Commit e9e608ad authored by Ronald S. Bultje's avatar Ronald S. Bultje Committed by Michael Niedermayer

Move bfin half-pel assembly from dsputil to hpeldsp.

parent de99545f
......@@ -3,6 +3,8 @@ OBJS += bfin/dsputil_bfin.o \
bfin/idct_bfin.o \
bfin/pixels_bfin.o \
OBJS-$(CONFIG_HPELDS) += bfin/hpeldsp_bfin.o \
bfin/hpel_pixels_bfin.o
OBJS-$(CONFIG_MPEGVIDEOENC) += bfin/mpegvideo_bfin.o
OBJS-$(CONFIG_VP3DSP) += bfin/vp3_bfin.o \
bfin/vp3_idct_bfin.o
......@@ -54,79 +54,6 @@ static void bfin_clear_blocks (int16_t *blocks)
::"a" (blocks):"P0","I0","R0");
}
static void bfin_put_pixels8 (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels8uc (block, pixels, pixels, line_size, line_size, h);
}
static void bfin_put_pixels8_x2(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels8uc (block, pixels, pixels+1, line_size, line_size, h);
}
static void bfin_put_pixels8_y2 (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels8uc (block, pixels, pixels+line_size, line_size, line_size, h);
}
static void bfin_put_pixels8_xy2 (uint8_t *block, const uint8_t *s0, ptrdiff_t line_size, int h)
{
ff_bfin_z_put_pixels8_xy2 (block,s0,line_size, line_size, h);
}
static void bfin_put_pixels16 (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels16uc (block, pixels, pixels, line_size, line_size, h);
}
static void bfin_put_pixels16_x2 (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels16uc (block, pixels, pixels+1, line_size, line_size, h);
}
static void bfin_put_pixels16_y2 (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels16uc (block, pixels, pixels+line_size, line_size, line_size, h);
}
static void bfin_put_pixels16_xy2 (uint8_t *block, const uint8_t *s0, ptrdiff_t line_size, int h)
{
ff_bfin_z_put_pixels16_xy2 (block,s0,line_size, line_size, h);
}
static void bfin_put_pixels8_nornd (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels8uc_nornd (block, pixels, pixels, line_size, h);
}
static void bfin_put_pixels8_x2_nornd (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels8uc_nornd (block, pixels, pixels+1, line_size, h);
}
static void bfin_put_pixels8_y2_nornd (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels8uc_nornd (block, pixels, pixels+line_size, line_size, h);
}
static void bfin_put_pixels16_nornd (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels16uc_nornd (block, pixels, pixels, line_size, h);
}
static void bfin_put_pixels16_x2_nornd (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels16uc_nornd (block, pixels, pixels+1, line_size, h);
}
static void bfin_put_pixels16_y2_nornd (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels16uc_nornd (block, pixels, pixels+line_size, line_size, h);
}
static int bfin_pix_abs16 (void *c, uint8_t *blk1, uint8_t *blk2, int line_size, int h)
{
return ff_bfin_z_sad16x16 (blk1,blk2,line_size,line_size,h);
......@@ -232,28 +159,6 @@ av_cold void ff_dsputil_init_bfin(DSPContext *c, AVCodecContext *avctx)
c->sse[1] = ff_bfin_sse8;
c->sse[2] = ff_bfin_sse4;
if (!high_bit_depth) {
c->put_pixels_tab[0][0] = bfin_put_pixels16;
c->put_pixels_tab[0][1] = bfin_put_pixels16_x2;
c->put_pixels_tab[0][2] = bfin_put_pixels16_y2;
c->put_pixels_tab[0][3] = bfin_put_pixels16_xy2;
c->put_pixels_tab[1][0] = bfin_put_pixels8;
c->put_pixels_tab[1][1] = bfin_put_pixels8_x2;
c->put_pixels_tab[1][2] = bfin_put_pixels8_y2;
c->put_pixels_tab[1][3] = bfin_put_pixels8_xy2;
c->put_no_rnd_pixels_tab[1][0] = bfin_put_pixels8_nornd;
c->put_no_rnd_pixels_tab[1][1] = bfin_put_pixels8_x2_nornd;
c->put_no_rnd_pixels_tab[1][2] = bfin_put_pixels8_y2_nornd;
/* c->put_no_rnd_pixels_tab[1][3] = ff_bfin_put_pixels8_xy2_nornd; */
c->put_no_rnd_pixels_tab[0][0] = bfin_put_pixels16_nornd;
c->put_no_rnd_pixels_tab[0][1] = bfin_put_pixels16_x2_nornd;
c->put_no_rnd_pixels_tab[0][2] = bfin_put_pixels16_y2_nornd;
/* c->put_no_rnd_pixels_tab[0][3] = ff_bfin_put_pixels16_xy2_nornd; */
}
if (avctx->bits_per_raw_sample <= 8) {
if (avctx->dct_algo == FF_DCT_AUTO)
c->fdct = ff_bfin_fdct;
......
......@@ -46,19 +46,8 @@ int ff_bfin_pix_norm1 (uint8_t * pix, int line_size) attribute_l1_text;
int ff_bfin_z_sad8x8 (uint8_t *blk1, uint8_t *blk2, int dsz, int line_size, int h) attribute_l1_text;
int ff_bfin_z_sad16x16 (uint8_t *blk1, uint8_t *blk2, int dsz, int line_size, int h) attribute_l1_text;
void ff_bfin_z_put_pixels16_xy2 (uint8_t *block, const uint8_t *s0, int dest_size, int line_size, int h) attribute_l1_text;
void ff_bfin_z_put_pixels8_xy2 (uint8_t *block, const uint8_t *s0, int dest_size, int line_size, int h) attribute_l1_text;
void ff_bfin_put_pixels16_xy2_nornd (uint8_t *block, const uint8_t *s0, int line_size, int h) attribute_l1_text;
void ff_bfin_put_pixels8_xy2_nornd (uint8_t *block, const uint8_t *s0, int line_size, int h) attribute_l1_text;
int ff_bfin_pix_sum (uint8_t *p, int stride) attribute_l1_text;
void ff_bfin_put_pixels8uc (uint8_t *block, const uint8_t *s0, const uint8_t *s1, int dest_size, int line_size, int h) attribute_l1_text;
void ff_bfin_put_pixels16uc (uint8_t *block, const uint8_t *s0, const uint8_t *s1, int dest_size, int line_size, int h) attribute_l1_text;
void ff_bfin_put_pixels8uc_nornd (uint8_t *block, const uint8_t *s0, const uint8_t *s1, int line_size, int h) attribute_l1_text;
void ff_bfin_put_pixels16uc_nornd (uint8_t *block, const uint8_t *s0, const uint8_t *s1, int line_size, int h) attribute_l1_text;
int ff_bfin_sse4 (void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h) attribute_l1_text;
int ff_bfin_sse8 (void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h) attribute_l1_text;
int ff_bfin_sse16 (void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h) attribute_l1_text;
......
/*
* Blackfin Pixel Operations
* Copyright (C) 2007 Marc Hoffman <marc.hoffman@analog.com>
*
* This file is part of FFmpeg.
*
* FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "config_bfin.h"
/**
motion compensation
primitives
* Halfpel motion compensation with rounding (a+b+1)>>1.
* This is an array[4][4] of motion compensation funcions for 4
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
* @param block destination where the result is stored
* @param pixels source
* @param line_size number of bytes in a horizontal line of block
* @param h height
*/
DEFUN(put_pixels8uc,mL1,
(uint8_t *block, const uint8_t *s0, const uint8_t *s1,
int dest_size, int line_size, int h)):
i3=r0; // dest
i0=r1; // src0
i1=r2; // src1
r0=[sp+12]; // dest_size
r2=[sp+16]; // line_size
p0=[sp+20]; // h
[--sp] = (r7:6);
r0+=-4;
m3=r0;
r2+=-8;
m0=r2;
LSETUP(pp8$0,pp8$1) LC0=P0;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
pp8$0: DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R6 = BYTEOP1P(R1:0,R3:2) || R0 = [I0++M0]|| R2 =[I1++M0];
R7 = BYTEOP1P(R1:0,R3:2)(R) || R0 = [I0++] || [I3++] = R6 ;
pp8$1: DISALGNEXCPT || R2 = [I1++] || [I3++M3] = R7;
(r7:6) = [sp++];
RTS;
DEFUN_END(put_pixels8uc)
DEFUN(put_pixels16uc,mL1,
(uint8_t *block, const uint8_t *s0, const uint8_t *s1,
int dest_size, int line_size, int h)):
link 0;
[--sp] = (r7:6);
i3=r0; // dest
i0=r1; // src0
i1=r2; // src1
r0=[fp+20]; // dest_size
r2=[fp+24]; // line_size
p0=[fp+28]; // h
r0+=-12;
m3=r0; // line_size
r2+=-16;
m0=r2;
LSETUP(pp16$0,pp16$1) LC0=P0;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
pp16$0: DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R6 = BYTEOP1P(R1:0,R3:2) || R0 = [I0++] || R2 =[I1++];
R7 = BYTEOP1P(R1:0,R3:2)(R) || R1 = [I0++] || R3 =[I1++];
[I3++] = R6;
R6 = BYTEOP1P(R1:0,R3:2) || R0 = [I0++M0] || R2 =[I1++M0];
R7 = BYTEOP1P(R1:0,R3:2)(R) || R0 = [I0++] || [I3++] = R7 ;
[I3++] = R6;
pp16$1: DISALGNEXCPT || R2 = [I1++] || [I3++M3] = R7;
(r7:6) = [sp++];
unlink;
RTS;
DEFUN_END(put_pixels16uc)
DEFUN(put_pixels8uc_nornd,mL1,
(uint8_t *block, const uint8_t *s0, const uint8_t *s1,
int line_size, int h)):
i3=r0; // dest
i0=r1; // src0
i1=r2; // src1
r2=[sp+12]; // line_size
p0=[sp+16]; // h
[--sp] = (r7:6);
r2+=-4;
m3=r2;
r2+=-4;
m0=r2;
LSETUP(pp8$2,pp8$3) LC0=P0;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
pp8$2: DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R6 = BYTEOP1P(R1:0,R3:2)(T) || R0 = [I0++M0]|| R2 =[I1++M0];
R7 = BYTEOP1P(R1:0,R3:2)(T,R) || R0 = [I0++] || [I3++] = R6 ;
pp8$3: DISALGNEXCPT || R2 = [I1++] || [I3++M3] = R7;
(r7:6) = [sp++];
RTS;
DEFUN_END(put_pixels8uc_nornd)
DEFUN(put_pixels16uc_nornd,mL1,
(uint8_t *block, const uint8_t *s0, const uint8_t *s1,
int line_size, int h)):
i3=r0; // dest
i0=r1; // src0
i1=r2; // src1
r2=[sp+12]; // line_size
p0=[sp+16]; // h
[--sp] = (r7:6);
r2+=-12;
m3=r2; // line_size
r2+=-4;
m0=r2;
LSETUP(pp16$2,pp16$3) LC0=P0;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
pp16$2:
DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R6 = BYTEOP1P(R1:0,R3:2)(T) || R0 = [I0++] || R2 =[I1++];
R7 = BYTEOP1P(R1:0,R3:2)(T,R) || R1 = [I0++] || R3 =[I1++];
[I3++] = R6;
R6 = BYTEOP1P(R1:0,R3:2)(T) || R0 = [I0++M0] || R2 =[I1++M0];
R7 = BYTEOP1P(R1:0,R3:2)(T,R) || R0 = [I0++] || [I3++] = R7 ;
[I3++] = R6;
pp16$3: DISALGNEXCPT || R2 = [I1++] || [I3++M3] = R7;
(r7:6) = [sp++];
RTS;
DEFUN_END(put_pixels16uc_nornd)
DEFUN(z_put_pixels16_xy2,mL1,
(uint8_t *block, const uint8_t *s0,
int dest_size, int line_size, int h)):
link 0;
[--sp] = (r7:4);
i3=r0; // dest
i0=r1; // src0--> pixels
i1=r1; // src1--> pixels + line_size
r2+=-12;
m2=r2; // m2=dest_width-4
r2=[fp+20];
m3=r2; // line_size
p0=[fp+24]; // h
r2+=-16;
i1+=m3; /* src1 + line_size */
m0=r2; /* line-size - 20 */
B0 = I0;
B1 = I1;
B3 = I3;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
LSETUP(LS$16E,LE$16E) LC0=P0;
LS$16E: DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R4 = BYTEOP2P (R3:2,R1:0) (RNDL) || R0 = [I0++] || R2 =[I1++];
R5 = BYTEOP2P (R3:2,R1:0) (RNDL,R) || R1 = [I0++] || [I3++] = R4 ;
DISALGNEXCPT || R3 = [I1++] || [I3++] = R5;
R4 = BYTEOP2P (R3:2,R1:0) (RNDL) || R0 = [I0++M0]|| R2 = [I1++M0];
R5 = BYTEOP2P (R3:2,R1:0) (RNDL,R) || R0 = [I0++] || [I3++] = R4 ;
LE$16E: DISALGNEXCPT || R2 = [I1++] || [I3++M2] = R5;
M1 = 1;
I3 = B3;
I1 = B1;
I0 = B0;
I0 += M1;
I1 += M1;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
LSETUP(LS$16O,LE$16O) LC0=P0;
LS$16O: DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R4 = BYTEOP2P (R3:2,R1:0) (RNDH) || R0 = [I0++] || R2 =[I1++];
R5 = BYTEOP2P (R3:2,R1:0) (RNDH,R) || R1 = [I0++] || R6 =[I3++];
R4 = R4 +|+ R6 || R7 = [I3--];
R5 = R5 +|+ R7 || [I3++] = R4;
DISALGNEXCPT || R3 =[I1++] || [I3++] = R5;
R4 = BYTEOP2P (R3:2,R1:0) (RNDH) || R0 = [I0++M0]|| R2 = [I1++M0];
R5 = BYTEOP2P (R3:2,R1:0) (RNDH,R) || R0 = [I0++] || R6 = [I3++];
R4 = R4 +|+ R6 || R7 = [I3--];
R5 = R5 +|+ R7 || [I3++] = R4;
LE$16O: DISALGNEXCPT || R2 = [I1++] || [I3++M2] = R5;
(r7:4) = [sp++];
unlink;
rts;
DEFUN_END(z_put_pixels16_xy2)
DEFUN(put_pixels16_xy2_nornd,mL1,
(uint8_t *block, const uint8_t *s0,
int line_size, int h)):
link 0;
[--sp] = (r7:4);
i3=r0; // dest
i0=r1; // src0--> pixels
i1=r1; // src1--> pixels + line_size
m3=r2;
r2+=-12;
m2=r2;
r2+=-4;
i1+=m3; /* src1 + line_size */
m0=r2; /* line-size - 20 */
p0=[fp+20]; // h
B0=I0;
B1=I1;
B3=I3;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
LSETUP(LS$16ET,LE$16ET) LC0=P0;
LS$16ET:DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R4 = BYTEOP2P (R3:2,R1:0) (TL) || R0 = [I0++] || R2 =[I1++];
R5 = BYTEOP2P (R3:2,R1:0) (TL,R) || R1 = [I0++] || [I3++] = R4 ;
DISALGNEXCPT || R3 = [I1++] || [I3++] = R5;
R4 = BYTEOP2P (R3:2,R1:0) (TL) || R0 = [I0++M0]|| R2 = [I1++M0];
R5 = BYTEOP2P (R3:2,R1:0) (TL,R) || R0 = [I0++] || [I3++] = R4 ;
LE$16ET:DISALGNEXCPT || R2 = [I1++] || [I3++M2] = R5;
M1 = 1;
I3=B3;
I1=B1;
I0=B0;
I0 += M1;
I1 += M1;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
LSETUP(LS$16OT,LE$16OT) LC0=P0;
LS$16OT:DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R4 = BYTEOP2P (R3:2,R1:0) (TH) || R0 = [I0++] || R2 =[I1++];
R5 = BYTEOP2P (R3:2,R1:0) (TH,R) || R1 = [I0++] || R6 =[I3++];
R4 = R4 +|+ R6 || R7 = [I3--];
R5 = R5 +|+ R7 || [I3++] = R4;
DISALGNEXCPT || R3 =[I1++] || [I3++] = R5;
R4 = BYTEOP2P (R3:2,R1:0) (TH) || R0 = [I0++M0]|| R2 = [I1++M0];
R5 = BYTEOP2P (R3:2,R1:0) (TH,R) || R0 = [I0++] || R6 = [I3++];
R4 = R4 +|+ R6 || R7 = [I3--];
R5 = R5 +|+ R7 || [I3++] = R4;
LE$16OT:DISALGNEXCPT || R2 = [I1++] || [I3++M2] = R5;
(r7:4) = [sp++];
unlink;
rts;
DEFUN_END(put_pixels16_xy2_nornd)
DEFUN(z_put_pixels8_xy2,mL1,
(uint8_t *block, const uint8_t *s0,
int dest_size, int line_size, int h)):
link 0;
[--sp] = (r7:4);
i3=r0; // dest
i0=r1; // src0--> pixels
i1=r1; // src1--> pixels + line_size
r2+=-4;
m2=r2; // m2=dest_width-4
r2=[fp+20];
m3=r2; // line_size
p0=[fp+24]; // h
r2+=-8;
i1+=m3; /* src1 + line_size */
m0=r2; /* line-size - 20 */
b0 = I0;
b1 = I1;
b3 = I3;
LSETUP(LS$8E,LE$8E) LC0=P0;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
LS$8E: DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R4 = BYTEOP2P (R3:2,R1:0) (RNDL) || R0 = [I0++M0] || R2 =[I1++M0];
R5 = BYTEOP2P (R3:2,R1:0) (RNDL,R) || R0 = [I0++] || [I3++] = R4 ;
LE$8E: DISALGNEXCPT || R2 = [I1++] || [I3++M2] = R5;
M1 = 1;
I3 = b3;
I1 = b1;
I0 = b0;
I0 += M1;
I1 += M1;
LSETUP(LS$8O,LE$8O) LC0=P0;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
LS$8O: DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R4 = BYTEOP2P (R3:2,R1:0) (RNDH) || R0 = [I0++M0] || R2 =[I1++M0];
R5 = BYTEOP2P (R3:2,R1:0) (RNDH,R) || R0 = [I0++] || R6 =[I3++];
R4 = R4 +|+ R6 || R7 = [I3--];
R5 = R5 +|+ R7 || [I3++] = R4;
LE$8O: DISALGNEXCPT || R2 =[I1++] || [I3++M2] = R5;
(r7:4) = [sp++];
unlink;
rts;
DEFUN_END(z_put_pixels8_xy2)
DEFUN(put_pixels8_xy2_nornd,mL1,
(uint8_t *block, const uint8_t *s0, int line_size, int h)):
link 0;
[--sp] = (r7:4);
i3=r0; // dest
i0=r1; // src0--> pixels
i1=r1; // src1--> pixels + line_size
m3=r2;
r2+=-4;
m2=r2;
r2+=-4;
i1+=m3; /* src1 + line_size */
m0=r2; /* line-size - 20 */
p0=[fp+20]; // h
b0 = I0;
b1 = I1;
b3 = I3;
LSETUP(LS$8ET,LE$8ET) LC0=P0;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
LS$8ET: DISALGNEXCPT || R1 = [I0++] || R3 = [I1++];
R4 = BYTEOP2P (R3:2,R1:0) (TL) || R0 = [I0++M0] || R2 = [I1++M0];
R5 = BYTEOP2P (R3:2,R1:0) (TL,R) || R0 = [I0++] || [I3++] = R4 ;
LE$8ET: DISALGNEXCPT || R2 = [I1++] || [I3++M2] = R5;
M1 = 1;
I3 = b3;
I1 = b1;
I0 = b0;
I0 += M1;
I1 += M1;
LSETUP(LS$8OT,LE$8OT) LC0=P0;
DISALGNEXCPT || R0 = [I0++] || R2 = [I1++];
LS$8OT: DISALGNEXCPT || R1 = [I0++] || R3 = [I1++];
R4 = BYTEOP2P (R3:2,R1:0) (TH) || R0 = [I0++M0] || R2 = [I1++M0];
R5 = BYTEOP2P (R3:2,R1:0) (TH,R) || R0 = [I0++] || R6 = [I3++];
R4 = R4 +|+ R6 || R7 = [I3--];
R5 = R5 +|+ R7 || [I3++] = R4;
LE$8OT: DISALGNEXCPT || R2 =[I1++] || [I3++M2] = R5;
(r7:4) = [sp++];
unlink;
rts;
DEFUN_END(put_pixels8_xy2_nornd)
/*
* BlackFin DSPUTILS
*
* Copyright (C) 2007 Marc Hoffman <marc.hoffman@analog.com>
* Copyright (c) 2006 Michael Benjamin <michael.benjamin@analog.com>
*
* This file is part of FFmpeg.
*
* FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "libavcodec/hpeldsp.h"
#include "hpeldsp_bfin.h"
static void bfin_put_pixels8 (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels8uc (block, pixels, pixels, line_size, line_size, h);
}
static void bfin_put_pixels8_x2(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels8uc (block, pixels, pixels+1, line_size, line_size, h);
}
static void bfin_put_pixels8_y2 (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels8uc (block, pixels, pixels+line_size, line_size, line_size, h);
}
static void bfin_put_pixels8_xy2 (uint8_t *block, const uint8_t *s0, ptrdiff_t line_size, int h)
{
ff_bfin_z_put_pixels8_xy2 (block,s0,line_size, line_size, h);
}
static void bfin_put_pixels16 (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels16uc (block, pixels, pixels, line_size, line_size, h);
}
static void bfin_put_pixels16_x2 (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels16uc (block, pixels, pixels+1, line_size, line_size, h);
}
static void bfin_put_pixels16_y2 (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels16uc (block, pixels, pixels+line_size, line_size, line_size, h);
}
static void bfin_put_pixels16_xy2 (uint8_t *block, const uint8_t *s0, ptrdiff_t line_size, int h)
{
ff_bfin_z_put_pixels16_xy2 (block,s0,line_size, line_size, h);
}
static void bfin_put_pixels8_nornd (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels8uc_nornd (block, pixels, pixels, line_size, h);
}
static void bfin_put_pixels8_x2_nornd (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels8uc_nornd (block, pixels, pixels+1, line_size, h);
}
static void bfin_put_pixels8_y2_nornd (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels8uc_nornd (block, pixels, pixels+line_size, line_size, h);
}
static void bfin_put_pixels16_nornd (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels16uc_nornd (block, pixels, pixels, line_size, h);
}
static void bfin_put_pixels16_x2_nornd (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels16uc_nornd (block, pixels, pixels+1, line_size, h);
}
static void bfin_put_pixels16_y2_nornd (uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)
{
ff_bfin_put_pixels16uc_nornd (block, pixels, pixels+line_size, line_size, h);
}
void ff_hpeldsp_init_bfin(HpelDSPContext* c, int flags)
{
c->put_pixels_tab[0][0] = bfin_put_pixels16;
c->put_pixels_tab[0][1] = bfin_put_pixels16_x2;
c->put_pixels_tab[0][2] = bfin_put_pixels16_y2;
c->put_pixels_tab[0][3] = bfin_put_pixels16_xy2;
c->put_pixels_tab[1][0] = bfin_put_pixels8;
c->put_pixels_tab[1][1] = bfin_put_pixels8_x2;
c->put_pixels_tab[1][2] = bfin_put_pixels8_y2;
c->put_pixels_tab[1][3] = bfin_put_pixels8_xy2;
c->put_no_rnd_pixels_tab[1][0] = bfin_put_pixels8_nornd;
c->put_no_rnd_pixels_tab[1][1] = bfin_put_pixels8_x2_nornd;
c->put_no_rnd_pixels_tab[1][2] = bfin_put_pixels8_y2_nornd;
/* c->put_no_rnd_pixels_tab[1][3] = ff_bfin_put_pixels8_xy2_nornd; */
c->put_no_rnd_pixels_tab[0][0] = bfin_put_pixels16_nornd;
c->put_no_rnd_pixels_tab[0][1] = bfin_put_pixels16_x2_nornd;
c->put_no_rnd_pixels_tab[0][2] = bfin_put_pixels16_y2_nornd;
/* c->put_no_rnd_pixels_tab[0][3] = ff_bfin_put_pixels16_xy2_nornd; */
}
/*
* BlackFin DSPUTILS COMMON OPTIMIZATIONS HEADER
*
* Copyright (C) 2007 Marc Hoffman <mmh@pleasantst.com>
*
* This file is part of FFmpeg.
*
* FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#ifndef AVCODEC_BFIN_HPELDSP_BFIN_H
#define AVCODEC_BFIN_HPELDSP_BFIN_H
#include <stdint.h>
#include "config.h"
#if defined(__FDPIC__) && CONFIG_SRAM
#define attribute_l1_text __attribute__ ((l1_text))
#define attribute_l1_data_b __attribute__((l1_data_B))
#else
#define attribute_l1_text
#define attribute_l1_data_b
#endif
void ff_bfin_z_put_pixels16_xy2 (uint8_t *block, const uint8_t *s0, int dest_size, int line_size, int h) attribute_l1_text;
void ff_bfin_z_put_pixels8_xy2 (uint8_t *block, const uint8_t *s0, int dest_size, int line_size, int h) attribute_l1_text;
void ff_bfin_put_pixels16_xy2_nornd (uint8_t *block, const uint8_t *s0, int line_size, int h) attribute_l1_text;
void ff_bfin_put_pixels8_xy2_nornd (uint8_t *block, const uint8_t *s0, int line_size, int h) attribute_l1_text;
void ff_bfin_put_pixels8uc (uint8_t *block, const uint8_t *s0, const uint8_t *s1, int dest_size, int line_size, int h) attribute_l1_text;
void ff_bfin_put_pixels16uc (uint8_t *block, const uint8_t *s0, const uint8_t *s1, int dest_size, int line_size, int h) attribute_l1_text;
void ff_bfin_put_pixels8uc_nornd (uint8_t *block, const uint8_t *s0, const uint8_t *s1, int line_size, int h) attribute_l1_text;
void ff_bfin_put_pixels16uc_nornd (uint8_t *block, const uint8_t *s0, const uint8_t *s1, int line_size, int h) attribute_l1_text;
#endif /* AVCODEC_BFIN_HPELDSP_BFIN_H */
......@@ -83,364 +83,6 @@ apc$3: R6 = BYTEOP3P(R1:0, R3:2) (LO) || [I2++M0] = R6 || R2 = [I1];
RTS;
DEFUN_END(add_pixels_clamped)
/**
motion compensation
primitives
* Halfpel motion compensation with rounding (a+b+1)>>1.
* This is an array[4][4] of motion compensation funcions for 4
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
* @param block destination where the result is stored
* @param pixels source
* @param line_size number of bytes in a horizontal line of block
* @param h height
*/
DEFUN(put_pixels8uc,mL1,
(uint8_t *block, const uint8_t *s0, const uint8_t *s1,
int dest_size, int line_size, int h)):
i3=r0; // dest
i0=r1; // src0
i1=r2; // src1
r0=[sp+12]; // dest_size
r2=[sp+16]; // line_size
p0=[sp+20]; // h
[--sp] = (r7:6);
r0+=-4;
m3=r0;
r2+=-8;
m0=r2;
LSETUP(pp8$0,pp8$1) LC0=P0;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
pp8$0: DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R6 = BYTEOP1P(R1:0,R3:2) || R0 = [I0++M0]|| R2 =[I1++M0];
R7 = BYTEOP1P(R1:0,R3:2)(R) || R0 = [I0++] || [I3++] = R6 ;
pp8$1: DISALGNEXCPT || R2 = [I1++] || [I3++M3] = R7;
(r7:6) = [sp++];
RTS;
DEFUN_END(put_pixels8uc)
DEFUN(put_pixels16uc,mL1,
(uint8_t *block, const uint8_t *s0, const uint8_t *s1,
int dest_size, int line_size, int h)):
link 0;
[--sp] = (r7:6);
i3=r0; // dest
i0=r1; // src0
i1=r2; // src1
r0=[fp+20]; // dest_size
r2=[fp+24]; // line_size
p0=[fp+28]; // h
r0+=-12;
m3=r0; // line_size
r2+=-16;
m0=r2;
LSETUP(pp16$0,pp16$1) LC0=P0;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
pp16$0: DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R6 = BYTEOP1P(R1:0,R3:2) || R0 = [I0++] || R2 =[I1++];
R7 = BYTEOP1P(R1:0,R3:2)(R) || R1 = [I0++] || R3 =[I1++];
[I3++] = R6;
R6 = BYTEOP1P(R1:0,R3:2) || R0 = [I0++M0] || R2 =[I1++M0];
R7 = BYTEOP1P(R1:0,R3:2)(R) || R0 = [I0++] || [I3++] = R7 ;
[I3++] = R6;
pp16$1: DISALGNEXCPT || R2 = [I1++] || [I3++M3] = R7;
(r7:6) = [sp++];
unlink;
RTS;
DEFUN_END(put_pixels16uc)
DEFUN(put_pixels8uc_nornd,mL1,
(uint8_t *block, const uint8_t *s0, const uint8_t *s1,
int line_size, int h)):
i3=r0; // dest
i0=r1; // src0
i1=r2; // src1
r2=[sp+12]; // line_size
p0=[sp+16]; // h
[--sp] = (r7:6);
r2+=-4;
m3=r2;
r2+=-4;
m0=r2;
LSETUP(pp8$2,pp8$3) LC0=P0;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
pp8$2: DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R6 = BYTEOP1P(R1:0,R3:2)(T) || R0 = [I0++M0]|| R2 =[I1++M0];
R7 = BYTEOP1P(R1:0,R3:2)(T,R) || R0 = [I0++] || [I3++] = R6 ;
pp8$3: DISALGNEXCPT || R2 = [I1++] || [I3++M3] = R7;
(r7:6) = [sp++];
RTS;
DEFUN_END(put_pixels8uc_nornd)
DEFUN(put_pixels16uc_nornd,mL1,
(uint8_t *block, const uint8_t *s0, const uint8_t *s1,
int line_size, int h)):
i3=r0; // dest
i0=r1; // src0
i1=r2; // src1
r2=[sp+12]; // line_size
p0=[sp+16]; // h
[--sp] = (r7:6);
r2+=-12;
m3=r2; // line_size
r2+=-4;
m0=r2;
LSETUP(pp16$2,pp16$3) LC0=P0;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
pp16$2:
DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R6 = BYTEOP1P(R1:0,R3:2)(T) || R0 = [I0++] || R2 =[I1++];
R7 = BYTEOP1P(R1:0,R3:2)(T,R) || R1 = [I0++] || R3 =[I1++];
[I3++] = R6;
R6 = BYTEOP1P(R1:0,R3:2)(T) || R0 = [I0++M0] || R2 =[I1++M0];
R7 = BYTEOP1P(R1:0,R3:2)(T,R) || R0 = [I0++] || [I3++] = R7 ;
[I3++] = R6;
pp16$3: DISALGNEXCPT || R2 = [I1++] || [I3++M3] = R7;
(r7:6) = [sp++];
RTS;
DEFUN_END(put_pixels16uc_nornd)
DEFUN(z_put_pixels16_xy2,mL1,
(uint8_t *block, const uint8_t *s0,
int dest_size, int line_size, int h)):
link 0;
[--sp] = (r7:4);
i3=r0; // dest
i0=r1; // src0--> pixels
i1=r1; // src1--> pixels + line_size
r2+=-12;
m2=r2; // m2=dest_width-4
r2=[fp+20];
m3=r2; // line_size
p0=[fp+24]; // h
r2+=-16;
i1+=m3; /* src1 + line_size */
m0=r2; /* line-size - 20 */
B0 = I0;
B1 = I1;
B3 = I3;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
LSETUP(LS$16E,LE$16E) LC0=P0;
LS$16E: DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R4 = BYTEOP2P (R3:2,R1:0) (RNDL) || R0 = [I0++] || R2 =[I1++];
R5 = BYTEOP2P (R3:2,R1:0) (RNDL,R) || R1 = [I0++] || [I3++] = R4 ;
DISALGNEXCPT || R3 = [I1++] || [I3++] = R5;
R4 = BYTEOP2P (R3:2,R1:0) (RNDL) || R0 = [I0++M0]|| R2 = [I1++M0];
R5 = BYTEOP2P (R3:2,R1:0) (RNDL,R) || R0 = [I0++] || [I3++] = R4 ;
LE$16E: DISALGNEXCPT || R2 = [I1++] || [I3++M2] = R5;
M1 = 1;
I3 = B3;
I1 = B1;
I0 = B0;
I0 += M1;
I1 += M1;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
LSETUP(LS$16O,LE$16O) LC0=P0;
LS$16O: DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R4 = BYTEOP2P (R3:2,R1:0) (RNDH) || R0 = [I0++] || R2 =[I1++];
R5 = BYTEOP2P (R3:2,R1:0) (RNDH,R) || R1 = [I0++] || R6 =[I3++];
R4 = R4 +|+ R6 || R7 = [I3--];
R5 = R5 +|+ R7 || [I3++] = R4;
DISALGNEXCPT || R3 =[I1++] || [I3++] = R5;
R4 = BYTEOP2P (R3:2,R1:0) (RNDH) || R0 = [I0++M0]|| R2 = [I1++M0];
R5 = BYTEOP2P (R3:2,R1:0) (RNDH,R) || R0 = [I0++] || R6 = [I3++];
R4 = R4 +|+ R6 || R7 = [I3--];
R5 = R5 +|+ R7 || [I3++] = R4;
LE$16O: DISALGNEXCPT || R2 = [I1++] || [I3++M2] = R5;
(r7:4) = [sp++];
unlink;
rts;
DEFUN_END(z_put_pixels16_xy2)
DEFUN(put_pixels16_xy2_nornd,mL1,
(uint8_t *block, const uint8_t *s0,
int line_size, int h)):
link 0;
[--sp] = (r7:4);
i3=r0; // dest
i0=r1; // src0--> pixels
i1=r1; // src1--> pixels + line_size
m3=r2;
r2+=-12;
m2=r2;
r2+=-4;
i1+=m3; /* src1 + line_size */
m0=r2; /* line-size - 20 */
p0=[fp+20]; // h
B0=I0;
B1=I1;
B3=I3;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
LSETUP(LS$16ET,LE$16ET) LC0=P0;
LS$16ET:DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R4 = BYTEOP2P (R3:2,R1:0) (TL) || R0 = [I0++] || R2 =[I1++];
R5 = BYTEOP2P (R3:2,R1:0) (TL,R) || R1 = [I0++] || [I3++] = R4 ;
DISALGNEXCPT || R3 = [I1++] || [I3++] = R5;
R4 = BYTEOP2P (R3:2,R1:0) (TL) || R0 = [I0++M0]|| R2 = [I1++M0];
R5 = BYTEOP2P (R3:2,R1:0) (TL,R) || R0 = [I0++] || [I3++] = R4 ;
LE$16ET:DISALGNEXCPT || R2 = [I1++] || [I3++M2] = R5;
M1 = 1;
I3=B3;
I1=B1;
I0=B0;
I0 += M1;
I1 += M1;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
LSETUP(LS$16OT,LE$16OT) LC0=P0;
LS$16OT:DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R4 = BYTEOP2P (R3:2,R1:0) (TH) || R0 = [I0++] || R2 =[I1++];
R5 = BYTEOP2P (R3:2,R1:0) (TH,R) || R1 = [I0++] || R6 =[I3++];
R4 = R4 +|+ R6 || R7 = [I3--];
R5 = R5 +|+ R7 || [I3++] = R4;
DISALGNEXCPT || R3 =[I1++] || [I3++] = R5;
R4 = BYTEOP2P (R3:2,R1:0) (TH) || R0 = [I0++M0]|| R2 = [I1++M0];
R5 = BYTEOP2P (R3:2,R1:0) (TH,R) || R0 = [I0++] || R6 = [I3++];
R4 = R4 +|+ R6 || R7 = [I3--];
R5 = R5 +|+ R7 || [I3++] = R4;
LE$16OT:DISALGNEXCPT || R2 = [I1++] || [I3++M2] = R5;
(r7:4) = [sp++];
unlink;
rts;
DEFUN_END(put_pixels16_xy2_nornd)
DEFUN(z_put_pixels8_xy2,mL1,
(uint8_t *block, const uint8_t *s0,
int dest_size, int line_size, int h)):
link 0;
[--sp] = (r7:4);
i3=r0; // dest
i0=r1; // src0--> pixels
i1=r1; // src1--> pixels + line_size
r2+=-4;
m2=r2; // m2=dest_width-4
r2=[fp+20];
m3=r2; // line_size
p0=[fp+24]; // h
r2+=-8;
i1+=m3; /* src1 + line_size */
m0=r2; /* line-size - 20 */
b0 = I0;
b1 = I1;
b3 = I3;
LSETUP(LS$8E,LE$8E) LC0=P0;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
LS$8E: DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R4 = BYTEOP2P (R3:2,R1:0) (RNDL) || R0 = [I0++M0] || R2 =[I1++M0];
R5 = BYTEOP2P (R3:2,R1:0) (RNDL,R) || R0 = [I0++] || [I3++] = R4 ;
LE$8E: DISALGNEXCPT || R2 = [I1++] || [I3++M2] = R5;
M1 = 1;
I3 = b3;
I1 = b1;
I0 = b0;
I0 += M1;
I1 += M1;
LSETUP(LS$8O,LE$8O) LC0=P0;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
LS$8O: DISALGNEXCPT || R1 = [I0++] || R3 =[I1++];
R4 = BYTEOP2P (R3:2,R1:0) (RNDH) || R0 = [I0++M0] || R2 =[I1++M0];
R5 = BYTEOP2P (R3:2,R1:0) (RNDH,R) || R0 = [I0++] || R6 =[I3++];
R4 = R4 +|+ R6 || R7 = [I3--];
R5 = R5 +|+ R7 || [I3++] = R4;
LE$8O: DISALGNEXCPT || R2 =[I1++] || [I3++M2] = R5;
(r7:4) = [sp++];
unlink;
rts;
DEFUN_END(z_put_pixels8_xy2)
DEFUN(put_pixels8_xy2_nornd,mL1,
(uint8_t *block, const uint8_t *s0, int line_size, int h)):
link 0;
[--sp] = (r7:4);
i3=r0; // dest
i0=r1; // src0--> pixels
i1=r1; // src1--> pixels + line_size
m3=r2;
r2+=-4;
m2=r2;
r2+=-4;
i1+=m3; /* src1 + line_size */
m0=r2; /* line-size - 20 */
p0=[fp+20]; // h
b0 = I0;
b1 = I1;
b3 = I3;
LSETUP(LS$8ET,LE$8ET) LC0=P0;
DISALGNEXCPT || R0 = [I0++] || R2 =[I1++];
LS$8ET: DISALGNEXCPT || R1 = [I0++] || R3 = [I1++];
R4 = BYTEOP2P (R3:2,R1:0) (TL) || R0 = [I0++M0] || R2 = [I1++M0];
R5 = BYTEOP2P (R3:2,R1:0) (TL,R) || R0 = [I0++] || [I3++] = R4 ;
LE$8ET: DISALGNEXCPT || R2 = [I1++] || [I3++M2] = R5;
M1 = 1;
I3 = b3;
I1 = b1;
I0 = b0;
I0 += M1;
I1 += M1;
LSETUP(LS$8OT,LE$8OT) LC0=P0;
DISALGNEXCPT || R0 = [I0++] || R2 = [I1++];
LS$8OT: DISALGNEXCPT || R1 = [I0++] || R3 = [I1++];
R4 = BYTEOP2P (R3:2,R1:0) (TH) || R0 = [I0++M0] || R2 = [I1++M0];
R5 = BYTEOP2P (R3:2,R1:0) (TH,R) || R0 = [I0++] || R6 = [I3++];
R4 = R4 +|+ R6 || R7 = [I3--];
R5 = R5 +|+ R7 || [I3++] = R4;
LE$8OT: DISALGNEXCPT || R2 =[I1++] || [I3++M2] = R5;
(r7:4) = [sp++];
unlink;
rts;
DEFUN(diff_pixels,mL1,
(int16_t *block, uint8_t *s1, uint8_t *s2, int stride)):
link 0;
......@@ -467,7 +109,7 @@ DEFUN(diff_pixels,mL1,
(r7:4) = [sp++];
unlink;
rts;
DEFUN_END(put_pixels8_xy2_nornd)
DEFUN_END(diff_pixels)
/*
for (i = 0; i < 16; i++) {
......
......@@ -62,6 +62,6 @@ av_cold void ff_hpeldsp_init(HpelDSPContext* c, int flags)
if (ARCH_PPC) ff_hpeldsp_init_ppc (c, flags);
#if 0
if (ARCH_SH4) ff_hpeldsp_init_sh4 (c, flags);
if (ARCH_BFIN) ff_hpeldsp_init_bfin (c, flags);
#endif
if (ARCH_BFIN) ff_hpeldsp_init_bfin (c, flags);
}
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment