mirror of https://github.com/FFmpeg/FFmpeg.git
The sh4 optimizations are removed, because the code is 100% identical to the C code, so it is unlikely to provide any real practical benefit. Signed-off-by: Diego Biurrun <diego@biurrun.de> Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com> Signed-off-by: Luca Barbato <lu_zero@gentoo.org>pull/9/head
Mans Rullgard
12 years ago
committed by
Luca Barbato
parent
7a95afe433
commit
e9d817351b
27 changed files with 2387 additions and 2459 deletions
@ -0,0 +1,170 @@ |
||||
/*
|
||||
* ARM NEON optimised DSP functions |
||||
* Copyright (c) 2008 Mans Rullgard <mans@mansr.com> |
||||
* |
||||
* This file is part of Libav. |
||||
* |
||||
* 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. |
||||
* |
||||
* 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 Libav; if not, write to the Free Software |
||||
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
||||
*/ |
||||
|
||||
#include <stdint.h> |
||||
|
||||
#include "config.h" |
||||
#include "libavcodec/h264qpel.h" |
||||
|
||||
void ff_put_h264_qpel16_mc00_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel16_mc10_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel16_mc20_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel16_mc30_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel16_mc01_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel16_mc11_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel16_mc21_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel16_mc31_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel16_mc02_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel16_mc12_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel16_mc22_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel16_mc32_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel16_mc03_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel16_mc13_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel16_mc23_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel16_mc33_neon(uint8_t *, uint8_t *, int); |
||||
|
||||
void ff_put_h264_qpel8_mc00_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel8_mc10_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel8_mc20_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel8_mc30_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel8_mc01_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel8_mc11_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel8_mc21_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel8_mc31_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel8_mc02_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel8_mc12_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel8_mc22_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel8_mc32_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel8_mc03_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel8_mc13_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel8_mc23_neon(uint8_t *, uint8_t *, int); |
||||
void ff_put_h264_qpel8_mc33_neon(uint8_t *, uint8_t *, int); |
||||
|
||||
void ff_avg_h264_qpel16_mc00_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel16_mc10_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel16_mc20_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel16_mc30_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel16_mc01_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel16_mc11_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel16_mc21_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel16_mc31_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel16_mc02_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel16_mc12_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel16_mc22_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel16_mc32_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel16_mc03_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel16_mc13_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel16_mc23_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel16_mc33_neon(uint8_t *, uint8_t *, int); |
||||
|
||||
void ff_avg_h264_qpel8_mc00_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel8_mc10_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel8_mc20_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel8_mc30_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel8_mc01_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel8_mc11_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel8_mc21_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel8_mc31_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel8_mc02_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel8_mc12_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel8_mc22_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel8_mc32_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel8_mc03_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel8_mc13_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel8_mc23_neon(uint8_t *, uint8_t *, int); |
||||
void ff_avg_h264_qpel8_mc33_neon(uint8_t *, uint8_t *, int); |
||||
|
||||
void ff_h264qpel_init_arm(H264QpelContext *c, int bit_depth) |
||||
{ |
||||
#if HAVE_NEON |
||||
const int high_bit_depth = bit_depth > 8; |
||||
|
||||
if (!high_bit_depth) { |
||||
c->put_h264_qpel_pixels_tab[0][ 0] = ff_put_h264_qpel16_mc00_neon; |
||||
c->put_h264_qpel_pixels_tab[0][ 1] = ff_put_h264_qpel16_mc10_neon; |
||||
c->put_h264_qpel_pixels_tab[0][ 2] = ff_put_h264_qpel16_mc20_neon; |
||||
c->put_h264_qpel_pixels_tab[0][ 3] = ff_put_h264_qpel16_mc30_neon; |
||||
c->put_h264_qpel_pixels_tab[0][ 4] = ff_put_h264_qpel16_mc01_neon; |
||||
c->put_h264_qpel_pixels_tab[0][ 5] = ff_put_h264_qpel16_mc11_neon; |
||||
c->put_h264_qpel_pixels_tab[0][ 6] = ff_put_h264_qpel16_mc21_neon; |
||||
c->put_h264_qpel_pixels_tab[0][ 7] = ff_put_h264_qpel16_mc31_neon; |
||||
c->put_h264_qpel_pixels_tab[0][ 8] = ff_put_h264_qpel16_mc02_neon; |
||||
c->put_h264_qpel_pixels_tab[0][ 9] = ff_put_h264_qpel16_mc12_neon; |
||||
c->put_h264_qpel_pixels_tab[0][10] = ff_put_h264_qpel16_mc22_neon; |
||||
c->put_h264_qpel_pixels_tab[0][11] = ff_put_h264_qpel16_mc32_neon; |
||||
c->put_h264_qpel_pixels_tab[0][12] = ff_put_h264_qpel16_mc03_neon; |
||||
c->put_h264_qpel_pixels_tab[0][13] = ff_put_h264_qpel16_mc13_neon; |
||||
c->put_h264_qpel_pixels_tab[0][14] = ff_put_h264_qpel16_mc23_neon; |
||||
c->put_h264_qpel_pixels_tab[0][15] = ff_put_h264_qpel16_mc33_neon; |
||||
|
||||
c->put_h264_qpel_pixels_tab[1][ 0] = ff_put_h264_qpel8_mc00_neon; |
||||
c->put_h264_qpel_pixels_tab[1][ 1] = ff_put_h264_qpel8_mc10_neon; |
||||
c->put_h264_qpel_pixels_tab[1][ 2] = ff_put_h264_qpel8_mc20_neon; |
||||
c->put_h264_qpel_pixels_tab[1][ 3] = ff_put_h264_qpel8_mc30_neon; |
||||
c->put_h264_qpel_pixels_tab[1][ 4] = ff_put_h264_qpel8_mc01_neon; |
||||
c->put_h264_qpel_pixels_tab[1][ 5] = ff_put_h264_qpel8_mc11_neon; |
||||
c->put_h264_qpel_pixels_tab[1][ 6] = ff_put_h264_qpel8_mc21_neon; |
||||
c->put_h264_qpel_pixels_tab[1][ 7] = ff_put_h264_qpel8_mc31_neon; |
||||
c->put_h264_qpel_pixels_tab[1][ 8] = ff_put_h264_qpel8_mc02_neon; |
||||
c->put_h264_qpel_pixels_tab[1][ 9] = ff_put_h264_qpel8_mc12_neon; |
||||
c->put_h264_qpel_pixels_tab[1][10] = ff_put_h264_qpel8_mc22_neon; |
||||
c->put_h264_qpel_pixels_tab[1][11] = ff_put_h264_qpel8_mc32_neon; |
||||
c->put_h264_qpel_pixels_tab[1][12] = ff_put_h264_qpel8_mc03_neon; |
||||
c->put_h264_qpel_pixels_tab[1][13] = ff_put_h264_qpel8_mc13_neon; |
||||
c->put_h264_qpel_pixels_tab[1][14] = ff_put_h264_qpel8_mc23_neon; |
||||
c->put_h264_qpel_pixels_tab[1][15] = ff_put_h264_qpel8_mc33_neon; |
||||
|
||||
c->avg_h264_qpel_pixels_tab[0][ 0] = ff_avg_h264_qpel16_mc00_neon; |
||||
c->avg_h264_qpel_pixels_tab[0][ 1] = ff_avg_h264_qpel16_mc10_neon; |
||||
c->avg_h264_qpel_pixels_tab[0][ 2] = ff_avg_h264_qpel16_mc20_neon; |
||||
c->avg_h264_qpel_pixels_tab[0][ 3] = ff_avg_h264_qpel16_mc30_neon; |
||||
c->avg_h264_qpel_pixels_tab[0][ 4] = ff_avg_h264_qpel16_mc01_neon; |
||||
c->avg_h264_qpel_pixels_tab[0][ 5] = ff_avg_h264_qpel16_mc11_neon; |
||||
c->avg_h264_qpel_pixels_tab[0][ 6] = ff_avg_h264_qpel16_mc21_neon; |
||||
c->avg_h264_qpel_pixels_tab[0][ 7] = ff_avg_h264_qpel16_mc31_neon; |
||||
c->avg_h264_qpel_pixels_tab[0][ 8] = ff_avg_h264_qpel16_mc02_neon; |
||||
c->avg_h264_qpel_pixels_tab[0][ 9] = ff_avg_h264_qpel16_mc12_neon; |
||||
c->avg_h264_qpel_pixels_tab[0][10] = ff_avg_h264_qpel16_mc22_neon; |
||||
c->avg_h264_qpel_pixels_tab[0][11] = ff_avg_h264_qpel16_mc32_neon; |
||||
c->avg_h264_qpel_pixels_tab[0][12] = ff_avg_h264_qpel16_mc03_neon; |
||||
c->avg_h264_qpel_pixels_tab[0][13] = ff_avg_h264_qpel16_mc13_neon; |
||||
c->avg_h264_qpel_pixels_tab[0][14] = ff_avg_h264_qpel16_mc23_neon; |
||||
c->avg_h264_qpel_pixels_tab[0][15] = ff_avg_h264_qpel16_mc33_neon; |
||||
|
||||
c->avg_h264_qpel_pixels_tab[1][ 0] = ff_avg_h264_qpel8_mc00_neon; |
||||
c->avg_h264_qpel_pixels_tab[1][ 1] = ff_avg_h264_qpel8_mc10_neon; |
||||
c->avg_h264_qpel_pixels_tab[1][ 2] = ff_avg_h264_qpel8_mc20_neon; |
||||
c->avg_h264_qpel_pixels_tab[1][ 3] = ff_avg_h264_qpel8_mc30_neon; |
||||
c->avg_h264_qpel_pixels_tab[1][ 4] = ff_avg_h264_qpel8_mc01_neon; |
||||
c->avg_h264_qpel_pixels_tab[1][ 5] = ff_avg_h264_qpel8_mc11_neon; |
||||
c->avg_h264_qpel_pixels_tab[1][ 6] = ff_avg_h264_qpel8_mc21_neon; |
||||
c->avg_h264_qpel_pixels_tab[1][ 7] = ff_avg_h264_qpel8_mc31_neon; |
||||
c->avg_h264_qpel_pixels_tab[1][ 8] = ff_avg_h264_qpel8_mc02_neon; |
||||
c->avg_h264_qpel_pixels_tab[1][ 9] = ff_avg_h264_qpel8_mc12_neon; |
||||
c->avg_h264_qpel_pixels_tab[1][10] = ff_avg_h264_qpel8_mc22_neon; |
||||
c->avg_h264_qpel_pixels_tab[1][11] = ff_avg_h264_qpel8_mc32_neon; |
||||
c->avg_h264_qpel_pixels_tab[1][12] = ff_avg_h264_qpel8_mc03_neon; |
||||
c->avg_h264_qpel_pixels_tab[1][13] = ff_avg_h264_qpel8_mc13_neon; |
||||
c->avg_h264_qpel_pixels_tab[1][14] = ff_avg_h264_qpel8_mc23_neon; |
||||
c->avg_h264_qpel_pixels_tab[1][15] = ff_avg_h264_qpel8_mc33_neon; |
||||
} |
||||
#endif /* HAVE_NEON */ |
||||
} |
||||
@ -0,0 +1,955 @@ |
||||
/* |
||||
* Copyright (c) 2008 Mans Rullgard <mans@mansr.com>
|
||||
* |
||||
* This file is part of Libav. |
||||
* |
||||
* 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. |
||||
* |
||||
* 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 Libav; if not, write to the Free Software
|
||||
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
||||
*/ |
||||
|
||||
#include "libavutil/arm/asm.S" |
||||
#include "neon.S" |
||||
|
||||
/* H.264 qpel MC */ |
||||
|
||||
.macro lowpass_const r |
||||
movw \r, #5 |
||||
movt \r, #20 |
||||
vmov.32 d6[0], \r |
||||
.endm |
||||
|
||||
.macro lowpass_8 r0, r1, r2, r3, d0, d1, narrow=1 |
||||
.if \narrow |
||||
t0 .req q0 |
||||
t1 .req q8 |
||||
.else |
||||
t0 .req \d0 |
||||
t1 .req \d1 |
||||
.endif |
||||
vext.8 d2, \r0, \r1, #2 |
||||
vext.8 d3, \r0, \r1, #3 |
||||
vaddl.u8 q1, d2, d3 |
||||
vext.8 d4, \r0, \r1, #1 |
||||
vext.8 d5, \r0, \r1, #4 |
||||
vaddl.u8 q2, d4, d5 |
||||
vext.8 d30, \r0, \r1, #5 |
||||
vaddl.u8 t0, \r0, d30 |
||||
vext.8 d18, \r2, \r3, #2 |
||||
vmla.i16 t0, q1, d6[1] |
||||
vext.8 d19, \r2, \r3, #3 |
||||
vaddl.u8 q9, d18, d19 |
||||
vext.8 d20, \r2, \r3, #1 |
||||
vmls.i16 t0, q2, d6[0] |
||||
vext.8 d21, \r2, \r3, #4 |
||||
vaddl.u8 q10, d20, d21 |
||||
vext.8 d31, \r2, \r3, #5 |
||||
vaddl.u8 t1, \r2, d31 |
||||
vmla.i16 t1, q9, d6[1] |
||||
vmls.i16 t1, q10, d6[0] |
||||
.if \narrow |
||||
vqrshrun.s16 \d0, t0, #5 |
||||
vqrshrun.s16 \d1, t1, #5 |
||||
.endif |
||||
.unreq t0
|
||||
.unreq t1
|
||||
.endm |
||||
|
||||
.macro lowpass_8_1 r0, r1, d0, narrow=1 |
||||
.if \narrow |
||||
t0 .req q0 |
||||
.else |
||||
t0 .req \d0 |
||||
.endif |
||||
vext.8 d2, \r0, \r1, #2 |
||||
vext.8 d3, \r0, \r1, #3 |
||||
vaddl.u8 q1, d2, d3 |
||||
vext.8 d4, \r0, \r1, #1 |
||||
vext.8 d5, \r0, \r1, #4 |
||||
vaddl.u8 q2, d4, d5 |
||||
vext.8 d30, \r0, \r1, #5 |
||||
vaddl.u8 t0, \r0, d30 |
||||
vmla.i16 t0, q1, d6[1] |
||||
vmls.i16 t0, q2, d6[0] |
||||
.if \narrow |
||||
vqrshrun.s16 \d0, t0, #5 |
||||
.endif |
||||
.unreq t0
|
||||
.endm |
||||
|
||||
.macro lowpass_8.16 r0, r1, l0, h0, l1, h1, d |
||||
vext.16 q1, \r0, \r1, #2 |
||||
vext.16 q0, \r0, \r1, #3 |
||||
vaddl.s16 q9, d2, d0 |
||||
vext.16 q2, \r0, \r1, #1 |
||||
vaddl.s16 q1, d3, d1 |
||||
vext.16 q3, \r0, \r1, #4 |
||||
vaddl.s16 q10, d4, d6 |
||||
vext.16 \r1, \r0, \r1, #5 |
||||
vaddl.s16 q2, d5, d7 |
||||
vaddl.s16 q0, \h0, \h1 |
||||
vaddl.s16 q8, \l0, \l1 |
||||
|
||||
vshl.i32 q3, q9, #4 |
||||
vshl.i32 q9, q9, #2 |
||||
vshl.i32 q15, q10, #2 |
||||
vadd.i32 q9, q9, q3 |
||||
vadd.i32 q10, q10, q15 |
||||
|
||||
vshl.i32 q3, q1, #4 |
||||
vshl.i32 q1, q1, #2 |
||||
vshl.i32 q15, q2, #2 |
||||
vadd.i32 q1, q1, q3 |
||||
vadd.i32 q2, q2, q15 |
||||
|
||||
vadd.i32 q9, q9, q8 |
||||
vsub.i32 q9, q9, q10 |
||||
|
||||
vadd.i32 q1, q1, q0 |
||||
vsub.i32 q1, q1, q2 |
||||
|
||||
vrshrn.s32 d18, q9, #10 |
||||
vrshrn.s32 d19, q1, #10 |
||||
|
||||
vqmovun.s16 \d, q9 |
||||
.endm |
||||
|
||||
function put_h264_qpel16_h_lowpass_neon_packed |
||||
mov r4, lr |
||||
mov r12, #16 |
||||
mov r3, #8 |
||||
bl put_h264_qpel8_h_lowpass_neon |
||||
sub r1, r1, r2, lsl #4 |
||||
add r1, r1, #8 |
||||
mov r12, #16 |
||||
mov lr, r4 |
||||
b put_h264_qpel8_h_lowpass_neon |
||||
endfunc |
||||
|
||||
.macro h264_qpel_h_lowpass type |
||||
function \type\()_h264_qpel16_h_lowpass_neon |
||||
push {lr} |
||||
mov r12, #16 |
||||
bl \type\()_h264_qpel8_h_lowpass_neon |
||||
sub r0, r0, r3, lsl #4 |
||||
sub r1, r1, r2, lsl #4 |
||||
add r0, r0, #8 |
||||
add r1, r1, #8 |
||||
mov r12, #16 |
||||
pop {lr} |
||||
endfunc |
||||
|
||||
function \type\()_h264_qpel8_h_lowpass_neon |
||||
1: vld1.8 {d0, d1}, [r1], r2 |
||||
vld1.8 {d16,d17}, [r1], r2 |
||||
subs r12, r12, #2 |
||||
lowpass_8 d0, d1, d16, d17, d0, d16 |
||||
.ifc \type,avg |
||||
vld1.8 {d2}, [r0,:64], r3 |
||||
vrhadd.u8 d0, d0, d2 |
||||
vld1.8 {d3}, [r0,:64] |
||||
vrhadd.u8 d16, d16, d3 |
||||
sub r0, r0, r3 |
||||
.endif |
||||
vst1.8 {d0}, [r0,:64], r3 |
||||
vst1.8 {d16}, [r0,:64], r3 |
||||
bne 1b |
||||
bx lr |
||||
endfunc |
||||
.endm |
||||
|
||||
h264_qpel_h_lowpass put |
||||
h264_qpel_h_lowpass avg |
||||
|
||||
.macro h264_qpel_h_lowpass_l2 type |
||||
function \type\()_h264_qpel16_h_lowpass_l2_neon |
||||
push {lr} |
||||
mov r12, #16 |
||||
bl \type\()_h264_qpel8_h_lowpass_l2_neon |
||||
sub r0, r0, r2, lsl #4 |
||||
sub r1, r1, r2, lsl #4 |
||||
sub r3, r3, r2, lsl #4 |
||||
add r0, r0, #8 |
||||
add r1, r1, #8 |
||||
add r3, r3, #8 |
||||
mov r12, #16 |
||||
pop {lr} |
||||
endfunc |
||||
|
||||
function \type\()_h264_qpel8_h_lowpass_l2_neon |
||||
1: vld1.8 {d0, d1}, [r1], r2 |
||||
vld1.8 {d16,d17}, [r1], r2 |
||||
vld1.8 {d28}, [r3], r2 |
||||
vld1.8 {d29}, [r3], r2 |
||||
subs r12, r12, #2 |
||||
lowpass_8 d0, d1, d16, d17, d0, d1 |
||||
vrhadd.u8 q0, q0, q14 |
||||
.ifc \type,avg |
||||
vld1.8 {d2}, [r0,:64], r2 |
||||
vrhadd.u8 d0, d0, d2 |
||||
vld1.8 {d3}, [r0,:64] |
||||
vrhadd.u8 d1, d1, d3 |
||||
sub r0, r0, r2 |
||||
.endif |
||||
vst1.8 {d0}, [r0,:64], r2 |
||||
vst1.8 {d1}, [r0,:64], r2 |
||||
bne 1b |
||||
bx lr |
||||
endfunc |
||||
.endm |
||||
|
||||
h264_qpel_h_lowpass_l2 put |
||||
h264_qpel_h_lowpass_l2 avg |
||||
|
||||
function put_h264_qpel16_v_lowpass_neon_packed |
||||
mov r4, lr |
||||
mov r2, #8 |
||||
bl put_h264_qpel8_v_lowpass_neon |
||||
sub r1, r1, r3, lsl #2 |
||||
bl put_h264_qpel8_v_lowpass_neon |
||||
sub r1, r1, r3, lsl #4 |
||||
sub r1, r1, r3, lsl #2 |
||||
add r1, r1, #8 |
||||
bl put_h264_qpel8_v_lowpass_neon |
||||
sub r1, r1, r3, lsl #2 |
||||
mov lr, r4 |
||||
b put_h264_qpel8_v_lowpass_neon |
||||
endfunc |
||||
|
||||
.macro h264_qpel_v_lowpass type |
||||
function \type\()_h264_qpel16_v_lowpass_neon |
||||
mov r4, lr |
||||
bl \type\()_h264_qpel8_v_lowpass_neon |
||||
sub r1, r1, r3, lsl #2 |
||||
bl \type\()_h264_qpel8_v_lowpass_neon |
||||
sub r0, r0, r2, lsl #4 |
||||
add r0, r0, #8 |
||||
sub r1, r1, r3, lsl #4 |
||||
sub r1, r1, r3, lsl #2 |
||||
add r1, r1, #8 |
||||
bl \type\()_h264_qpel8_v_lowpass_neon |
||||
sub r1, r1, r3, lsl #2 |
||||
mov lr, r4 |
||||
endfunc |
||||
|
||||
function \type\()_h264_qpel8_v_lowpass_neon |
||||
vld1.8 {d8}, [r1], r3 |
||||
vld1.8 {d10}, [r1], r3 |
||||
vld1.8 {d12}, [r1], r3 |
||||
vld1.8 {d14}, [r1], r3 |
||||
vld1.8 {d22}, [r1], r3 |
||||
vld1.8 {d24}, [r1], r3 |
||||
vld1.8 {d26}, [r1], r3 |
||||
vld1.8 {d28}, [r1], r3 |
||||
vld1.8 {d9}, [r1], r3 |
||||
vld1.8 {d11}, [r1], r3 |
||||
vld1.8 {d13}, [r1], r3 |
||||
vld1.8 {d15}, [r1], r3 |
||||
vld1.8 {d23}, [r1] |
||||
|
||||
transpose_8x8 q4, q5, q6, q7, q11, q12, q13, q14 |
||||
lowpass_8 d8, d9, d10, d11, d8, d10 |
||||
lowpass_8 d12, d13, d14, d15, d12, d14 |
||||
lowpass_8 d22, d23, d24, d25, d22, d24 |
||||
lowpass_8 d26, d27, d28, d29, d26, d28 |
||||
transpose_8x8 d8, d10, d12, d14, d22, d24, d26, d28 |
||||
|
||||
.ifc \type,avg |
||||
vld1.8 {d9}, [r0,:64], r2 |
||||
vrhadd.u8 d8, d8, d9 |
||||
vld1.8 {d11}, [r0,:64], r2 |
||||
vrhadd.u8 d10, d10, d11 |
||||
vld1.8 {d13}, [r0,:64], r2 |
||||
vrhadd.u8 d12, d12, d13 |
||||
vld1.8 {d15}, [r0,:64], r2 |
||||
vrhadd.u8 d14, d14, d15 |
||||
vld1.8 {d23}, [r0,:64], r2 |
||||
vrhadd.u8 d22, d22, d23 |
||||
vld1.8 {d25}, [r0,:64], r2 |
||||
vrhadd.u8 d24, d24, d25 |
||||
vld1.8 {d27}, [r0,:64], r2 |
||||
vrhadd.u8 d26, d26, d27 |
||||
vld1.8 {d29}, [r0,:64], r2 |
||||
vrhadd.u8 d28, d28, d29 |
||||
sub r0, r0, r2, lsl #3 |
||||
.endif |
||||
|
||||
vst1.8 {d8}, [r0,:64], r2 |
||||
vst1.8 {d10}, [r0,:64], r2 |
||||
vst1.8 {d12}, [r0,:64], r2 |
||||
vst1.8 {d14}, [r0,:64], r2 |
||||
vst1.8 {d22}, [r0,:64], r2 |
||||
vst1.8 {d24}, [r0,:64], r2 |
||||
vst1.8 {d26}, [r0,:64], r2 |
||||
vst1.8 {d28}, [r0,:64], r2 |
||||
|
||||
bx lr |
||||
endfunc |
||||
.endm |
||||
|
||||
h264_qpel_v_lowpass put |
||||
h264_qpel_v_lowpass avg |
||||
|
||||
.macro h264_qpel_v_lowpass_l2 type |
||||
function \type\()_h264_qpel16_v_lowpass_l2_neon |
||||
mov r4, lr |
||||
bl \type\()_h264_qpel8_v_lowpass_l2_neon |
||||
sub r1, r1, r3, lsl #2 |
||||
bl \type\()_h264_qpel8_v_lowpass_l2_neon |
||||
sub r0, r0, r3, lsl #4 |
||||
sub r12, r12, r2, lsl #4 |
||||
add r0, r0, #8 |
||||
add r12, r12, #8 |
||||
sub r1, r1, r3, lsl #4 |
||||
sub r1, r1, r3, lsl #2 |
||||
add r1, r1, #8 |
||||
bl \type\()_h264_qpel8_v_lowpass_l2_neon |
||||
sub r1, r1, r3, lsl #2 |
||||
mov lr, r4 |
||||
endfunc |
||||
|
||||
function \type\()_h264_qpel8_v_lowpass_l2_neon |
||||
vld1.8 {d8}, [r1], r3 |
||||
vld1.8 {d10}, [r1], r3 |
||||
vld1.8 {d12}, [r1], r3 |
||||
vld1.8 {d14}, [r1], r3 |
||||
vld1.8 {d22}, [r1], r3 |
||||
vld1.8 {d24}, [r1], r3 |
||||
vld1.8 {d26}, [r1], r3 |
||||
vld1.8 {d28}, [r1], r3 |
||||
vld1.8 {d9}, [r1], r3 |
||||
vld1.8 {d11}, [r1], r3 |
||||
vld1.8 {d13}, [r1], r3 |
||||
vld1.8 {d15}, [r1], r3 |
||||
vld1.8 {d23}, [r1] |
||||
|
||||
transpose_8x8 q4, q5, q6, q7, q11, q12, q13, q14 |
||||
lowpass_8 d8, d9, d10, d11, d8, d9 |
||||
lowpass_8 d12, d13, d14, d15, d12, d13 |
||||
lowpass_8 d22, d23, d24, d25, d22, d23 |
||||
lowpass_8 d26, d27, d28, d29, d26, d27 |
||||
transpose_8x8 d8, d9, d12, d13, d22, d23, d26, d27 |
||||
|
||||
vld1.8 {d0}, [r12], r2 |
||||
vld1.8 {d1}, [r12], r2 |
||||
vld1.8 {d2}, [r12], r2 |
||||
vld1.8 {d3}, [r12], r2 |
||||
vld1.8 {d4}, [r12], r2 |
||||
vrhadd.u8 q0, q0, q4 |
||||
vld1.8 {d5}, [r12], r2 |
||||
vrhadd.u8 q1, q1, q6 |
||||
vld1.8 {d10}, [r12], r2 |
||||
vrhadd.u8 q2, q2, q11 |
||||
vld1.8 {d11}, [r12], r2 |
||||
vrhadd.u8 q5, q5, q13 |
||||
|
||||
.ifc \type,avg |
||||
vld1.8 {d16}, [r0,:64], r3 |
||||
vrhadd.u8 d0, d0, d16 |
||||
vld1.8 {d17}, [r0,:64], r3 |
||||
vrhadd.u8 d1, d1, d17 |
||||
vld1.8 {d16}, [r0,:64], r3 |
||||
vrhadd.u8 d2, d2, d16 |
||||
vld1.8 {d17}, [r0,:64], r3 |
||||
vrhadd.u8 d3, d3, d17 |
||||
vld1.8 {d16}, [r0,:64], r3 |
||||
vrhadd.u8 d4, d4, d16 |
||||
vld1.8 {d17}, [r0,:64], r3 |
||||
vrhadd.u8 d5, d5, d17 |
||||
vld1.8 {d16}, [r0,:64], r3 |
||||
vrhadd.u8 d10, d10, d16 |
||||
vld1.8 {d17}, [r0,:64], r3 |
||||
vrhadd.u8 d11, d11, d17 |
||||
sub r0, r0, r3, lsl #3 |
||||
.endif |
||||
|
||||
vst1.8 {d0}, [r0,:64], r3 |
||||
vst1.8 {d1}, [r0,:64], r3 |
||||
vst1.8 {d2}, [r0,:64], r3 |
||||
vst1.8 {d3}, [r0,:64], r3 |
||||
vst1.8 {d4}, [r0,:64], r3 |
||||
vst1.8 {d5}, [r0,:64], r3 |
||||
vst1.8 {d10}, [r0,:64], r3 |
||||
vst1.8 {d11}, [r0,:64], r3 |
||||
|
||||
bx lr |
||||
endfunc |
||||
.endm |
||||
|
||||
h264_qpel_v_lowpass_l2 put |
||||
h264_qpel_v_lowpass_l2 avg |
||||
|
||||
function put_h264_qpel8_hv_lowpass_neon_top |
||||
lowpass_const r12 |
||||
mov r12, #12 |
||||
1: vld1.8 {d0, d1}, [r1], r3 |
||||
vld1.8 {d16,d17}, [r1], r3 |
||||
subs r12, r12, #2 |
||||
lowpass_8 d0, d1, d16, d17, q11, q12, narrow=0 |
||||
vst1.8 {d22-d25}, [r4,:128]! |
||||
bne 1b |
||||
|
||||
vld1.8 {d0, d1}, [r1] |
||||
lowpass_8_1 d0, d1, q12, narrow=0 |
||||
|
||||
mov r12, #-16 |
||||
add r4, r4, r12 |
||||
vld1.8 {d30,d31}, [r4,:128], r12 |
||||
vld1.8 {d20,d21}, [r4,:128], r12 |
||||
vld1.8 {d18,d19}, [r4,:128], r12 |
||||
vld1.8 {d16,d17}, [r4,:128], r12 |
||||
vld1.8 {d14,d15}, [r4,:128], r12 |
||||
vld1.8 {d12,d13}, [r4,:128], r12 |
||||
vld1.8 {d10,d11}, [r4,:128], r12 |
||||
vld1.8 {d8, d9}, [r4,:128], r12 |
||||
vld1.8 {d6, d7}, [r4,:128], r12 |
||||
vld1.8 {d4, d5}, [r4,:128], r12 |
||||
vld1.8 {d2, d3}, [r4,:128], r12 |
||||
vld1.8 {d0, d1}, [r4,:128] |
||||
|
||||
swap4 d1, d3, d5, d7, d8, d10, d12, d14 |
||||
transpose16_4x4 q0, q1, q2, q3, q4, q5, q6, q7 |
||||
|
||||
swap4 d17, d19, d21, d31, d24, d26, d28, d22 |
||||
transpose16_4x4 q8, q9, q10, q15, q12, q13, q14, q11 |
||||
|
||||
vst1.8 {d30,d31}, [r4,:128]! |
||||
vst1.8 {d6, d7}, [r4,:128]! |
||||
vst1.8 {d20,d21}, [r4,:128]! |
||||
vst1.8 {d4, d5}, [r4,:128]! |
||||
vst1.8 {d18,d19}, [r4,:128]! |
||||
vst1.8 {d2, d3}, [r4,:128]! |
||||
vst1.8 {d16,d17}, [r4,:128]! |
||||
vst1.8 {d0, d1}, [r4,:128] |
||||
|
||||
lowpass_8.16 q4, q12, d8, d9, d24, d25, d8 |
||||
lowpass_8.16 q5, q13, d10, d11, d26, d27, d9 |
||||
lowpass_8.16 q6, q14, d12, d13, d28, d29, d10 |
||||
lowpass_8.16 q7, q11, d14, d15, d22, d23, d11 |
||||
|
||||
vld1.8 {d16,d17}, [r4,:128], r12 |
||||
vld1.8 {d30,d31}, [r4,:128], r12 |
||||
lowpass_8.16 q8, q15, d16, d17, d30, d31, d12 |
||||
vld1.8 {d16,d17}, [r4,:128], r12 |
||||
vld1.8 {d30,d31}, [r4,:128], r12 |
||||
lowpass_8.16 q8, q15, d16, d17, d30, d31, d13 |
||||
vld1.8 {d16,d17}, [r4,:128], r12 |
||||
vld1.8 {d30,d31}, [r4,:128], r12 |
||||
lowpass_8.16 q8, q15, d16, d17, d30, d31, d14 |
||||
vld1.8 {d16,d17}, [r4,:128], r12 |
||||
vld1.8 {d30,d31}, [r4,:128] |
||||
lowpass_8.16 q8, q15, d16, d17, d30, d31, d15 |
||||
|
||||
transpose_8x8 d12, d13, d14, d15, d8, d9, d10, d11 |
||||
|
||||
bx lr |
||||
endfunc |
||||
|
||||
.macro h264_qpel8_hv_lowpass type |
||||
function \type\()_h264_qpel8_hv_lowpass_neon |
||||
mov r10, lr |
||||
bl put_h264_qpel8_hv_lowpass_neon_top |
||||
.ifc \type,avg |
||||
vld1.8 {d0}, [r0,:64], r2 |
||||
vrhadd.u8 d12, d12, d0 |
||||
vld1.8 {d1}, [r0,:64], r2 |
||||
vrhadd.u8 d13, d13, d1 |
||||
vld1.8 {d2}, [r0,:64], r2 |
||||
vrhadd.u8 d14, d14, d2 |
||||
vld1.8 {d3}, [r0,:64], r2 |
||||
vrhadd.u8 d15, d15, d3 |
||||
vld1.8 {d4}, [r0,:64], r2 |
||||
vrhadd.u8 d8, d8, d4 |
||||
vld1.8 {d5}, [r0,:64], r2 |
||||
vrhadd.u8 d9, d9, d5 |
||||
vld1.8 {d6}, [r0,:64], r2 |
||||
vrhadd.u8 d10, d10, d6 |
||||
vld1.8 {d7}, [r0,:64], r2 |
||||
vrhadd.u8 d11, d11, d7 |
||||
sub r0, r0, r2, lsl #3 |
||||
.endif |
||||
|
||||
vst1.8 {d12}, [r0,:64], r2 |
||||
vst1.8 {d13}, [r0,:64], r2 |
||||
vst1.8 {d14}, [r0,:64], r2 |
||||
vst1.8 {d15}, [r0,:64], r2 |
||||
vst1.8 {d8}, [r0,:64], r2 |
||||
vst1.8 {d9}, [r0,:64], r2 |
||||
vst1.8 {d10}, [r0,:64], r2 |
||||
vst1.8 {d11}, [r0,:64], r2 |
||||
|
||||
mov lr, r10 |
||||
bx lr |
||||
endfunc |
||||
.endm |
||||
|
||||
h264_qpel8_hv_lowpass put |
||||
h264_qpel8_hv_lowpass avg |
||||
|
||||
.macro h264_qpel8_hv_lowpass_l2 type |
||||
function \type\()_h264_qpel8_hv_lowpass_l2_neon |
||||
mov r10, lr |
||||
bl put_h264_qpel8_hv_lowpass_neon_top |
||||
|
||||
vld1.8 {d0, d1}, [r2,:128]! |
||||
vld1.8 {d2, d3}, [r2,:128]! |
||||
vrhadd.u8 q0, q0, q6 |
||||
vld1.8 {d4, d5}, [r2,:128]! |
||||
vrhadd.u8 q1, q1, q7 |
||||
vld1.8 {d6, d7}, [r2,:128]! |
||||
vrhadd.u8 q2, q2, q4 |
||||
vrhadd.u8 q3, q3, q5 |
||||
.ifc \type,avg |
||||
vld1.8 {d16}, [r0,:64], r3 |
||||
vrhadd.u8 d0, d0, d16 |
||||
vld1.8 {d17}, [r0,:64], r3 |
||||
vrhadd.u8 d1, d1, d17 |
||||
vld1.8 {d18}, [r0,:64], r3 |
||||
vrhadd.u8 d2, d2, d18 |
||||
vld1.8 {d19}, [r0,:64], r3 |
||||
vrhadd.u8 d3, d3, d19 |
||||
vld1.8 {d20}, [r0,:64], r3 |
||||
vrhadd.u8 d4, d4, d20 |
||||
vld1.8 {d21}, [r0,:64], r3 |
||||
vrhadd.u8 d5, d5, d21 |
||||
vld1.8 {d22}, [r0,:64], r3 |
||||
vrhadd.u8 d6, d6, d22 |
||||
vld1.8 {d23}, [r0,:64], r3 |
||||
vrhadd.u8 d7, d7, d23 |
||||
sub r0, r0, r3, lsl #3 |
||||
.endif |
||||
vst1.8 {d0}, [r0,:64], r3 |
||||
vst1.8 {d1}, [r0,:64], r3 |
||||
vst1.8 {d2}, [r0,:64], r3 |
||||
vst1.8 {d3}, [r0,:64], r3 |
||||
vst1.8 {d4}, [r0,:64], r3 |
||||
vst1.8 {d5}, [r0,:64], r3 |
||||
vst1.8 {d6}, [r0,:64], r3 |
||||
vst1.8 {d7}, [r0,:64], r3 |
||||
|
||||
mov lr, r10 |
||||
bx lr |
||||
endfunc |
||||
.endm |
||||
|
||||
h264_qpel8_hv_lowpass_l2 put |
||||
h264_qpel8_hv_lowpass_l2 avg |
||||
|
||||
.macro h264_qpel16_hv type |
||||
function \type\()_h264_qpel16_hv_lowpass_neon |
||||
mov r9, lr |
||||
bl \type\()_h264_qpel8_hv_lowpass_neon |
||||
sub r1, r1, r3, lsl #2 |
||||
bl \type\()_h264_qpel8_hv_lowpass_neon |
||||
sub r1, r1, r3, lsl #4 |
||||
sub r1, r1, r3, lsl #2 |
||||
add r1, r1, #8 |
||||
sub r0, r0, r2, lsl #4 |
||||
add r0, r0, #8 |
||||
bl \type\()_h264_qpel8_hv_lowpass_neon |
||||
sub r1, r1, r3, lsl #2 |
||||
mov lr, r9 |
||||
b \type\()_h264_qpel8_hv_lowpass_neon |
||||
endfunc |
||||
|
||||
function \type\()_h264_qpel16_hv_lowpass_l2_neon |
||||
mov r9, lr |
||||
sub r2, r4, #256 |
||||
bl \type\()_h264_qpel8_hv_lowpass_l2_neon |
||||
sub r1, r1, r3, lsl #2 |
||||
bl \type\()_h264_qpel8_hv_lowpass_l2_neon |
||||
sub r1, r1, r3, lsl #4 |
||||
sub r1, r1, r3, lsl #2 |
||||
add r1, r1, #8 |
||||
sub r0, r0, r3, lsl #4 |
||||
add r0, r0, #8 |
||||
bl \type\()_h264_qpel8_hv_lowpass_l2_neon |
||||
sub r1, r1, r3, lsl #2 |
||||
mov lr, r9 |
||||
b \type\()_h264_qpel8_hv_lowpass_l2_neon |
||||
endfunc |
||||
.endm |
||||
|
||||
h264_qpel16_hv put |
||||
h264_qpel16_hv avg |
||||
|
||||
.macro h264_qpel8 type |
||||
function ff_\type\()_h264_qpel8_mc10_neon, export=1 |
||||
lowpass_const r3 |
||||
mov r3, r1 |
||||
sub r1, r1, #2 |
||||
mov r12, #8 |
||||
b \type\()_h264_qpel8_h_lowpass_l2_neon |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel8_mc20_neon, export=1 |
||||
lowpass_const r3 |
||||
sub r1, r1, #2 |
||||
mov r3, r2 |
||||
mov r12, #8 |
||||
b \type\()_h264_qpel8_h_lowpass_neon |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel8_mc30_neon, export=1 |
||||
lowpass_const r3 |
||||
add r3, r1, #1 |
||||
sub r1, r1, #2 |
||||
mov r12, #8 |
||||
b \type\()_h264_qpel8_h_lowpass_l2_neon |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel8_mc01_neon, export=1 |
||||
push {lr} |
||||
mov r12, r1 |
||||
\type\()_h264_qpel8_mc01: |
||||
lowpass_const r3 |
||||
mov r3, r2 |
||||
sub r1, r1, r2, lsl #1 |
||||
vpush {d8-d15} |
||||
bl \type\()_h264_qpel8_v_lowpass_l2_neon |
||||
vpop {d8-d15} |
||||
pop {pc} |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel8_mc11_neon, export=1 |
||||
push {r0, r1, r11, lr} |
||||
\type\()_h264_qpel8_mc11: |
||||
lowpass_const r3 |
||||
mov r11, sp |
||||
A bic sp, sp, #15 |
||||
T bic r0, r11, #15 |
||||
T mov sp, r0 |
||||
sub sp, sp, #64 |
||||
mov r0, sp |
||||
sub r1, r1, #2 |
||||
mov r3, #8 |
||||
mov r12, #8 |
||||
vpush {d8-d15} |
||||
bl put_h264_qpel8_h_lowpass_neon |
||||
ldrd r0, r1, [r11], #8 |
||||
mov r3, r2 |
||||
add r12, sp, #64 |
||||
sub r1, r1, r2, lsl #1 |
||||
mov r2, #8 |
||||
bl \type\()_h264_qpel8_v_lowpass_l2_neon |
||||
vpop {d8-d15} |
||||
mov sp, r11 |
||||
pop {r11, pc} |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel8_mc21_neon, export=1 |
||||
push {r0, r1, r4, r10, r11, lr} |
||||
\type\()_h264_qpel8_mc21: |
||||
lowpass_const r3 |
||||
mov r11, sp |
||||
A bic sp, sp, #15 |
||||
T bic r0, r11, #15 |
||||
T mov sp, r0 |
||||
sub sp, sp, #(8*8+16*12) |
||||
sub r1, r1, #2 |
||||
mov r3, #8 |
||||
mov r0, sp |
||||
mov r12, #8 |
||||
vpush {d8-d15} |
||||
bl put_h264_qpel8_h_lowpass_neon |
||||
mov r4, r0 |
||||
ldrd r0, r1, [r11], #8 |
||||
sub r1, r1, r2, lsl #1 |
||||
sub r1, r1, #2 |
||||
mov r3, r2 |
||||
sub r2, r4, #64 |
||||
bl \type\()_h264_qpel8_hv_lowpass_l2_neon |
||||
vpop {d8-d15} |
||||
mov sp, r11 |
||||
pop {r4, r10, r11, pc} |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel8_mc31_neon, export=1 |
||||
add r1, r1, #1 |
||||
push {r0, r1, r11, lr} |
||||
sub r1, r1, #1 |
||||
b \type\()_h264_qpel8_mc11 |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel8_mc02_neon, export=1 |
||||
push {lr} |
||||
lowpass_const r3 |
||||
sub r1, r1, r2, lsl #1 |
||||
mov r3, r2 |
||||
vpush {d8-d15} |
||||
bl \type\()_h264_qpel8_v_lowpass_neon |
||||
vpop {d8-d15} |
||||
pop {pc} |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel8_mc12_neon, export=1 |
||||
push {r0, r1, r4, r10, r11, lr} |
||||
\type\()_h264_qpel8_mc12: |
||||
lowpass_const r3 |
||||
mov r11, sp |
||||
A bic sp, sp, #15 |
||||
T bic r0, r11, #15 |
||||
T mov sp, r0 |
||||
sub sp, sp, #(8*8+16*12) |
||||
sub r1, r1, r2, lsl #1 |
||||
mov r3, r2 |
||||
mov r2, #8 |
||||
mov r0, sp |
||||
vpush {d8-d15} |
||||
bl put_h264_qpel8_v_lowpass_neon |
||||
mov r4, r0 |
||||
ldrd r0, r1, [r11], #8 |
||||
sub r1, r1, r3, lsl #1 |
||||
sub r1, r1, #2 |
||||
sub r2, r4, #64 |
||||
bl \type\()_h264_qpel8_hv_lowpass_l2_neon |
||||
vpop {d8-d15} |
||||
mov sp, r11 |
||||
pop {r4, r10, r11, pc} |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel8_mc22_neon, export=1 |
||||
push {r4, r10, r11, lr} |
||||
mov r11, sp |
||||
A bic sp, sp, #15 |
||||
T bic r4, r11, #15 |
||||
T mov sp, r4 |
||||
sub r1, r1, r2, lsl #1 |
||||
sub r1, r1, #2 |
||||
mov r3, r2 |
||||
sub sp, sp, #(16*12) |
||||
mov r4, sp |
||||
vpush {d8-d15} |
||||
bl \type\()_h264_qpel8_hv_lowpass_neon |
||||
vpop {d8-d15} |
||||
mov sp, r11 |
||||
pop {r4, r10, r11, pc} |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel8_mc32_neon, export=1 |
||||
push {r0, r1, r4, r10, r11, lr} |
||||
add r1, r1, #1 |
||||
b \type\()_h264_qpel8_mc12 |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel8_mc03_neon, export=1 |
||||
push {lr} |
||||
add r12, r1, r2 |
||||
b \type\()_h264_qpel8_mc01 |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel8_mc13_neon, export=1 |
||||
push {r0, r1, r11, lr} |
||||
add r1, r1, r2 |
||||
b \type\()_h264_qpel8_mc11 |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel8_mc23_neon, export=1 |
||||
push {r0, r1, r4, r10, r11, lr} |
||||
add r1, r1, r2 |
||||
b \type\()_h264_qpel8_mc21 |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel8_mc33_neon, export=1 |
||||
add r1, r1, #1 |
||||
push {r0, r1, r11, lr} |
||||
add r1, r1, r2 |
||||
sub r1, r1, #1 |
||||
b \type\()_h264_qpel8_mc11 |
||||
endfunc |
||||
.endm |
||||
|
||||
h264_qpel8 put |
||||
h264_qpel8 avg |
||||
|
||||
.macro h264_qpel16 type |
||||
function ff_\type\()_h264_qpel16_mc10_neon, export=1 |
||||
lowpass_const r3 |
||||
mov r3, r1 |
||||
sub r1, r1, #2 |
||||
b \type\()_h264_qpel16_h_lowpass_l2_neon |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel16_mc20_neon, export=1 |
||||
lowpass_const r3 |
||||
sub r1, r1, #2 |
||||
mov r3, r2 |
||||
b \type\()_h264_qpel16_h_lowpass_neon |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel16_mc30_neon, export=1 |
||||
lowpass_const r3 |
||||
add r3, r1, #1 |
||||
sub r1, r1, #2 |
||||
b \type\()_h264_qpel16_h_lowpass_l2_neon |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel16_mc01_neon, export=1 |
||||
push {r4, lr} |
||||
mov r12, r1 |
||||
\type\()_h264_qpel16_mc01: |
||||
lowpass_const r3 |
||||
mov r3, r2 |
||||
sub r1, r1, r2, lsl #1 |
||||
vpush {d8-d15} |
||||
bl \type\()_h264_qpel16_v_lowpass_l2_neon |
||||
vpop {d8-d15} |
||||
pop {r4, pc} |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel16_mc11_neon, export=1 |
||||
push {r0, r1, r4, r11, lr} |
||||
\type\()_h264_qpel16_mc11: |
||||
lowpass_const r3 |
||||
mov r11, sp |
||||
A bic sp, sp, #15 |
||||
T bic r0, r11, #15 |
||||
T mov sp, r0 |
||||
sub sp, sp, #256 |
||||
mov r0, sp |
||||
sub r1, r1, #2 |
||||
mov r3, #16 |
||||
vpush {d8-d15} |
||||
bl put_h264_qpel16_h_lowpass_neon |
||||
ldrd r0, r1, [r11], #8 |
||||
mov r3, r2 |
||||
add r12, sp, #64 |
||||
sub r1, r1, r2, lsl #1 |
||||
mov r2, #16 |
||||
bl \type\()_h264_qpel16_v_lowpass_l2_neon |
||||
vpop {d8-d15} |
||||
mov sp, r11 |
||||
pop {r4, r11, pc} |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel16_mc21_neon, export=1 |
||||
push {r0, r1, r4-r5, r9-r11, lr} |
||||
\type\()_h264_qpel16_mc21: |
||||
lowpass_const r3 |
||||
mov r11, sp |
||||
A bic sp, sp, #15 |
||||
T bic r0, r11, #15 |
||||
T mov sp, r0 |
||||
sub sp, sp, #(16*16+16*12) |
||||
sub r1, r1, #2 |
||||
mov r0, sp |
||||
vpush {d8-d15} |
||||
bl put_h264_qpel16_h_lowpass_neon_packed |
||||
mov r4, r0 |
||||
ldrd r0, r1, [r11], #8 |
||||
sub r1, r1, r2, lsl #1 |
||||
sub r1, r1, #2 |
||||
mov r3, r2 |
||||
bl \type\()_h264_qpel16_hv_lowpass_l2_neon |
||||
vpop {d8-d15} |
||||
mov sp, r11 |
||||
pop {r4-r5, r9-r11, pc} |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel16_mc31_neon, export=1 |
||||
add r1, r1, #1 |
||||
push {r0, r1, r4, r11, lr} |
||||
sub r1, r1, #1 |
||||
b \type\()_h264_qpel16_mc11 |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel16_mc02_neon, export=1 |
||||
push {r4, lr} |
||||
lowpass_const r3 |
||||
sub r1, r1, r2, lsl #1 |
||||
mov r3, r2 |
||||
vpush {d8-d15} |
||||
bl \type\()_h264_qpel16_v_lowpass_neon |
||||
vpop {d8-d15} |
||||
pop {r4, pc} |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel16_mc12_neon, export=1 |
||||
push {r0, r1, r4-r5, r9-r11, lr} |
||||
\type\()_h264_qpel16_mc12: |
||||
lowpass_const r3 |
||||
mov r11, sp |
||||
A bic sp, sp, #15 |
||||
T bic r0, r11, #15 |
||||
T mov sp, r0 |
||||
sub sp, sp, #(16*16+16*12) |
||||
sub r1, r1, r2, lsl #1 |
||||
mov r0, sp |
||||
mov r3, r2 |
||||
vpush {d8-d15} |
||||
bl put_h264_qpel16_v_lowpass_neon_packed |
||||
mov r4, r0 |
||||
ldrd r0, r1, [r11], #8 |
||||
sub r1, r1, r3, lsl #1 |
||||
sub r1, r1, #2 |
||||
mov r2, r3 |
||||
bl \type\()_h264_qpel16_hv_lowpass_l2_neon |
||||
vpop {d8-d15} |
||||
mov sp, r11 |
||||
pop {r4-r5, r9-r11, pc} |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel16_mc22_neon, export=1 |
||||
push {r4, r9-r11, lr} |
||||
lowpass_const r3 |
||||
mov r11, sp |
||||
A bic sp, sp, #15 |
||||
T bic r4, r11, #15 |
||||
T mov sp, r4 |
||||
sub r1, r1, r2, lsl #1 |
||||
sub r1, r1, #2 |
||||
mov r3, r2 |
||||
sub sp, sp, #(16*12) |
||||
mov r4, sp |
||||
vpush {d8-d15} |
||||
bl \type\()_h264_qpel16_hv_lowpass_neon |
||||
vpop {d8-d15} |
||||
mov sp, r11 |
||||
pop {r4, r9-r11, pc} |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel16_mc32_neon, export=1 |
||||
push {r0, r1, r4-r5, r9-r11, lr} |
||||
add r1, r1, #1 |
||||
b \type\()_h264_qpel16_mc12 |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel16_mc03_neon, export=1 |
||||
push {r4, lr} |
||||
add r12, r1, r2 |
||||
b \type\()_h264_qpel16_mc01 |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel16_mc13_neon, export=1 |
||||
push {r0, r1, r4, r11, lr} |
||||
add r1, r1, r2 |
||||
b \type\()_h264_qpel16_mc11 |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel16_mc23_neon, export=1 |
||||
push {r0, r1, r4-r5, r9-r11, lr} |
||||
add r1, r1, r2 |
||||
b \type\()_h264_qpel16_mc21 |
||||
endfunc |
||||
|
||||
function ff_\type\()_h264_qpel16_mc33_neon, export=1 |
||||
add r1, r1, #1 |
||||
push {r0, r1, r4, r11, lr} |
||||
add r1, r1, r2 |
||||
sub r1, r1, #1 |
||||
b \type\()_h264_qpel16_mc11 |
||||
endfunc |
||||
.endm |
||||
|
||||
h264_qpel16 put |
||||
h264_qpel16 avg |
||||
@ -0,0 +1,86 @@ |
||||
/*
|
||||
* H.26L/H.264/AVC/JVT/14496-10/... encoder/decoder |
||||
* Copyright (c) 2003-2010 Michael Niedermayer <michaelni@gmx.at> |
||||
* |
||||
* This file is part of Libav. |
||||
* |
||||
* 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. |
||||
* |
||||
* 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 Libav; if not, write to the Free Software |
||||
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
||||
*/ |
||||
|
||||
#include "h264qpel.h" |
||||
|
||||
#define BIT_DEPTH 8 |
||||
#include "h264qpel_template.c" |
||||
#undef BIT_DEPTH |
||||
|
||||
#define BIT_DEPTH 9 |
||||
#include "h264qpel_template.c" |
||||
#undef BIT_DEPTH |
||||
|
||||
#define BIT_DEPTH 10 |
||||
#include "h264qpel_template.c" |
||||
#undef BIT_DEPTH |
||||
|
||||
void ff_h264qpel_init(H264QpelContext *c, int bit_depth) |
||||
{ |
||||
#undef FUNCC |
||||
#define FUNCC(f, depth) f ## _ ## depth ## _c |
||||
|
||||
#define dspfunc2(PFX, IDX, NUM, depth) \ |
||||
c->PFX ## _pixels_tab[IDX][ 0] = FUNCC(PFX ## NUM ## _mc00, depth); \
|
||||
c->PFX ## _pixels_tab[IDX][ 1] = FUNCC(PFX ## NUM ## _mc10, depth); \
|
||||
c->PFX ## _pixels_tab[IDX][ 2] = FUNCC(PFX ## NUM ## _mc20, depth); \
|
||||
c->PFX ## _pixels_tab[IDX][ 3] = FUNCC(PFX ## NUM ## _mc30, depth); \
|
||||
c->PFX ## _pixels_tab[IDX][ 4] = FUNCC(PFX ## NUM ## _mc01, depth); \
|
||||
c->PFX ## _pixels_tab[IDX][ 5] = FUNCC(PFX ## NUM ## _mc11, depth); \
|
||||
c->PFX ## _pixels_tab[IDX][ 6] = FUNCC(PFX ## NUM ## _mc21, depth); \
|
||||
c->PFX ## _pixels_tab[IDX][ 7] = FUNCC(PFX ## NUM ## _mc31, depth); \
|
||||
c->PFX ## _pixels_tab[IDX][ 8] = FUNCC(PFX ## NUM ## _mc02, depth); \
|
||||
c->PFX ## _pixels_tab[IDX][ 9] = FUNCC(PFX ## NUM ## _mc12, depth); \
|
||||
c->PFX ## _pixels_tab[IDX][10] = FUNCC(PFX ## NUM ## _mc22, depth); \
|
||||
c->PFX ## _pixels_tab[IDX][11] = FUNCC(PFX ## NUM ## _mc32, depth); \
|
||||
c->PFX ## _pixels_tab[IDX][12] = FUNCC(PFX ## NUM ## _mc03, depth); \
|
||||
c->PFX ## _pixels_tab[IDX][13] = FUNCC(PFX ## NUM ## _mc13, depth); \
|
||||
c->PFX ## _pixels_tab[IDX][14] = FUNCC(PFX ## NUM ## _mc23, depth); \
|
||||
c->PFX ## _pixels_tab[IDX][15] = FUNCC(PFX ## NUM ## _mc33, depth) |
||||
|
||||
#define SET_QPEL(depth) \ |
||||
dspfunc2(put_h264_qpel, 0, 16, depth); \
|
||||
dspfunc2(put_h264_qpel, 1, 8, depth); \
|
||||
dspfunc2(put_h264_qpel, 2, 4, depth); \
|
||||
dspfunc2(put_h264_qpel, 3, 2, depth); \
|
||||
dspfunc2(avg_h264_qpel, 0, 16, depth); \
|
||||
dspfunc2(avg_h264_qpel, 1, 8, depth); \
|
||||
dspfunc2(avg_h264_qpel, 2, 4, depth) |
||||
|
||||
switch (bit_depth) { |
||||
default: |
||||
SET_QPEL(8); |
||||
break; |
||||
case 9: |
||||
SET_QPEL(9); |
||||
break; |
||||
case 10: |
||||
SET_QPEL(10); |
||||
break; |
||||
} |
||||
|
||||
if (ARCH_ARM) |
||||
ff_h264qpel_init_arm(c, bit_depth); |
||||
if (ARCH_PPC) |
||||
ff_h264qpel_init_ppc(c, bit_depth); |
||||
if (ARCH_X86) |
||||
ff_h264qpel_init_x86(c, bit_depth); |
||||
} |
||||
@ -0,0 +1,38 @@ |
||||
/*
|
||||
* H.26L/H.264/AVC/JVT/14496-10/... encoder/decoder |
||||
* Copyright (c) 2003-2010 Michael Niedermayer <michaelni@gmx.at> |
||||
* |
||||
* This file is part of Libav. |
||||
* |
||||
* 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. |
||||
* |
||||
* 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 Libav; if not, write to the Free Software |
||||
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
||||
*/ |
||||
|
||||
#ifndef AVCODEC_H264QPEL_H |
||||
#define AVCODEC_H264QPEL_H |
||||
|
||||
#include "dsputil.h" |
||||
|
||||
typedef struct H264QpelContext { |
||||
qpel_mc_func put_h264_qpel_pixels_tab[4][16]; |
||||
qpel_mc_func avg_h264_qpel_pixels_tab[4][16]; |
||||
} H264QpelContext; |
||||
|
||||
void ff_h264qpel_init(H264QpelContext *c, int bit_depth); |
||||
|
||||
void ff_h264qpel_init_arm(H264QpelContext *c, int bit_depth); |
||||
void ff_h264qpel_init_ppc(H264QpelContext *c, int bit_depth); |
||||
void ff_h264qpel_init_x86(H264QpelContext *c, int bit_depth); |
||||
|
||||
#endif /* AVCODEC_H264QPEL_H */ |
||||
@ -0,0 +1,550 @@ |
||||
/*
|
||||
* H.26L/H.264/AVC/JVT/14496-10/... encoder/decoder |
||||
* Copyright (c) 2003-2010 Michael Niedermayer <michaelni@gmx.at> |
||||
* |
||||
* This file is part of Libav. |
||||
* |
||||
* 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. |
||||
* |
||||
* 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 Libav; if not, write to the Free Software |
||||
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
||||
*/ |
||||
|
||||
#include "libavutil/common.h" |
||||
#include "bit_depth_template.c" |
||||
#include "hpel_template.c" |
||||
|
||||
static inline void FUNC(copy_block2)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h) |
||||
{ |
||||
int i; |
||||
for(i=0; i<h; i++) |
||||
{ |
||||
AV_WN2P(dst , AV_RN2P(src )); |
||||
dst+=dstStride; |
||||
src+=srcStride; |
||||
} |
||||
} |
||||
|
||||
static inline void FUNC(copy_block4)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h) |
||||
{ |
||||
int i; |
||||
for(i=0; i<h; i++) |
||||
{ |
||||
AV_WN4P(dst , AV_RN4P(src )); |
||||
dst+=dstStride; |
||||
src+=srcStride; |
||||
} |
||||
} |
||||
|
||||
static inline void FUNC(copy_block8)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h) |
||||
{ |
||||
int i; |
||||
for(i=0; i<h; i++) |
||||
{ |
||||
AV_WN4P(dst , AV_RN4P(src )); |
||||
AV_WN4P(dst+4*sizeof(pixel), AV_RN4P(src+4*sizeof(pixel))); |
||||
dst+=dstStride; |
||||
src+=srcStride; |
||||
} |
||||
} |
||||
|
||||
static inline void FUNC(copy_block16)(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h) |
||||
{ |
||||
int i; |
||||
for(i=0; i<h; i++) |
||||
{ |
||||
AV_WN4P(dst , AV_RN4P(src )); |
||||
AV_WN4P(dst+ 4*sizeof(pixel), AV_RN4P(src+ 4*sizeof(pixel))); |
||||
AV_WN4P(dst+ 8*sizeof(pixel), AV_RN4P(src+ 8*sizeof(pixel))); |
||||
AV_WN4P(dst+12*sizeof(pixel), AV_RN4P(src+12*sizeof(pixel))); |
||||
dst+=dstStride; |
||||
src+=srcStride; |
||||
} |
||||
} |
||||
|
||||
#define H264_LOWPASS(OPNAME, OP, OP2) \ |
||||
static av_unused void FUNC(OPNAME ## h264_qpel2_h_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\
|
||||
const int h=2;\
|
||||
INIT_CLIP\
|
||||
int i;\
|
||||
pixel *dst = (pixel*)_dst;\
|
||||
pixel *src = (pixel*)_src;\
|
||||
dstStride /= sizeof(pixel);\
|
||||
srcStride /= sizeof(pixel);\
|
||||
for(i=0; i<h; i++)\
|
||||
{\
|
||||
OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]));\
|
||||
OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]));\
|
||||
dst+=dstStride;\
|
||||
src+=srcStride;\
|
||||
}\
|
||||
}\
|
||||
\
|
||||
static av_unused void FUNC(OPNAME ## h264_qpel2_v_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\
|
||||
const int w=2;\
|
||||
INIT_CLIP\
|
||||
int i;\
|
||||
pixel *dst = (pixel*)_dst;\
|
||||
pixel *src = (pixel*)_src;\
|
||||
dstStride /= sizeof(pixel);\
|
||||
srcStride /= sizeof(pixel);\
|
||||
for(i=0; i<w; i++)\
|
||||
{\
|
||||
const int srcB= src[-2*srcStride];\
|
||||
const int srcA= src[-1*srcStride];\
|
||||
const int src0= src[0 *srcStride];\
|
||||
const int src1= src[1 *srcStride];\
|
||||
const int src2= src[2 *srcStride];\
|
||||
const int src3= src[3 *srcStride];\
|
||||
const int src4= src[4 *srcStride];\
|
||||
OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
|
||||
OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
|
||||
dst++;\
|
||||
src++;\
|
||||
}\
|
||||
}\
|
||||
\
|
||||
static av_unused void FUNC(OPNAME ## h264_qpel2_hv_lowpass)(uint8_t *_dst, int16_t *tmp, uint8_t *_src, int dstStride, int tmpStride, int srcStride){\
|
||||
const int h=2;\
|
||||
const int w=2;\
|
||||
const int pad = (BIT_DEPTH > 9) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\
|
||||
INIT_CLIP\
|
||||
int i;\
|
||||
pixel *dst = (pixel*)_dst;\
|
||||
pixel *src = (pixel*)_src;\
|
||||
dstStride /= sizeof(pixel);\
|
||||
srcStride /= sizeof(pixel);\
|
||||
src -= 2*srcStride;\
|
||||
for(i=0; i<h+5; i++)\
|
||||
{\
|
||||
tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]) + pad;\
|
||||
tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]) + pad;\
|
||||
tmp+=tmpStride;\
|
||||
src+=srcStride;\
|
||||
}\
|
||||
tmp -= tmpStride*(h+5-2);\
|
||||
for(i=0; i<w; i++)\
|
||||
{\
|
||||
const int tmpB= tmp[-2*tmpStride] - pad;\
|
||||
const int tmpA= tmp[-1*tmpStride] - pad;\
|
||||
const int tmp0= tmp[0 *tmpStride] - pad;\
|
||||
const int tmp1= tmp[1 *tmpStride] - pad;\
|
||||
const int tmp2= tmp[2 *tmpStride] - pad;\
|
||||
const int tmp3= tmp[3 *tmpStride] - pad;\
|
||||
const int tmp4= tmp[4 *tmpStride] - pad;\
|
||||
OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\
|
||||
OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\
|
||||
dst++;\
|
||||
tmp++;\
|
||||
}\
|
||||
}\
|
||||
static void FUNC(OPNAME ## h264_qpel4_h_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\
|
||||
const int h=4;\
|
||||
INIT_CLIP\
|
||||
int i;\
|
||||
pixel *dst = (pixel*)_dst;\
|
||||
pixel *src = (pixel*)_src;\
|
||||
dstStride /= sizeof(pixel);\
|
||||
srcStride /= sizeof(pixel);\
|
||||
for(i=0; i<h; i++)\
|
||||
{\
|
||||
OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]));\
|
||||
OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]));\
|
||||
OP(dst[2], (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5]));\
|
||||
OP(dst[3], (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6]));\
|
||||
dst+=dstStride;\
|
||||
src+=srcStride;\
|
||||
}\
|
||||
}\
|
||||
\
|
||||
static void FUNC(OPNAME ## h264_qpel4_v_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\
|
||||
const int w=4;\
|
||||
INIT_CLIP\
|
||||
int i;\
|
||||
pixel *dst = (pixel*)_dst;\
|
||||
pixel *src = (pixel*)_src;\
|
||||
dstStride /= sizeof(pixel);\
|
||||
srcStride /= sizeof(pixel);\
|
||||
for(i=0; i<w; i++)\
|
||||
{\
|
||||
const int srcB= src[-2*srcStride];\
|
||||
const int srcA= src[-1*srcStride];\
|
||||
const int src0= src[0 *srcStride];\
|
||||
const int src1= src[1 *srcStride];\
|
||||
const int src2= src[2 *srcStride];\
|
||||
const int src3= src[3 *srcStride];\
|
||||
const int src4= src[4 *srcStride];\
|
||||
const int src5= src[5 *srcStride];\
|
||||
const int src6= src[6 *srcStride];\
|
||||
OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
|
||||
OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
|
||||
OP(dst[2*dstStride], (src2+src3)*20 - (src1+src4)*5 + (src0+src5));\
|
||||
OP(dst[3*dstStride], (src3+src4)*20 - (src2+src5)*5 + (src1+src6));\
|
||||
dst++;\
|
||||
src++;\
|
||||
}\
|
||||
}\
|
||||
\
|
||||
static void FUNC(OPNAME ## h264_qpel4_hv_lowpass)(uint8_t *_dst, int16_t *tmp, uint8_t *_src, int dstStride, int tmpStride, int srcStride){\
|
||||
const int h=4;\
|
||||
const int w=4;\
|
||||
const int pad = (BIT_DEPTH > 9) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\
|
||||
INIT_CLIP\
|
||||
int i;\
|
||||
pixel *dst = (pixel*)_dst;\
|
||||
pixel *src = (pixel*)_src;\
|
||||
dstStride /= sizeof(pixel);\
|
||||
srcStride /= sizeof(pixel);\
|
||||
src -= 2*srcStride;\
|
||||
for(i=0; i<h+5; i++)\
|
||||
{\
|
||||
tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]) + pad;\
|
||||
tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]) + pad;\
|
||||
tmp[2]= (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5]) + pad;\
|
||||
tmp[3]= (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6]) + pad;\
|
||||
tmp+=tmpStride;\
|
||||
src+=srcStride;\
|
||||
}\
|
||||
tmp -= tmpStride*(h+5-2);\
|
||||
for(i=0; i<w; i++)\
|
||||
{\
|
||||
const int tmpB= tmp[-2*tmpStride] - pad;\
|
||||
const int tmpA= tmp[-1*tmpStride] - pad;\
|
||||
const int tmp0= tmp[0 *tmpStride] - pad;\
|
||||
const int tmp1= tmp[1 *tmpStride] - pad;\
|
||||
const int tmp2= tmp[2 *tmpStride] - pad;\
|
||||
const int tmp3= tmp[3 *tmpStride] - pad;\
|
||||
const int tmp4= tmp[4 *tmpStride] - pad;\
|
||||
const int tmp5= tmp[5 *tmpStride] - pad;\
|
||||
const int tmp6= tmp[6 *tmpStride] - pad;\
|
||||
OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\
|
||||
OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\
|
||||
OP2(dst[2*dstStride], (tmp2+tmp3)*20 - (tmp1+tmp4)*5 + (tmp0+tmp5));\
|
||||
OP2(dst[3*dstStride], (tmp3+tmp4)*20 - (tmp2+tmp5)*5 + (tmp1+tmp6));\
|
||||
dst++;\
|
||||
tmp++;\
|
||||
}\
|
||||
}\
|
||||
\
|
||||
static void FUNC(OPNAME ## h264_qpel8_h_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\
|
||||
const int h=8;\
|
||||
INIT_CLIP\
|
||||
int i;\
|
||||
pixel *dst = (pixel*)_dst;\
|
||||
pixel *src = (pixel*)_src;\
|
||||
dstStride /= sizeof(pixel);\
|
||||
srcStride /= sizeof(pixel);\
|
||||
for(i=0; i<h; i++)\
|
||||
{\
|
||||
OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3 ]));\
|
||||
OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4 ]));\
|
||||
OP(dst[2], (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5 ]));\
|
||||
OP(dst[3], (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6 ]));\
|
||||
OP(dst[4], (src[4]+src[5])*20 - (src[3 ]+src[6])*5 + (src[2 ]+src[7 ]));\
|
||||
OP(dst[5], (src[5]+src[6])*20 - (src[4 ]+src[7])*5 + (src[3 ]+src[8 ]));\
|
||||
OP(dst[6], (src[6]+src[7])*20 - (src[5 ]+src[8])*5 + (src[4 ]+src[9 ]));\
|
||||
OP(dst[7], (src[7]+src[8])*20 - (src[6 ]+src[9])*5 + (src[5 ]+src[10]));\
|
||||
dst+=dstStride;\
|
||||
src+=srcStride;\
|
||||
}\
|
||||
}\
|
||||
\
|
||||
static void FUNC(OPNAME ## h264_qpel8_v_lowpass)(uint8_t *_dst, uint8_t *_src, int dstStride, int srcStride){\
|
||||
const int w=8;\
|
||||
INIT_CLIP\
|
||||
int i;\
|
||||
pixel *dst = (pixel*)_dst;\
|
||||
pixel *src = (pixel*)_src;\
|
||||
dstStride /= sizeof(pixel);\
|
||||
srcStride /= sizeof(pixel);\
|
||||
for(i=0; i<w; i++)\
|
||||
{\
|
||||
const int srcB= src[-2*srcStride];\
|
||||
const int srcA= src[-1*srcStride];\
|
||||
const int src0= src[0 *srcStride];\
|
||||
const int src1= src[1 *srcStride];\
|
||||
const int src2= src[2 *srcStride];\
|
||||
const int src3= src[3 *srcStride];\
|
||||
const int src4= src[4 *srcStride];\
|
||||
const int src5= src[5 *srcStride];\
|
||||
const int src6= src[6 *srcStride];\
|
||||
const int src7= src[7 *srcStride];\
|
||||
const int src8= src[8 *srcStride];\
|
||||
const int src9= src[9 *srcStride];\
|
||||
const int src10=src[10*srcStride];\
|
||||
OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
|
||||
OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
|
||||
OP(dst[2*dstStride], (src2+src3)*20 - (src1+src4)*5 + (src0+src5));\
|
||||
OP(dst[3*dstStride], (src3+src4)*20 - (src2+src5)*5 + (src1+src6));\
|
||||
OP(dst[4*dstStride], (src4+src5)*20 - (src3+src6)*5 + (src2+src7));\
|
||||
OP(dst[5*dstStride], (src5+src6)*20 - (src4+src7)*5 + (src3+src8));\
|
||||
OP(dst[6*dstStride], (src6+src7)*20 - (src5+src8)*5 + (src4+src9));\
|
||||
OP(dst[7*dstStride], (src7+src8)*20 - (src6+src9)*5 + (src5+src10));\
|
||||
dst++;\
|
||||
src++;\
|
||||
}\
|
||||
}\
|
||||
\
|
||||
static void FUNC(OPNAME ## h264_qpel8_hv_lowpass)(uint8_t *_dst, int16_t *tmp, uint8_t *_src, int dstStride, int tmpStride, int srcStride){\
|
||||
const int h=8;\
|
||||
const int w=8;\
|
||||
const int pad = (BIT_DEPTH > 9) ? (-10 * ((1<<BIT_DEPTH)-1)) : 0;\
|
||||
INIT_CLIP\
|
||||
int i;\
|
||||
pixel *dst = (pixel*)_dst;\
|
||||
pixel *src = (pixel*)_src;\
|
||||
dstStride /= sizeof(pixel);\
|
||||
srcStride /= sizeof(pixel);\
|
||||
src -= 2*srcStride;\
|
||||
for(i=0; i<h+5; i++)\
|
||||
{\
|
||||
tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3 ]) + pad;\
|
||||
tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4 ]) + pad;\
|
||||
tmp[2]= (src[2]+src[3])*20 - (src[1 ]+src[4])*5 + (src[0 ]+src[5 ]) + pad;\
|
||||
tmp[3]= (src[3]+src[4])*20 - (src[2 ]+src[5])*5 + (src[1 ]+src[6 ]) + pad;\
|
||||
tmp[4]= (src[4]+src[5])*20 - (src[3 ]+src[6])*5 + (src[2 ]+src[7 ]) + pad;\
|
||||
tmp[5]= (src[5]+src[6])*20 - (src[4 ]+src[7])*5 + (src[3 ]+src[8 ]) + pad;\
|
||||
tmp[6]= (src[6]+src[7])*20 - (src[5 ]+src[8])*5 + (src[4 ]+src[9 ]) + pad;\
|
||||
tmp[7]= (src[7]+src[8])*20 - (src[6 ]+src[9])*5 + (src[5 ]+src[10]) + pad;\
|
||||
tmp+=tmpStride;\
|
||||
src+=srcStride;\
|
||||
}\
|
||||
tmp -= tmpStride*(h+5-2);\
|
||||
for(i=0; i<w; i++)\
|
||||
{\
|
||||
const int tmpB= tmp[-2*tmpStride] - pad;\
|
||||
const int tmpA= tmp[-1*tmpStride] - pad;\
|
||||
const int tmp0= tmp[0 *tmpStride] - pad;\
|
||||
const int tmp1= tmp[1 *tmpStride] - pad;\
|
||||
const int tmp2= tmp[2 *tmpStride] - pad;\
|
||||
const int tmp3= tmp[3 *tmpStride] - pad;\
|
||||
const int tmp4= tmp[4 *tmpStride] - pad;\
|
||||
const int tmp5= tmp[5 *tmpStride] - pad;\
|
||||
const int tmp6= tmp[6 *tmpStride] - pad;\
|
||||
const int tmp7= tmp[7 *tmpStride] - pad;\
|
||||
const int tmp8= tmp[8 *tmpStride] - pad;\
|
||||
const int tmp9= tmp[9 *tmpStride] - pad;\
|
||||
const int tmp10=tmp[10*tmpStride] - pad;\
|
||||
OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\
|
||||
OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\
|
||||
OP2(dst[2*dstStride], (tmp2+tmp3)*20 - (tmp1+tmp4)*5 + (tmp0+tmp5));\
|
||||
OP2(dst[3*dstStride], (tmp3+tmp4)*20 - (tmp2+tmp5)*5 + (tmp1+tmp6));\
|
||||
OP2(dst[4*dstStride], (tmp4+tmp5)*20 - (tmp3+tmp6)*5 + (tmp2+tmp7));\
|
||||
OP2(dst[5*dstStride], (tmp5+tmp6)*20 - (tmp4+tmp7)*5 + (tmp3+tmp8));\
|
||||
OP2(dst[6*dstStride], (tmp6+tmp7)*20 - (tmp5+tmp8)*5 + (tmp4+tmp9));\
|
||||
OP2(dst[7*dstStride], (tmp7+tmp8)*20 - (tmp6+tmp9)*5 + (tmp5+tmp10));\
|
||||
dst++;\
|
||||
tmp++;\
|
||||
}\
|
||||
}\
|
||||
\
|
||||
static void FUNC(OPNAME ## h264_qpel16_v_lowpass)(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
|
||||
FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst , src , dstStride, srcStride);\
|
||||
FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\
|
||||
src += 8*srcStride;\
|
||||
dst += 8*dstStride;\
|
||||
FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst , src , dstStride, srcStride);\
|
||||
FUNC(OPNAME ## h264_qpel8_v_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\
|
||||
}\
|
||||
\
|
||||
static void FUNC(OPNAME ## h264_qpel16_h_lowpass)(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
|
||||
FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst , src , dstStride, srcStride);\
|
||||
FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\
|
||||
src += 8*srcStride;\
|
||||
dst += 8*dstStride;\
|
||||
FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst , src , dstStride, srcStride);\
|
||||
FUNC(OPNAME ## h264_qpel8_h_lowpass)(dst+8*sizeof(pixel), src+8*sizeof(pixel), dstStride, srcStride);\
|
||||
}\
|
||||
\
|
||||
static void FUNC(OPNAME ## h264_qpel16_hv_lowpass)(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\
|
||||
FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst , tmp , src , dstStride, tmpStride, srcStride);\
|
||||
FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst+8*sizeof(pixel), tmp+8, src+8*sizeof(pixel), dstStride, tmpStride, srcStride);\
|
||||
src += 8*srcStride;\
|
||||
dst += 8*dstStride;\
|
||||
FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst , tmp , src , dstStride, tmpStride, srcStride);\
|
||||
FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst+8*sizeof(pixel), tmp+8, src+8*sizeof(pixel), dstStride, tmpStride, srcStride);\
|
||||
}\
|
||||
|
||||
#define H264_MC(OPNAME, SIZE) \ |
||||
static av_unused void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc00)(uint8_t *dst, uint8_t *src, int stride){\
|
||||
FUNCC(OPNAME ## pixels ## SIZE)(dst, src, stride, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc10)(uint8_t *dst, uint8_t *src, int stride){\
|
||||
uint8_t half[SIZE*SIZE*sizeof(pixel)];\
|
||||
FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(half, src, SIZE*sizeof(pixel), stride);\
|
||||
FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, src, half, stride, stride, SIZE*sizeof(pixel), SIZE);\
|
||||
}\
|
||||
\
|
||||
static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc20)(uint8_t *dst, uint8_t *src, int stride){\
|
||||
FUNC(OPNAME ## h264_qpel ## SIZE ## _h_lowpass)(dst, src, stride, stride);\
|
||||
}\
|
||||
\
|
||||
static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc30)(uint8_t *dst, uint8_t *src, int stride){\
|
||||
uint8_t half[SIZE*SIZE*sizeof(pixel)];\
|
||||
FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(half, src, SIZE*sizeof(pixel), stride);\
|
||||
FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, src+sizeof(pixel), half, stride, stride, SIZE*sizeof(pixel), SIZE);\
|
||||
}\
|
||||
\
|
||||
static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc01)(uint8_t *dst, uint8_t *src, int stride){\
|
||||
uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
|
||||
uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
|
||||
uint8_t half[SIZE*SIZE*sizeof(pixel)];\
|
||||
FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel), stride, SIZE + 5);\
|
||||
FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(half, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
|
||||
FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, full_mid, half, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
|
||||
}\
|
||||
\
|
||||
static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc02)(uint8_t *dst, uint8_t *src, int stride){\
|
||||
uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
|
||||
uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
|
||||
FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel), stride, SIZE + 5);\
|
||||
FUNC(OPNAME ## h264_qpel ## SIZE ## _v_lowpass)(dst, full_mid, stride, SIZE*sizeof(pixel));\
|
||||
}\
|
||||
\
|
||||
static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc03)(uint8_t *dst, uint8_t *src, int stride){\
|
||||
uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
|
||||
uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
|
||||
uint8_t half[SIZE*SIZE*sizeof(pixel)];\
|
||||
FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel), stride, SIZE + 5);\
|
||||
FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(half, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
|
||||
FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, full_mid+SIZE*sizeof(pixel), half, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
|
||||
}\
|
||||
\
|
||||
static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc11)(uint8_t *dst, uint8_t *src, int stride){\
|
||||
uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
|
||||
uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
|
||||
uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
|
||||
uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
|
||||
FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\
|
||||
FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel), stride, SIZE + 5);\
|
||||
FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
|
||||
FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
|
||||
}\
|
||||
\
|
||||
static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc31)(uint8_t *dst, uint8_t *src, int stride){\
|
||||
uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
|
||||
uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
|
||||
uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
|
||||
uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
|
||||
FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\
|
||||
FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel), stride, SIZE + 5);\
|
||||
FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
|
||||
FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
|
||||
}\
|
||||
\
|
||||
static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc13)(uint8_t *dst, uint8_t *src, int stride){\
|
||||
uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
|
||||
uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
|
||||
uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
|
||||
uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
|
||||
FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\
|
||||
FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel), stride, SIZE + 5);\
|
||||
FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
|
||||
FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
|
||||
}\
|
||||
\
|
||||
static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc33)(uint8_t *dst, uint8_t *src, int stride){\
|
||||
uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
|
||||
uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
|
||||
uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
|
||||
uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
|
||||
FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\
|
||||
FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel), stride, SIZE + 5);\
|
||||
FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
|
||||
FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
|
||||
}\
|
||||
\
|
||||
static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc22)(uint8_t *dst, uint8_t *src, int stride){\
|
||||
int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
|
||||
FUNC(OPNAME ## h264_qpel ## SIZE ## _hv_lowpass)(dst, tmp, src, stride, SIZE*sizeof(pixel), stride);\
|
||||
}\
|
||||
\
|
||||
static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc21)(uint8_t *dst, uint8_t *src, int stride){\
|
||||
int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
|
||||
uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
|
||||
uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\
|
||||
FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src, SIZE*sizeof(pixel), stride);\
|
||||
FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\
|
||||
FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
|
||||
}\
|
||||
\
|
||||
static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc23)(uint8_t *dst, uint8_t *src, int stride){\
|
||||
int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
|
||||
uint8_t halfH[SIZE*SIZE*sizeof(pixel)];\
|
||||
uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\
|
||||
FUNC(put_h264_qpel ## SIZE ## _h_lowpass)(halfH, src + stride, SIZE*sizeof(pixel), stride);\
|
||||
FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\
|
||||
FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfH, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
|
||||
}\
|
||||
\
|
||||
static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc12)(uint8_t *dst, uint8_t *src, int stride){\
|
||||
uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
|
||||
uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
|
||||
int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
|
||||
uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
|
||||
uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\
|
||||
FUNC(copy_block ## SIZE )(full, src - stride*2, SIZE*sizeof(pixel), stride, SIZE + 5);\
|
||||
FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
|
||||
FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\
|
||||
FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfV, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
|
||||
}\
|
||||
\
|
||||
static void FUNCC(OPNAME ## h264_qpel ## SIZE ## _mc32)(uint8_t *dst, uint8_t *src, int stride){\
|
||||
uint8_t full[SIZE*(SIZE+5)*sizeof(pixel)];\
|
||||
uint8_t * const full_mid= full + SIZE*2*sizeof(pixel);\
|
||||
int16_t tmp[SIZE*(SIZE+5)*sizeof(pixel)];\
|
||||
uint8_t halfV[SIZE*SIZE*sizeof(pixel)];\
|
||||
uint8_t halfHV[SIZE*SIZE*sizeof(pixel)];\
|
||||
FUNC(copy_block ## SIZE )(full, src - stride*2 + sizeof(pixel), SIZE*sizeof(pixel), stride, SIZE + 5);\
|
||||
FUNC(put_h264_qpel ## SIZE ## _v_lowpass)(halfV, full_mid, SIZE*sizeof(pixel), SIZE*sizeof(pixel));\
|
||||
FUNC(put_h264_qpel ## SIZE ## _hv_lowpass)(halfHV, tmp, src, SIZE*sizeof(pixel), SIZE*sizeof(pixel), stride);\
|
||||
FUNC(OPNAME ## pixels ## SIZE ## _l2)(dst, halfV, halfHV, stride, SIZE*sizeof(pixel), SIZE*sizeof(pixel), SIZE);\
|
||||
}\
|
||||
|
||||
#define op_avg(a, b) a = (((a)+CLIP(((b) + 16)>>5)+1)>>1) |
||||
//#define op_avg2(a, b) a = (((a)*w1+cm[((b) + 16)>>5]*w2 + o + 64)>>7)
|
||||
#define op_put(a, b) a = CLIP(((b) + 16)>>5) |
||||
#define op2_avg(a, b) a = (((a)+CLIP(((b) + 512)>>10)+1)>>1) |
||||
#define op2_put(a, b) a = CLIP(((b) + 512)>>10) |
||||
|
||||
H264_LOWPASS(put_ , op_put, op2_put) |
||||
H264_LOWPASS(avg_ , op_avg, op2_avg) |
||||
H264_MC(put_, 2) |
||||
H264_MC(put_, 4) |
||||
H264_MC(put_, 8) |
||||
H264_MC(put_, 16) |
||||
H264_MC(avg_, 4) |
||||
H264_MC(avg_, 8) |
||||
H264_MC(avg_, 16) |
||||
|
||||
#undef op_avg |
||||
#undef op_put |
||||
#undef op2_avg |
||||
#undef op2_put |
||||
|
||||
#if BIT_DEPTH == 8 |
||||
# define put_h264_qpel8_mc00_8_c ff_put_pixels8x8_8_c |
||||
# define avg_h264_qpel8_mc00_8_c ff_avg_pixels8x8_8_c |
||||
# define put_h264_qpel16_mc00_8_c ff_put_pixels16x16_8_c |
||||
# define avg_h264_qpel16_mc00_8_c ff_avg_pixels16x16_8_c |
||||
#elif BIT_DEPTH == 9 |
||||
# define put_h264_qpel8_mc00_9_c ff_put_pixels8x8_9_c |
||||
# define avg_h264_qpel8_mc00_9_c ff_avg_pixels8x8_9_c |
||||
# define put_h264_qpel16_mc00_9_c ff_put_pixels16x16_9_c |
||||
# define avg_h264_qpel16_mc00_9_c ff_avg_pixels16x16_9_c |
||||
#elif BIT_DEPTH == 10 |
||||
# define put_h264_qpel8_mc00_10_c ff_put_pixels8x8_10_c |
||||
# define avg_h264_qpel8_mc00_10_c ff_avg_pixels8x8_10_c |
||||
# define put_h264_qpel16_mc00_10_c ff_put_pixels16x16_10_c |
||||
# define avg_h264_qpel16_mc00_10_c ff_avg_pixels16x16_10_c |
||||
#endif |
||||
@ -0,0 +1,100 @@ |
||||
/*
|
||||
* Copyright (c) 2000, 2001 Fabrice Bellard |
||||
* Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at> |
||||
* |
||||
* This file is part of Libav. |
||||
* |
||||
* 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. |
||||
* |
||||
* 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 Libav; if not, write to the Free Software |
||||
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
||||
*/ |
||||
|
||||
#define DEF_HPEL(OPNAME, OP) \ |
||||
static inline void FUNCC(OPNAME ## _pixels2)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
|
||||
int i;\
|
||||
for(i=0; i<h; i++){\
|
||||
OP(*((pixel2*)(block )), AV_RN2P(pixels ));\
|
||||
pixels+=line_size;\
|
||||
block +=line_size;\
|
||||
}\
|
||||
}\
|
||||
static inline void FUNCC(OPNAME ## _pixels4)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
|
||||
int i;\
|
||||
for(i=0; i<h; i++){\
|
||||
OP(*((pixel4*)(block )), AV_RN4P(pixels ));\
|
||||
pixels+=line_size;\
|
||||
block +=line_size;\
|
||||
}\
|
||||
}\
|
||||
static inline void FUNCC(OPNAME ## _pixels8)(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
|
||||
int i;\
|
||||
for(i=0; i<h; i++){\
|
||||
OP(*((pixel4*)(block )), AV_RN4P(pixels ));\
|
||||
OP(*((pixel4*)(block+4*sizeof(pixel))), AV_RN4P(pixels+4*sizeof(pixel)));\
|
||||
pixels+=line_size;\
|
||||
block +=line_size;\
|
||||
}\
|
||||
}\
|
||||
\
|
||||
static inline void FUNC(OPNAME ## _pixels8_l2)(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++){\
|
||||
pixel4 a,b;\
|
||||
a= AV_RN4P(&src1[i*src_stride1 ]);\
|
||||
b= AV_RN4P(&src2[i*src_stride2 ]);\
|
||||
OP(*((pixel4*)&dst[i*dst_stride ]), rnd_avg_pixel4(a, b));\
|
||||
a= AV_RN4P(&src1[i*src_stride1+4*sizeof(pixel)]);\
|
||||
b= AV_RN4P(&src2[i*src_stride2+4*sizeof(pixel)]);\
|
||||
OP(*((pixel4*)&dst[i*dst_stride+4*sizeof(pixel)]), rnd_avg_pixel4(a, b));\
|
||||
}\
|
||||
}\
|
||||
\
|
||||
static inline void FUNC(OPNAME ## _pixels4_l2)(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++){\
|
||||
pixel4 a,b;\
|
||||
a= AV_RN4P(&src1[i*src_stride1 ]);\
|
||||
b= AV_RN4P(&src2[i*src_stride2 ]);\
|
||||
OP(*((pixel4*)&dst[i*dst_stride ]), rnd_avg_pixel4(a, b));\
|
||||
}\
|
||||
}\
|
||||
\
|
||||
static inline void FUNC(OPNAME ## _pixels2_l2)(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++){\
|
||||
pixel4 a,b;\
|
||||
a= AV_RN2P(&src1[i*src_stride1 ]);\
|
||||
b= AV_RN2P(&src2[i*src_stride2 ]);\
|
||||
OP(*((pixel2*)&dst[i*dst_stride ]), rnd_avg_pixel4(a, b));\
|
||||
}\
|
||||
}\
|
||||
\
|
||||
static inline void FUNC(OPNAME ## _pixels16_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
|
||||
int src_stride1, int src_stride2, int h){\
|
||||
FUNC(OPNAME ## _pixels8_l2)(dst , src1 , src2 , dst_stride, src_stride1, src_stride2, h);\
|
||||
FUNC(OPNAME ## _pixels8_l2)(dst+8*sizeof(pixel), src1+8*sizeof(pixel), src2+8*sizeof(pixel), dst_stride, src_stride1, src_stride2, h);\
|
||||
}\
|
||||
\
|
||||
CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16) , FUNCC(OPNAME ## _pixels8) , 8*sizeof(pixel)) |
||||
|
||||
|
||||
#define op_avg(a, b) a = rnd_avg_pixel4(a, b) |
||||
#define op_put(a, b) a = b |
||||
|
||||
DEF_HPEL(avg, op_avg) |
||||
DEF_HPEL(put, op_put) |
||||
#undef op_avg |
||||
#undef op_put |
||||
@ -0,0 +1,319 @@ |
||||
/*
|
||||
* Copyright (c) 2004 Romain Dolbeau <romain@dolbeau.org> |
||||
* |
||||
* This file is part of Libav. |
||||
* |
||||
* 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. |
||||
* |
||||
* 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 Libav; if not, write to the Free Software |
||||
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
||||
*/ |
||||
|
||||
#include "config.h" |
||||
#include "libavcodec/h264qpel.h" |
||||
|
||||
#if HAVE_ALTIVEC |
||||
#include "libavutil/cpu.h" |
||||
#include "libavutil/intreadwrite.h" |
||||
#include "libavutil/ppc/types_altivec.h" |
||||
#include "libavutil/ppc/util_altivec.h" |
||||
#include "dsputil_altivec.h" |
||||
|
||||
#define PUT_OP_U8_ALTIVEC(d, s, dst) d = s |
||||
#define AVG_OP_U8_ALTIVEC(d, s, dst) d = vec_avg(dst, s) |
||||
|
||||
#define OP_U8_ALTIVEC PUT_OP_U8_ALTIVEC |
||||
#define PREFIX_h264_chroma_mc8_altivec put_h264_chroma_mc8_altivec |
||||
#define PREFIX_h264_chroma_mc8_num altivec_put_h264_chroma_mc8_num |
||||
#define PREFIX_h264_qpel16_h_lowpass_altivec put_h264_qpel16_h_lowpass_altivec |
||||
#define PREFIX_h264_qpel16_h_lowpass_num altivec_put_h264_qpel16_h_lowpass_num |
||||
#define PREFIX_h264_qpel16_v_lowpass_altivec put_h264_qpel16_v_lowpass_altivec |
||||
#define PREFIX_h264_qpel16_v_lowpass_num altivec_put_h264_qpel16_v_lowpass_num |
||||
#define PREFIX_h264_qpel16_hv_lowpass_altivec put_h264_qpel16_hv_lowpass_altivec |
||||
#define PREFIX_h264_qpel16_hv_lowpass_num altivec_put_h264_qpel16_hv_lowpass_num |
||||
#include "h264_qpel_template.c" |
||||
#undef OP_U8_ALTIVEC |
||||
#undef PREFIX_h264_chroma_mc8_altivec |
||||
#undef PREFIX_h264_chroma_mc8_num |
||||
#undef PREFIX_h264_qpel16_h_lowpass_altivec |
||||
#undef PREFIX_h264_qpel16_h_lowpass_num |
||||
#undef PREFIX_h264_qpel16_v_lowpass_altivec |
||||
#undef PREFIX_h264_qpel16_v_lowpass_num |
||||
#undef PREFIX_h264_qpel16_hv_lowpass_altivec |
||||
#undef PREFIX_h264_qpel16_hv_lowpass_num |
||||
|
||||
#define OP_U8_ALTIVEC AVG_OP_U8_ALTIVEC |
||||
#define PREFIX_h264_chroma_mc8_altivec avg_h264_chroma_mc8_altivec |
||||
#define PREFIX_h264_chroma_mc8_num altivec_avg_h264_chroma_mc8_num |
||||
#define PREFIX_h264_qpel16_h_lowpass_altivec avg_h264_qpel16_h_lowpass_altivec |
||||
#define PREFIX_h264_qpel16_h_lowpass_num altivec_avg_h264_qpel16_h_lowpass_num |
||||
#define PREFIX_h264_qpel16_v_lowpass_altivec avg_h264_qpel16_v_lowpass_altivec |
||||
#define PREFIX_h264_qpel16_v_lowpass_num altivec_avg_h264_qpel16_v_lowpass_num |
||||
#define PREFIX_h264_qpel16_hv_lowpass_altivec avg_h264_qpel16_hv_lowpass_altivec |
||||
#define PREFIX_h264_qpel16_hv_lowpass_num altivec_avg_h264_qpel16_hv_lowpass_num |
||||
#include "h264_qpel_template.c" |
||||
#undef OP_U8_ALTIVEC |
||||
#undef PREFIX_h264_chroma_mc8_altivec |
||||
#undef PREFIX_h264_chroma_mc8_num |
||||
#undef PREFIX_h264_qpel16_h_lowpass_altivec |
||||
#undef PREFIX_h264_qpel16_h_lowpass_num |
||||
#undef PREFIX_h264_qpel16_v_lowpass_altivec |
||||
#undef PREFIX_h264_qpel16_v_lowpass_num |
||||
#undef PREFIX_h264_qpel16_hv_lowpass_altivec |
||||
#undef PREFIX_h264_qpel16_hv_lowpass_num |
||||
|
||||
#define H264_MC(OPNAME, SIZE, CODETYPE) \ |
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc00_ ## CODETYPE (uint8_t *dst, uint8_t *src, int stride){\
|
||||
ff_ ## OPNAME ## pixels ## SIZE ## _ ## CODETYPE(dst, src, stride, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc10_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){ \
|
||||
DECLARE_ALIGNED(16, uint8_t, half)[SIZE*SIZE];\
|
||||
put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(half, src, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, src, half, stride, stride, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc20_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
OPNAME ## h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(dst, src, stride, stride);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc30_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED(16, uint8_t, half)[SIZE*SIZE];\
|
||||
put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(half, src, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, src+1, half, stride, stride, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc01_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED(16, uint8_t, half)[SIZE*SIZE];\
|
||||
put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(half, src, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, src, half, stride, stride, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc02_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
OPNAME ## h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(dst, src, stride, stride);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc03_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED(16, uint8_t, half)[SIZE*SIZE];\
|
||||
put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(half, src, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, src+stride, half, stride, stride, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc11_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED(16, uint8_t, halfH)[SIZE*SIZE];\
|
||||
DECLARE_ALIGNED(16, uint8_t, halfV)[SIZE*SIZE];\
|
||||
put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src, SIZE, stride);\
|
||||
put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfV, stride, SIZE, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc31_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED(16, uint8_t, halfH)[SIZE*SIZE];\
|
||||
DECLARE_ALIGNED(16, uint8_t, halfV)[SIZE*SIZE];\
|
||||
put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src, SIZE, stride);\
|
||||
put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src+1, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfV, stride, SIZE, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc13_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED(16, uint8_t, halfH)[SIZE*SIZE];\
|
||||
DECLARE_ALIGNED(16, uint8_t, halfV)[SIZE*SIZE];\
|
||||
put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src + stride, SIZE, stride);\
|
||||
put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfV, stride, SIZE, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc33_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED(16, uint8_t, halfH)[SIZE*SIZE];\
|
||||
DECLARE_ALIGNED(16, uint8_t, halfV)[SIZE*SIZE];\
|
||||
put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src + stride, SIZE, stride);\
|
||||
put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src+1, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfV, stride, SIZE, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc22_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED(16, int16_t, tmp)[SIZE*(SIZE+8)];\
|
||||
OPNAME ## h264_qpel ## SIZE ## _hv_lowpass_ ## CODETYPE(dst, tmp, src, stride, SIZE, stride);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc21_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED(16, uint8_t, halfH)[SIZE*SIZE];\
|
||||
DECLARE_ALIGNED(16, uint8_t, halfHV)[SIZE*SIZE];\
|
||||
DECLARE_ALIGNED(16, int16_t, tmp)[SIZE*(SIZE+8)];\
|
||||
put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src, SIZE, stride);\
|
||||
put_h264_qpel ## SIZE ## _hv_lowpass_ ## CODETYPE(halfHV, tmp, src, SIZE, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfHV, stride, SIZE, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc23_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED(16, uint8_t, halfH)[SIZE*SIZE];\
|
||||
DECLARE_ALIGNED(16, uint8_t, halfHV)[SIZE*SIZE];\
|
||||
DECLARE_ALIGNED(16, int16_t, tmp)[SIZE*(SIZE+8)];\
|
||||
put_h264_qpel ## SIZE ## _h_lowpass_ ## CODETYPE(halfH, src + stride, SIZE, stride);\
|
||||
put_h264_qpel ## SIZE ## _hv_lowpass_ ## CODETYPE(halfHV, tmp, src, SIZE, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfH, halfHV, stride, SIZE, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc12_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED(16, uint8_t, halfV)[SIZE*SIZE];\
|
||||
DECLARE_ALIGNED(16, uint8_t, halfHV)[SIZE*SIZE];\
|
||||
DECLARE_ALIGNED(16, int16_t, tmp)[SIZE*(SIZE+8)];\
|
||||
put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src, SIZE, stride);\
|
||||
put_h264_qpel ## SIZE ## _hv_lowpass_ ## CODETYPE(halfHV, tmp, src, SIZE, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfV, halfHV, stride, SIZE, SIZE);\
|
||||
}\
|
||||
\
|
||||
static void OPNAME ## h264_qpel ## SIZE ## _mc32_ ## CODETYPE(uint8_t *dst, uint8_t *src, int stride){\
|
||||
DECLARE_ALIGNED(16, uint8_t, halfV)[SIZE*SIZE];\
|
||||
DECLARE_ALIGNED(16, uint8_t, halfHV)[SIZE*SIZE];\
|
||||
DECLARE_ALIGNED(16, int16_t, tmp)[SIZE*(SIZE+8)];\
|
||||
put_h264_qpel ## SIZE ## _v_lowpass_ ## CODETYPE(halfV, src+1, SIZE, stride);\
|
||||
put_h264_qpel ## SIZE ## _hv_lowpass_ ## CODETYPE(halfHV, tmp, src, SIZE, SIZE, stride);\
|
||||
OPNAME ## pixels ## SIZE ## _l2_ ## CODETYPE(dst, halfV, halfHV, stride, SIZE, SIZE);\
|
||||
}\
|
||||
|
||||
static inline void put_pixels16_l2_altivec( uint8_t * dst, const uint8_t * src1, |
||||
const uint8_t * src2, int dst_stride, |
||||
int src_stride1, int h) |
||||
{ |
||||
int i; |
||||
vec_u8 a, b, d, tmp1, tmp2, mask, mask_, edges, align; |
||||
|
||||
mask_ = vec_lvsl(0, src2); |
||||
|
||||
for (i = 0; i < h; i++) { |
||||
|
||||
tmp1 = vec_ld(i * src_stride1, src1); |
||||
mask = vec_lvsl(i * src_stride1, src1); |
||||
tmp2 = vec_ld(i * src_stride1 + 15, src1); |
||||
|
||||
a = vec_perm(tmp1, tmp2, mask); |
||||
|
||||
tmp1 = vec_ld(i * 16, src2); |
||||
tmp2 = vec_ld(i * 16 + 15, src2); |
||||
|
||||
b = vec_perm(tmp1, tmp2, mask_); |
||||
|
||||
tmp1 = vec_ld(0, dst); |
||||
mask = vec_lvsl(0, dst); |
||||
tmp2 = vec_ld(15, dst); |
||||
|
||||
d = vec_avg(a, b); |
||||
|
||||
edges = vec_perm(tmp2, tmp1, mask); |
||||
|
||||
align = vec_lvsr(0, dst); |
||||
|
||||
tmp2 = vec_perm(d, edges, align); |
||||
tmp1 = vec_perm(edges, d, align); |
||||
|
||||
vec_st(tmp2, 15, dst); |
||||
vec_st(tmp1, 0 , dst); |
||||
|
||||
dst += dst_stride; |
||||
} |
||||
} |
||||
|
||||
static inline void avg_pixels16_l2_altivec( uint8_t * dst, const uint8_t * src1, |
||||
const uint8_t * src2, int dst_stride, |
||||
int src_stride1, int h) |
||||
{ |
||||
int i; |
||||
vec_u8 a, b, d, tmp1, tmp2, mask, mask_, edges, align; |
||||
|
||||
mask_ = vec_lvsl(0, src2); |
||||
|
||||
for (i = 0; i < h; i++) { |
||||
|
||||
tmp1 = vec_ld(i * src_stride1, src1); |
||||
mask = vec_lvsl(i * src_stride1, src1); |
||||
tmp2 = vec_ld(i * src_stride1 + 15, src1); |
||||
|
||||
a = vec_perm(tmp1, tmp2, mask); |
||||
|
||||
tmp1 = vec_ld(i * 16, src2); |
||||
tmp2 = vec_ld(i * 16 + 15, src2); |
||||
|
||||
b = vec_perm(tmp1, tmp2, mask_); |
||||
|
||||
tmp1 = vec_ld(0, dst); |
||||
mask = vec_lvsl(0, dst); |
||||
tmp2 = vec_ld(15, dst); |
||||
|
||||
d = vec_avg(vec_perm(tmp1, tmp2, mask), vec_avg(a, b)); |
||||
|
||||
edges = vec_perm(tmp2, tmp1, mask); |
||||
|
||||
align = vec_lvsr(0, dst); |
||||
|
||||
tmp2 = vec_perm(d, edges, align); |
||||
tmp1 = vec_perm(edges, d, align); |
||||
|
||||
vec_st(tmp2, 15, dst); |
||||
vec_st(tmp1, 0 , dst); |
||||
|
||||
dst += dst_stride; |
||||
} |
||||
} |
||||
|
||||
/* Implemented but could be faster
|
||||
#define put_pixels16_l2_altivec(d,s1,s2,ds,s1s,h) put_pixels16_l2(d,s1,s2,ds,s1s,16,h) |
||||
#define avg_pixels16_l2_altivec(d,s1,s2,ds,s1s,h) avg_pixels16_l2(d,s1,s2,ds,s1s,16,h) |
||||
*/ |
||||
|
||||
H264_MC(put_, 16, altivec) |
||||
H264_MC(avg_, 16, altivec) |
||||
|
||||
void ff_dsputil_h264_init_ppc(DSPContext* c, AVCodecContext *avctx) |
||||
{ |
||||
const int high_bit_depth = avctx->bits_per_raw_sample > 8; |
||||
|
||||
if (av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC) { |
||||
if (!high_bit_depth) { |
||||
c->put_h264_chroma_pixels_tab[0] = put_h264_chroma_mc8_altivec; |
||||
c->avg_h264_chroma_pixels_tab[0] = avg_h264_chroma_mc8_altivec; |
||||
} |
||||
} |
||||
} |
||||
#endif /* HAVE_ALTIVEC */ |
||||
|
||||
void ff_h264qpel_init_ppc(H264QpelContext* c, int bit_depth) { |
||||
#if HAVE_ALTIVEC |
||||
const int high_bit_depth = bit_depth > 8; |
||||
|
||||
if (av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC) { |
||||
if (!high_bit_depth) { |
||||
#define dspfunc(PFX, IDX, NUM) \ |
||||
c->PFX ## _pixels_tab[IDX][ 0] = PFX ## NUM ## _mc00_altivec; \
|
||||
c->PFX ## _pixels_tab[IDX][ 1] = PFX ## NUM ## _mc10_altivec; \
|
||||
c->PFX ## _pixels_tab[IDX][ 2] = PFX ## NUM ## _mc20_altivec; \
|
||||
c->PFX ## _pixels_tab[IDX][ 3] = PFX ## NUM ## _mc30_altivec; \
|
||||
c->PFX ## _pixels_tab[IDX][ 4] = PFX ## NUM ## _mc01_altivec; \
|
||||
c->PFX ## _pixels_tab[IDX][ 5] = PFX ## NUM ## _mc11_altivec; \
|
||||
c->PFX ## _pixels_tab[IDX][ 6] = PFX ## NUM ## _mc21_altivec; \
|
||||
c->PFX ## _pixels_tab[IDX][ 7] = PFX ## NUM ## _mc31_altivec; \
|
||||
c->PFX ## _pixels_tab[IDX][ 8] = PFX ## NUM ## _mc02_altivec; \
|
||||
c->PFX ## _pixels_tab[IDX][ 9] = PFX ## NUM ## _mc12_altivec; \
|
||||
c->PFX ## _pixels_tab[IDX][10] = PFX ## NUM ## _mc22_altivec; \
|
||||
c->PFX ## _pixels_tab[IDX][11] = PFX ## NUM ## _mc32_altivec; \
|
||||
c->PFX ## _pixels_tab[IDX][12] = PFX ## NUM ## _mc03_altivec; \
|
||||
c->PFX ## _pixels_tab[IDX][13] = PFX ## NUM ## _mc13_altivec; \
|
||||
c->PFX ## _pixels_tab[IDX][14] = PFX ## NUM ## _mc23_altivec; \
|
||||
c->PFX ## _pixels_tab[IDX][15] = PFX ## NUM ## _mc33_altivec |
||||
|
||||
dspfunc(put_h264_qpel, 0, 16); |
||||
dspfunc(avg_h264_qpel, 0, 16); |
||||
#undef dspfunc |
||||
} |
||||
} |
||||
#endif /* HAVE_ALTIVEC */ |
||||
} |
||||
Loading…
Reference in new issue