %3, %2, %4 - SUMSUB_BADC %5, %7, %6, %8 - SUMSUB_BADC %1, %5, %2, %6 - SUMSUB_BADC %3, %7, %4, %8 -%endmacro - -%macro TRANS_SSE2 5-6 -; TRANSPOSE2x2 -; %1: transpose width (d/q) - use SBUTTERFLY qdq for dq -; %2: ord/unord (for compat with sse4, unused) -; %3/%4: source regs -; %5/%6: tmp regs -%ifidn %1, d -%define mask [mask_10] -%define shift 16 -%elifidn %1, q -%define mask [mask_1100] -%define shift 32 -%endif -%if %0==6 ; less dependency if we have two tmp - mova m%5, mask ; ff00 - mova m%6, m%4 ; x5x4 - psll%1 m%4, shift ; x4.. - pand m%6, m%5 ; x5.. - pandn m%5, m%3 ; ..x0 - psrl%1 m%3, shift ; ..x1 - por m%4, m%5 ; x4x0 - por m%3, m%6 ; x5x1 -%else ; more dependency, one insn less. sometimes faster, sometimes not - mova m%5, m%4 ; x5x4 - psll%1 m%4, shift ; x4.. - pxor m%4, m%3 ; (x4^x1)x0 - pand m%4, mask ; (x4^x1).. - pxor m%3, m%4 ; x4x0 - psrl%1 m%4, shift ; ..(x1^x4) - pxor m%5, m%4 ; x5x1 - SWAP %4, %3, %5 -%endif -%endmacro - -%macro TRANS_SSE4 5-6 ; see above -%ifidn %1, d - mova m%5, m%3 -%ifidn %2, ord - psrl%1 m%3, 16 -%endif - pblendw m%3, m%4, 10101010b - psll%1 m%4, 16 -%ifidn %2, ord - pblendw m%4, m%5, 01010101b -%else - psrl%1 m%5, 16 - por m%4, m%5 -%endif -%elifidn %1, q - mova m%5, m%3 - shufps m%3, m%4, 10001000b - shufps m%5, m%4, 11011101b - SWAP %4, %5 -%endif -%endmacro - -%macro HADAMARD 5-6 -; %1=distance in words (0 for vertical pass, 1/2/4 for horizontal passes) -; %2=sumsub/max/amax (sum and diff / maximum / maximum of absolutes) -; %3/%4: regs -; %5(%6): tmpregs -%if %1!=0 ; have to reorder stuff for horizontal op - %ifidn %2, sumsub - %define ORDER ord - ; sumsub needs order because a-b != b-a unless a=b - %else - %define ORDER unord - ; if we just max, order doesn't matter (allows pblendw+or in sse4) - %endif - %if %1==1 - TRANS d, ORDER, %3, %4, %5, %6 - %elif %1==2 - %if mmsize==8 - SBUTTERFLY dq, %3, %4, %5 - %else - TRANS q, ORDER, %3, %4, %5, %6 - %endif - %elif %1==4 - SBUTTERFLY qdq, %3, %4, %5 - %endif -%endif -%ifidn %2, sumsub - SUMSUB_BA m%3, m%4, m%5 -%else - %ifidn %2, amax - %if %0==6 - ABS2 m%3, m%4, m%5, m%6 - %else - ABS1 m%3, m%5 - ABS1 m%4, m%5 - %endif - %endif - pmaxsw m%3, m%4 -%endif -%endmacro - - -%macro HADAMARD2_2D 6-7 sumsub - HADAMARD 0, sumsub, %1, %2, %5 - HADAMARD 0, sumsub, %3, %4, %5 - SBUTTERFLY %6, %1, %2, %5 -%ifnum %7 - HADAMARD 0, amax, %1, %2, %5, %7 -%else - HADAMARD 0, %7, %1, %2, %5 -%endif - SBUTTERFLY %6, %3, %4, %5 -%ifnum %7 - HADAMARD 0, amax, %3, %4, %5, %7 -%else - HADAMARD 0, %7, %3, %4, %5 -%endif -%endmacro - -%macro HADAMARD4_2D 5-6 sumsub - HADAMARD2_2D %1, %2, %3, %4, %5, wd - HADAMARD2_2D %1, %3, %2, %4, %5, dq, %6 - SWAP %2, %3 -%endmacro - -%macro HADAMARD4_2D_SSE 5-6 sumsub - HADAMARD 0, sumsub, %1, %2, %5 ; 1st V row 0 + 1 - HADAMARD 0, sumsub, %3, %4, %5 ; 1st V row 2 + 3 - SBUTTERFLY wd, %1, %2, %5 ; %1: m0 1+0 %2: m1 1+0 - SBUTTERFLY wd, %3, %4, %5 ; %3: m0 3+2 %4: m1 3+2 - HADAMARD2_2D %1, %3, %2, %4, %5, dq - SBUTTERFLY qdq, %1, %2, %5 - HADAMARD 0, %6, %1, %2, %5 ; 2nd H m1/m0 row 0+1 - SBUTTERFLY qdq, %3, %4, %5 - HADAMARD 0, %6, %3, %4, %5 ; 2nd H m1/m0 row 2+3 -%endmacro - -%macro HADAMARD8_2D 9-10 sumsub - HADAMARD2_2D %1, %2, %3, %4, %9, wd - HADAMARD2_2D %5, %6, %7, %8, %9, wd - HADAMARD2_2D %1, %3, %2, %4, %9, dq - HADAMARD2_2D %5, %7, %6, %8, %9, dq - HADAMARD2_2D %1, %5, %3, %7, %9, qdq, %10 - HADAMARD2_2D %2, %6, %4, %8, %9, qdq, %10 -%ifnidn %10, amax - SWAP %2, %5 - SWAP %4, %7 -%endif -%endmacro - %macro SUMSUB2_AB 3 mova %3, %1 paddw %1, %1 @@ -483,32 +323,6 @@ %endif %endmacro -%macro LOAD_DIFF8x4_SSE2 8 - LOAD_DIFF m%1, m%5, m%6, [%7+%1*FENC_STRIDE], [%8+%1*FDEC_STRIDE] - LOAD_DIFF m%2, m%5, m%6, [%7+%2*FENC_STRIDE], [%8+%2*FDEC_STRIDE] - LOAD_DIFF m%3, m%5, m%6, [%7+%3*FENC_STRIDE], [%8+%3*FDEC_STRIDE] - LOAD_DIFF m%4, m%5, m%6, [%7+%4*FENC_STRIDE], [%8+%4*FDEC_STRIDE] -%endmacro - -%macro LOAD_DIFF8x4_SSSE3 8 ; 4x dst, 1x tmp, 1x mul, 2x ptr - movh m%2, [%8+%1*FDEC_STRIDE] - movh m%1, [%7+%1*FENC_STRIDE] - punpcklbw m%1, m%2 - movh m%3, [%8+%2*FDEC_STRIDE] - movh m%2, [%7+%2*FENC_STRIDE] - punpcklbw m%2, m%3 - movh m%4, [%8+%3*FDEC_STRIDE] - movh m%3, [%7+%3*FENC_STRIDE] - punpcklbw m%3, m%4 - movh m%5, [%8+%4*FDEC_STRIDE] - movh m%4, [%7+%4*FENC_STRIDE] - punpcklbw m%4, m%5 - pmaddubsw m%1, m%6 - pmaddubsw m%2, m%6 - pmaddubsw m%3, m%6 - pmaddubsw m%4, m%6 -%endmacro - %macro STORE_DCT 6 movq [%5+%6+ 0], m%1 movq [%5+%6+ 8], m%2 @@ -520,17 +334,6 @@ movhps [%5+%6+56], m%4 %endmacro -%macro STORE_IDCT 4 - movhps [r0-4*FDEC_STRIDE], %1 - movh [r0-3*FDEC_STRIDE], %1 - movhps [r0-2*FDEC_STRIDE], %2 - movh [r0-1*FDEC_STRIDE], %2 - movhps [r0+0*FDEC_STRIDE], %3 - movh [r0+1*FDEC_STRIDE], %3 - movhps [r0+2*FDEC_STRIDE], %4 - movh [r0+3*FDEC_STRIDE], %4 -%endmacro - %macro LOAD_DIFF_8x4P 7-10 r0,r2,0 ; 4x dest, 2x temp, 2x pointer, increment? LOAD_DIFF m%1, m%5, m%7, [%8], [%9] LOAD_DIFF m%2, m%6, m%7, [%8+r1], [%9+r3] |