view armv4l/dsputil_arm.c @ 3198:6b9f0c4fbdbe libavcodec

First part of a series of speed-enchancing patches. This one sets up a snow.h and makes snow use the dsputil function pointer framework to access the three functions that will be implemented in asm in the other parts of the patchset. Patch by Robert Edele < yartrebo AH earthlink POIS net> Original thread: Subject: [Ffmpeg-devel] [PATCH] Snow mmx+sse2 asm optimizations Date: Sun, 05 Feb 2006 12:47:14 -0500

/*
 * ARMv4L optimized DSP utils
 * Copyright (c) 2001 Lionel Ulmer.
 *
 * This library 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 of the License, or (at your option) any later version.
 *
 * This library 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 this library; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include "../dsputil.h"
#ifdef HAVE_IPP
#include "ipp.h"
#endif

extern void dsputil_init_iwmmxt(DSPContext* c, AVCodecContext *avctx);

extern void j_rev_dct_ARM(DCTELEM *data);
extern void simple_idct_ARM(DCTELEM *data);

/* XXX: local hack */
static void (*ff_put_pixels_clamped)(const DCTELEM *block, uint8_t *pixels, int line_size);
static void (*ff_add_pixels_clamped)(const DCTELEM *block, uint8_t *pixels, int line_size);

void put_pixels8_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
void put_pixels8_x2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
void put_pixels8_y2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
void put_pixels8_xy2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);

void put_no_rnd_pixels8_x2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
void put_no_rnd_pixels8_y2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
void put_no_rnd_pixels8_xy2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);

void put_pixels16_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);

CALL_2X_PIXELS(put_pixels16_x2_arm , put_pixels8_x2_arm , 8)
CALL_2X_PIXELS(put_pixels16_y2_arm , put_pixels8_y2_arm , 8)
CALL_2X_PIXELS(put_pixels16_xy2_arm, put_pixels8_xy2_arm, 8)
CALL_2X_PIXELS(put_no_rnd_pixels16_x2_arm , put_no_rnd_pixels8_x2_arm , 8)
CALL_2X_PIXELS(put_no_rnd_pixels16_y2_arm , put_no_rnd_pixels8_y2_arm , 8)
CALL_2X_PIXELS(put_no_rnd_pixels16_xy2_arm, put_no_rnd_pixels8_xy2_arm, 8)

static void add_pixels_clamped_ARM(short *block, unsigned char *dest, int line_size)
{
    asm volatile (
                  "mov r10, #8 \n\t"

                  "1: \n\t"

                  /* load dest */
                  "ldr r4, [%1] \n\t"
                  /* block[0] and block[1]*/
                  "ldrsh r5, [%0] \n\t"
                  "ldrsh r7, [%0, #2] \n\t"
                  "and r6, r4, #0xFF \n\t"
                  "and r8, r4, #0xFF00 \n\t"
                  "add r6, r5, r6 \n\t"
                  "add r8, r7, r8, lsr #8 \n\t"
                  "mvn r5, r5 \n\t"
                  "mvn r7, r7 \n\t"
                  "tst r6, #0x100 \n\t"
                  "movne r6, r5, lsr #24 \n\t"
                  "tst r8, #0x100 \n\t"
                  "movne r8, r7, lsr #24 \n\t"
                  "mov r9, r6 \n\t"
                  "ldrsh r5, [%0, #4] \n\t" /* moved form [A] */
                  "orr r9, r9, r8, lsl #8 \n\t"
                  /* block[2] and block[3] */
                  /* [A] */
                  "ldrsh r7, [%0, #6] \n\t"
                  "and r6, r4, #0xFF0000 \n\t"
                  "and r8, r4, #0xFF000000 \n\t"
                  "add r6, r5, r6, lsr #16 \n\t"
                  "add r8, r7, r8, lsr #24 \n\t"
                  "mvn r5, r5 \n\t"
                  "mvn r7, r7 \n\t"
                  "tst r6, #0x100 \n\t"
                  "movne r6, r5, lsr #24 \n\t"
                  "tst r8, #0x100 \n\t"
                  "movne r8, r7, lsr #24 \n\t"
                  "orr r9, r9, r6, lsl #16 \n\t"
                  "ldr r4, [%1, #4] \n\t"       /* moved form [B] */
                  "orr r9, r9, r8, lsl #24 \n\t"
                  /* store dest */
                  "ldrsh r5, [%0, #8] \n\t" /* moved form [C] */
                  "str r9, [%1] \n\t"

                  /* load dest */
                  /* [B] */
                  /* block[4] and block[5] */
                  /* [C] */
                  "ldrsh r7, [%0, #10] \n\t"
                  "and r6, r4, #0xFF \n\t"
                  "and r8, r4, #0xFF00 \n\t"
                  "add r6, r5, r6 \n\t"
                  "add r8, r7, r8, lsr #8 \n\t"
                  "mvn r5, r5 \n\t"
                  "mvn r7, r7 \n\t"
                  "tst r6, #0x100 \n\t"
                  "movne r6, r5, lsr #24 \n\t"
                  "tst r8, #0x100 \n\t"
                  "movne r8, r7, lsr #24 \n\t"
                  "mov r9, r6 \n\t"
                  "ldrsh r5, [%0, #12] \n\t" /* moved from [D] */
                  "orr r9, r9, r8, lsl #8 \n\t"
                  /* block[6] and block[7] */
                  /* [D] */
                  "ldrsh r7, [%0, #14] \n\t"
                  "and r6, r4, #0xFF0000 \n\t"
                  "and r8, r4, #0xFF000000 \n\t"
                  "add r6, r5, r6, lsr #16 \n\t"
                  "add r8, r7, r8, lsr #24 \n\t"
                  "mvn r5, r5 \n\t"
                  "mvn r7, r7 \n\t"
                  "tst r6, #0x100 \n\t"
                  "movne r6, r5, lsr #24 \n\t"
                  "tst r8, #0x100 \n\t"
                  "movne r8, r7, lsr #24 \n\t"
                  "orr r9, r9, r6, lsl #16 \n\t"
                  "add %0, %0, #16 \n\t" /* moved from [E] */
                  "orr r9, r9, r8, lsl #24 \n\t"
                  "subs r10, r10, #1 \n\t" /* moved from [F] */
                  /* store dest */
                  "str r9, [%1, #4] \n\t"

                  /* [E] */
                  /* [F] */
                  "add %1, %1, %2 \n\t"
                  "bne 1b \n\t"
                  : "+r"(block),
                    "+r"(dest)
                  : "r"(line_size)
                  : "r4", "r5", "r6", "r7", "r8", "r9", "r10", "cc", "memory" );
}

/* XXX: those functions should be suppressed ASAP when all IDCTs are
   converted */
static void j_rev_dct_ARM_put(uint8_t *dest, int line_size, DCTELEM *block)
{
    j_rev_dct_ARM (block);
    ff_put_pixels_clamped(block, dest, line_size);
}
static void j_rev_dct_ARM_add(uint8_t *dest, int line_size, DCTELEM *block)
{
    j_rev_dct_ARM (block);
    ff_add_pixels_clamped(block, dest, line_size);
}
static void simple_idct_ARM_put(uint8_t *dest, int line_size, DCTELEM *block)
{
    simple_idct_ARM (block);
    ff_put_pixels_clamped(block, dest, line_size);
}
static void simple_idct_ARM_add(uint8_t *dest, int line_size, DCTELEM *block)
{
    simple_idct_ARM (block);
    ff_add_pixels_clamped(block, dest, line_size);
}
static void simple_idct_ipp(DCTELEM *block)
{
#ifdef HAVE_IPP
    ippiDCT8x8Inv_Video_16s_C1I(block);
#endif
}
static void simple_idct_ipp_put(uint8_t *dest, int line_size, DCTELEM *block)
{
#ifdef HAVE_IPP
    ippiDCT8x8Inv_Video_16s8u_C1R(block, dest, line_size);
#endif
}

void add_pixels_clamped_iwmmxt(const DCTELEM *block, uint8_t *pixels, int line_size);

static void simple_idct_ipp_add(uint8_t *dest, int line_size, DCTELEM *block)
{
#ifdef HAVE_IPP
    ippiDCT8x8Inv_Video_16s_C1I(block);
#ifdef HAVE_IWMMXT
    add_pixels_clamped_iwmmxt(block, dest, line_size);
#else
    add_pixels_clamped_ARM(block, dest, line_size);
#endif
#endif
}

void dsputil_init_armv4l(DSPContext* c, AVCodecContext *avctx)
{
    const int idct_algo= avctx->idct_algo;

    ff_put_pixels_clamped = c->put_pixels_clamped;
    ff_add_pixels_clamped = c->add_pixels_clamped;

#ifdef HAVE_IPP
    if(idct_algo==FF_IDCT_ARM){
#else
    if(idct_algo==FF_IDCT_AUTO || idct_algo==FF_IDCT_ARM){
#endif
        c->idct_put= j_rev_dct_ARM_put;
        c->idct_add= j_rev_dct_ARM_add;
        c->idct    = j_rev_dct_ARM;
        c->idct_permutation_type= FF_LIBMPEG2_IDCT_PERM;/* FF_NO_IDCT_PERM */
    } else if (idct_algo==FF_IDCT_SIMPLEARM){
        c->idct_put= simple_idct_ARM_put;
        c->idct_add= simple_idct_ARM_add;
        c->idct    = simple_idct_ARM;
        c->idct_permutation_type= FF_NO_IDCT_PERM;
#ifdef HAVE_IPP
    } else if (idct_algo==FF_IDCT_AUTO || idct_algo==FF_IDCT_IPP){
#else
    } else if (idct_algo==FF_IDCT_IPP){
#endif
        c->idct_put= simple_idct_ipp_put;
        c->idct_add= simple_idct_ipp_add;
        c->idct    = simple_idct_ipp;
        c->idct_permutation_type= FF_NO_IDCT_PERM;
    }

/*     c->put_pixels_tab[0][0] = put_pixels16_arm; */ // NG!
    c->put_pixels_tab[0][1] = put_pixels16_x2_arm; //OK!
    c->put_pixels_tab[0][2] = put_pixels16_y2_arm; //OK!
/*     c->put_pixels_tab[0][3] = put_pixels16_xy2_arm; /\* NG *\/ */
/*     c->put_no_rnd_pixels_tab[0][0] = put_pixels16_arm; // ?(使われない) */
    c->put_no_rnd_pixels_tab[0][1] = put_no_rnd_pixels16_x2_arm; // OK
    c->put_no_rnd_pixels_tab[0][2] = put_no_rnd_pixels16_y2_arm; //OK
/*     c->put_no_rnd_pixels_tab[0][3] = put_no_rnd_pixels16_xy2_arm; //NG */
    c->put_pixels_tab[1][0] = put_pixels8_arm; //OK
    c->put_pixels_tab[1][1] = put_pixels8_x2_arm; //OK
/*     c->put_pixels_tab[1][2] = put_pixels8_y2_arm; //NG */
/*     c->put_pixels_tab[1][3] = put_pixels8_xy2_arm; //NG */
    c->put_no_rnd_pixels_tab[1][0] = put_pixels8_arm;//OK
    c->put_no_rnd_pixels_tab[1][1] = put_no_rnd_pixels8_x2_arm; //OK
    c->put_no_rnd_pixels_tab[1][2] = put_no_rnd_pixels8_y2_arm; //OK
/*     c->put_no_rnd_pixels_tab[1][3] = put_no_rnd_pixels8_xy2_arm;//NG */

#ifdef HAVE_IWMMXT
    dsputil_init_iwmmxt(c, avctx);
#endif
}