view libmpeg2/motion_comp_alpha.c @ 16946:47c5e9846cd3

ultra simple&slow pp filter, yes yet another spp like filter :) this one does actually compress&decompress the video at various shifts with lavc while the other spp filters are doing optimized intra only filtering limitations: mpeg4 is hardcoded, all options too, pretty trivial to change though, even filtering with non dct codecs like snow could be tried ... the qscale/qp is only taken fron the first MB of each image and then used for the whole image (would needs some small changes to lavc to let the user set the qscales for the mbs themselfs but iam to lazy ...) this needs ALOT of cpu time and memory especially at uspp=8 ...

/*
 * motion_comp_alpha.c
 * Copyright (C) 2002-2003 Falk Hueffner <[e-mail]>
 *
 * This file is part of mpeg2dec, a free MPEG-2 video stream decoder.
 * See http://libmpeg2.sourceforge.net/ for updates.
 *
 * mpeg2dec is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * mpeg2dec is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 */

#include "config.h"

#ifdef ARCH_ALPHA

#include <inttypes.h>

#include "mpeg2.h"
#include "attributes.h"
#include "mpeg2_internal.h"
#include "alpha_asm.h"

static inline uint64_t avg2 (uint64_t a, uint64_t b)
{
    return (a | b) - (((a ^ b) & BYTE_VEC (0xfe)) >> 1);
}

// Load two unaligned quadwords from addr. This macro only works if
// addr is actually unaligned.
#define ULOAD16(ret_l,ret_r,addr)			\
    do {						\
	uint64_t _l = ldq_u (addr +  0);		\
	uint64_t _m = ldq_u (addr +  8);		\
	uint64_t _r = ldq_u (addr + 16);		\
	ret_l = extql (_l, addr) | extqh (_m, addr);	\
	ret_r = extql (_m, addr) | extqh (_r, addr);	\
    } while (0)

// Load two aligned quadwords from addr.
#define ALOAD16(ret_l,ret_r,addr)			\
    do {						\
	ret_l = ldq (addr);				\
	ret_r = ldq (addr + 8);				\
    } while (0)

#define OP8(LOAD,LOAD16,STORE)			\
    do {					\
	STORE (LOAD (pixels), block);		\
	pixels += line_size;			\
	block += line_size;			\
    } while (--h)

#define OP16(LOAD,LOAD16,STORE)			\
    do {					\
	uint64_t l, r;				\
	LOAD16 (l, r, pixels);			\
	STORE (l, block);			\
	STORE (r, block + 8);			\
	pixels += line_size;			\
	block += line_size;			\
    } while (--h)

#define OP8_X2(LOAD,LOAD16,STORE)			\
    do {						\
	uint64_t p0, p1;				\
							\
	p0 = LOAD (pixels);				\
	p1 = p0 >> 8 | ((uint64_t) pixels[8] << 56);	\
	STORE (avg2 (p0, p1), block);			\
	pixels += line_size;				\
	block += line_size;				\
    } while (--h)

#define OP16_X2(LOAD,LOAD16,STORE)				\
    do {							\
	uint64_t p0, p1;					\
								\
	LOAD16 (p0, p1, pixels);				\
	STORE (avg2(p0, p0 >> 8 | p1 << 56), block);		\
	STORE (avg2(p1, p1 >> 8 | (uint64_t) pixels[16] << 56),	\
	       block + 8);					\
	pixels += line_size;					\
	block += line_size;					\
    } while (--h)

#define OP8_Y2(LOAD,LOAD16,STORE)		\
    do {					\
	uint64_t p0, p1;			\
	p0 = LOAD (pixels);			\
	pixels += line_size;			\
	p1 = LOAD (pixels);			\
	do {					\
	    uint64_t av = avg2 (p0, p1);	\
	    if (--h == 0) line_size = 0;	\
	    pixels += line_size;		\
	    p0 = p1;				\
	    p1 = LOAD (pixels);			\
	    STORE (av, block);			\
	    block += line_size;			\
	} while (h);				\
    } while (0)

#define OP16_Y2(LOAD,LOAD16,STORE)		\
    do {					\
	uint64_t p0l, p0r, p1l, p1r;		\
	LOAD16 (p0l, p0r, pixels);		\
	pixels += line_size;			\
	LOAD16 (p1l, p1r, pixels);		\
	do {					\
	    uint64_t avl, avr;			\
	    if (--h == 0) line_size = 0;	\
	    avl = avg2 (p0l, p1l);		\
	    avr = avg2 (p0r, p1r);		\
	    p0l = p1l;				\
	    p0r = p1r;				\
	    pixels += line_size;		\
	    LOAD16 (p1l, p1r, pixels);		\
	    STORE (avl, block);			\
	    STORE (avr, block + 8);		\
	    block += line_size;			\
	} while (h);				\
    } while (0)

#define OP8_XY2(LOAD,LOAD16,STORE)				\
    do {							\
	uint64_t pl, ph;					\
	uint64_t p1 = LOAD (pixels);				\
	uint64_t p2 = p1 >> 8 | ((uint64_t) pixels[8] << 56);	\
								\
	ph = (((p1 & ~BYTE_VEC (0x03)) >> 2) +			\
	      ((p2 & ~BYTE_VEC (0x03)) >> 2));			\
	pl = ((p1 & BYTE_VEC (0x03)) +				\
	      (p2 & BYTE_VEC (0x03)));				\
								\
	do {							\
	    uint64_t npl, nph;					\
								\
	    pixels += line_size;				\
	    p1 = LOAD (pixels);					\
	    p2 = (p1 >> 8) | ((uint64_t) pixels[8] << 56);	\
	    nph = (((p1 & ~BYTE_VEC (0x03)) >> 2) +		\
	           ((p2 & ~BYTE_VEC (0x03)) >> 2));		\
	    npl = ((p1 & BYTE_VEC (0x03)) +			\
	           (p2 & BYTE_VEC (0x03)));			\
								\
	    STORE (ph + nph +					\
		   (((pl + npl + BYTE_VEC (0x02)) >> 2) &	\
		    BYTE_VEC (0x03)), block);			\
								\
	    block += line_size;					\
            pl = npl;						\
	    ph = nph;						\
	} while (--h);						\
    } while (0)

#define OP16_XY2(LOAD,LOAD16,STORE)				\
    do {							\
	uint64_t p0, p1, p2, p3, pl_l, ph_l, pl_r, ph_r;	\
	LOAD16 (p0, p2, pixels);				\
	p1 = p0 >> 8 | (p2 << 56);				\
	p3 = p2 >> 8 | ((uint64_t)pixels[16] << 56);		\
								\
	ph_l = (((p0 & ~BYTE_VEC (0x03)) >> 2) +		\
	        ((p1 & ~BYTE_VEC (0x03)) >> 2));		\
	pl_l = ((p0 & BYTE_VEC (0x03)) +			\
	        (p1 & BYTE_VEC(0x03)));				\
	ph_r = (((p2 & ~BYTE_VEC (0x03)) >> 2) +		\
	        ((p3 & ~BYTE_VEC (0x03)) >> 2));		\
	pl_r = ((p2 & BYTE_VEC (0x03)) +			\
	        (p3 & BYTE_VEC (0x03)));			\
								\
	do {							\
	    uint64_t npl_l, nph_l, npl_r, nph_r;		\
								\
	    pixels += line_size;				\
	    LOAD16 (p0, p2, pixels);				\
	    p1 = p0 >> 8 | (p2 << 56);				\
	    p3 = p2 >> 8 | ((uint64_t)pixels[16] << 56);	\
	    nph_l = (((p0 & ~BYTE_VEC (0x03)) >> 2) +		\
		     ((p1 & ~BYTE_VEC (0x03)) >> 2));		\
	    npl_l = ((p0 & BYTE_VEC (0x03)) +			\
		     (p1 & BYTE_VEC (0x03)));			\
	    nph_r = (((p2 & ~BYTE_VEC (0x03)) >> 2) +		\
		     ((p3 & ~BYTE_VEC (0x03)) >> 2));		\
	    npl_r = ((p2 & BYTE_VEC (0x03)) +			\
		     (p3 & BYTE_VEC (0x03)));			\
								\
	    STORE (ph_l + nph_l +				\
		   (((pl_l + npl_l + BYTE_VEC (0x02)) >> 2) &	\
		    BYTE_VEC(0x03)), block);			\
	    STORE (ph_r + nph_r +				\
		   (((pl_r + npl_r + BYTE_VEC (0x02)) >> 2) &	\
		    BYTE_VEC(0x03)), block + 8);		\
								\
	    block += line_size;					\
	    pl_l = npl_l;					\
	    ph_l = nph_l;					\
	    pl_r = npl_r;					\
	    ph_r = nph_r;					\
	} while (--h);						\
    } while (0)

#define MAKE_OP(OPNAME,SIZE,SUFF,OPKIND,STORE)				\
static void MC_ ## OPNAME ## _ ## SUFF ## _ ## SIZE ## _alpha		\
	(uint8_t *restrict block, const uint8_t *restrict pixels,	\
	 int line_size, int h)						\
{									\
    if ((uint64_t) pixels & 0x7) {					\
	OPKIND (uldq, ULOAD16, STORE);					\
    } else {								\
	OPKIND (ldq, ALOAD16, STORE);					\
    }									\
}

#define PIXOP(OPNAME,STORE)			\
    MAKE_OP (OPNAME, 8,  o,  OP8,      STORE);	\
    MAKE_OP (OPNAME, 8,  x,  OP8_X2,   STORE);	\
    MAKE_OP (OPNAME, 8,  y,  OP8_Y2,   STORE);	\
    MAKE_OP (OPNAME, 8,  xy, OP8_XY2,  STORE);	\
    MAKE_OP (OPNAME, 16, o,  OP16,     STORE);	\
    MAKE_OP (OPNAME, 16, x,  OP16_X2,  STORE);	\
    MAKE_OP (OPNAME, 16, y,  OP16_Y2,  STORE);	\
    MAKE_OP (OPNAME, 16, xy, OP16_XY2, STORE);

#define STORE(l,b) stq (l, b)
PIXOP (put, STORE);
#undef STORE
#define STORE(l,b) stq (avg2 (l, ldq (b)), b);
PIXOP (avg, STORE);

mpeg2_mc_t mpeg2_mc_alpha = {
    { MC_put_o_16_alpha, MC_put_x_16_alpha,
      MC_put_y_16_alpha, MC_put_xy_16_alpha,
      MC_put_o_8_alpha, MC_put_x_8_alpha,
      MC_put_y_8_alpha, MC_put_xy_8_alpha },
    { MC_avg_o_16_alpha, MC_avg_x_16_alpha,
      MC_avg_y_16_alpha, MC_avg_xy_16_alpha,
      MC_avg_o_8_alpha, MC_avg_x_8_alpha,
      MC_avg_y_8_alpha, MC_avg_xy_8_alpha }
};

#endif