comparison i386/mpegvideo_mmx.c @ 2967:ef2149182f1c libavcodec

COSMETICS: Remove all trailing whitespace.

2966:564788471dd4

    long level, qmul, qadd, nCoeffs;

    qmul = qscale << 1;

    assert(s->block_last_index[n]>=0 || s->h263_aic);
        
    if (!s->h263_aic) {
        if (n < 4)
            level = block[0] * s->y_dc_scale;
        else
            level = block[0] * s->c_dc_scale;

    qmul = qscale << 1;
    qadd = (qscale - 1) | 1;

    assert(s->block_last_index[n]>=0 || s->h263_aic);
        
    nCoeffs= s->inter_scantable.raster_end[ s->block_last_index[n] ];
//printf("%d %d  ", qmul, qadd);
asm volatile(
		"movd %1, %%mm6			\n\t" //qmul
		"packssdw %%mm6, %%mm6		\n\t"

    assert(s->block_last_index[n]>=0);

    nCoeffs= s->intra_scantable.raster_end[ s->block_last_index[n] ]+1;

    if (n < 4) 
        block0 = block[0] * s->y_dc_scale;
    else
        block0 = block[0] * s->c_dc_scale;
    /* XXX: only mpeg1 */
    quant_matrix = s->intra_matrix;

		"add $16, %%"REG_a"		\n\t"
		"js 1b				\n\t"
		::"r" (block+nCoeffs), "r"(quant_matrix+nCoeffs), "g" (qscale), "g" (-2*nCoeffs)
		: "%"REG_a, "memory"
	);    
    block[0]= block0;
}

static void dct_unquantize_mpeg1_inter_mmx(MpegEncContext *s,
                                     DCTELEM *block, int n, int qscale)

                                     DCTELEM *block, int n, int qscale)
{
    long nCoeffs;
    const uint16_t *quant_matrix;
    int block0;
    
    assert(s->block_last_index[n]>=0);

    if(s->alternate_scan) nCoeffs= 63; //FIXME
    else nCoeffs= s->intra_scantable.raster_end[ s->block_last_index[n] ];

    if (n < 4) 
        block0 = block[0] * s->y_dc_scale;
    else
        block0 = block[0] * s->c_dc_scale;
    quant_matrix = s->intra_matrix;
asm volatile(

		"add $16, %%"REG_a"		\n\t"
		"jng 1b				\n\t"
		::"r" (block+nCoeffs), "r"(quant_matrix+nCoeffs), "g" (qscale), "g" (-2*nCoeffs)
		: "%"REG_a, "memory"
	);    
    block[0]= block0;
        //Note, we dont do mismatch control for intra as errors cannot accumulate
}

static void dct_unquantize_mpeg2_inter_mmx(MpegEncContext *s,
                                     DCTELEM *block, int n, int qscale)
{
    long nCoeffs;
    const uint16_t *quant_matrix;
    
    assert(s->block_last_index[n]>=0);

    if(s->alternate_scan) nCoeffs= 63; //FIXME
    else nCoeffs= s->intra_scantable.raster_end[ s->block_last_index[n] ];

                "pxor %%mm6, %%mm7		\n\t"
                "pslld $31, %%mm7		\n\t"
                "psrlq $15, %%mm7		\n\t"
                "pxor %%mm7, %%mm0		\n\t"
                "movd %%mm0, 124(%0, %3)	\n\t"
                
		::"r" (block+nCoeffs), "r"(quant_matrix+nCoeffs), "g" (qscale), "r" (-2*nCoeffs)
		: "%"REG_a, "memory"
	);
}

/* draw the edges of width 'w' of an image of size width, height 
   this mmx version can only handle w==8 || w==16 */
static void draw_edges_mmx(uint8_t *buf, int wrap, int width, int height, int w)
{
    uint8_t *ptr, *last_line;
    int i;

    if(w==8)
    {
	asm volatile(
		"1:				\n\t"
		"movd (%0), %%mm0		\n\t"
		"punpcklbw %%mm0, %%mm0		\n\t" 
		"punpcklwd %%mm0, %%mm0		\n\t"
		"punpckldq %%mm0, %%mm0		\n\t"
		"movq %%mm0, -8(%0)		\n\t"
		"movq -8(%0, %2), %%mm1		\n\t"
		"punpckhbw %%mm1, %%mm1		\n\t"

    else
    {
	asm volatile(
		"1:				\n\t"
		"movd (%0), %%mm0		\n\t"
		"punpcklbw %%mm0, %%mm0		\n\t" 
		"punpcklwd %%mm0, %%mm0		\n\t"
		"punpckldq %%mm0, %%mm0		\n\t"
		"movq %%mm0, -8(%0)		\n\t"
		"movq %%mm0, -16(%0)		\n\t"
		"movq -8(%0, %2), %%mm1		\n\t"

		"punpckhdq %%mm1, %%mm1		\n\t"
		"movq %%mm1, (%0, %2)		\n\t"
		"movq %%mm1, 8(%0, %2)		\n\t"
		"add %1, %0			\n\t"
		"cmp %3, %0			\n\t"
		" jb 1b				\n\t"		
		: "+r" (ptr)
		: "r" ((long)wrap), "r" ((long)width), "r" (ptr + wrap*height)
	);
    }
    
    for(i=0;i<w;i+=4) {
        /* top and bottom (and hopefully also the corners) */
	ptr= buf - (i + 1) * wrap - w;
	asm volatile(
		"1:				\n\t"

void MPV_common_init_mmx(MpegEncContext *s)
{
    if (mm_flags & MM_MMX) {
        const int dct_algo = s->avctx->dct_algo;
        
        s->dct_unquantize_h263_intra = dct_unquantize_h263_intra_mmx;
        s->dct_unquantize_h263_inter = dct_unquantize_h263_inter_mmx;
        s->dct_unquantize_mpeg1_intra = dct_unquantize_mpeg1_intra_mmx;
        s->dct_unquantize_mpeg1_inter = dct_unquantize_mpeg1_inter_mmx;
        s->dct_unquantize_mpeg2_intra = dct_unquantize_mpeg2_intra_mmx;
        s->dct_unquantize_mpeg2_inter = dct_unquantize_mpeg2_inter_mmx;

        draw_edges = draw_edges_mmx;
        
        if (mm_flags & MM_SSE2) {
	    s->denoise_dct= denoise_dct_sse2;
	} else {
    	    s->denoise_dct= denoise_dct_mmx;
	}

2967:ef2149182f1c

    long level, qmul, qadd, nCoeffs;

    qmul = qscale << 1;

    assert(s->block_last_index[n]>=0 || s->h263_aic);

    if (!s->h263_aic) {
        if (n < 4)
            level = block[0] * s->y_dc_scale;
        else
            level = block[0] * s->c_dc_scale;

    qmul = qscale << 1;
    qadd = (qscale - 1) | 1;

    assert(s->block_last_index[n]>=0 || s->h263_aic);

    nCoeffs= s->inter_scantable.raster_end[ s->block_last_index[n] ];
//printf("%d %d  ", qmul, qadd);
asm volatile(
		"movd %1, %%mm6			\n\t" //qmul
		"packssdw %%mm6, %%mm6		\n\t"

    assert(s->block_last_index[n]>=0);

    nCoeffs= s->intra_scantable.raster_end[ s->block_last_index[n] ]+1;

    if (n < 4)
        block0 = block[0] * s->y_dc_scale;
    else
        block0 = block[0] * s->c_dc_scale;
    /* XXX: only mpeg1 */
    quant_matrix = s->intra_matrix;

		"add $16, %%"REG_a"		\n\t"
		"js 1b				\n\t"
		::"r" (block+nCoeffs), "r"(quant_matrix+nCoeffs), "g" (qscale), "g" (-2*nCoeffs)
		: "%"REG_a, "memory"
	);
    block[0]= block0;
}

static void dct_unquantize_mpeg1_inter_mmx(MpegEncContext *s,
                                     DCTELEM *block, int n, int qscale)

                                     DCTELEM *block, int n, int qscale)
{
    long nCoeffs;
    const uint16_t *quant_matrix;
    int block0;

    assert(s->block_last_index[n]>=0);

    if(s->alternate_scan) nCoeffs= 63; //FIXME
    else nCoeffs= s->intra_scantable.raster_end[ s->block_last_index[n] ];

    if (n < 4)
        block0 = block[0] * s->y_dc_scale;
    else
        block0 = block[0] * s->c_dc_scale;
    quant_matrix = s->intra_matrix;
asm volatile(

		"add $16, %%"REG_a"		\n\t"
		"jng 1b				\n\t"
		::"r" (block+nCoeffs), "r"(quant_matrix+nCoeffs), "g" (qscale), "g" (-2*nCoeffs)
		: "%"REG_a, "memory"
	);
    block[0]= block0;
        //Note, we dont do mismatch control for intra as errors cannot accumulate
}

static void dct_unquantize_mpeg2_inter_mmx(MpegEncContext *s,
                                     DCTELEM *block, int n, int qscale)
{
    long nCoeffs;
    const uint16_t *quant_matrix;

    assert(s->block_last_index[n]>=0);

    if(s->alternate_scan) nCoeffs= 63; //FIXME
    else nCoeffs= s->intra_scantable.raster_end[ s->block_last_index[n] ];

                "pxor %%mm6, %%mm7		\n\t"
                "pslld $31, %%mm7		\n\t"
                "psrlq $15, %%mm7		\n\t"
                "pxor %%mm7, %%mm0		\n\t"
                "movd %%mm0, 124(%0, %3)	\n\t"

		::"r" (block+nCoeffs), "r"(quant_matrix+nCoeffs), "g" (qscale), "r" (-2*nCoeffs)
		: "%"REG_a, "memory"
	);
}

/* draw the edges of width 'w' of an image of size width, height
   this mmx version can only handle w==8 || w==16 */
static void draw_edges_mmx(uint8_t *buf, int wrap, int width, int height, int w)
{
    uint8_t *ptr, *last_line;
    int i;

    if(w==8)
    {
	asm volatile(
		"1:				\n\t"
		"movd (%0), %%mm0		\n\t"
		"punpcklbw %%mm0, %%mm0		\n\t"
		"punpcklwd %%mm0, %%mm0		\n\t"
		"punpckldq %%mm0, %%mm0		\n\t"
		"movq %%mm0, -8(%0)		\n\t"
		"movq -8(%0, %2), %%mm1		\n\t"
		"punpckhbw %%mm1, %%mm1		\n\t"

    else
    {
	asm volatile(
		"1:				\n\t"
		"movd (%0), %%mm0		\n\t"
		"punpcklbw %%mm0, %%mm0		\n\t"
		"punpcklwd %%mm0, %%mm0		\n\t"
		"punpckldq %%mm0, %%mm0		\n\t"
		"movq %%mm0, -8(%0)		\n\t"
		"movq %%mm0, -16(%0)		\n\t"
		"movq -8(%0, %2), %%mm1		\n\t"

		"punpckhdq %%mm1, %%mm1		\n\t"
		"movq %%mm1, (%0, %2)		\n\t"
		"movq %%mm1, 8(%0, %2)		\n\t"
		"add %1, %0			\n\t"
		"cmp %3, %0			\n\t"
		" jb 1b				\n\t"
		: "+r" (ptr)
		: "r" ((long)wrap), "r" ((long)width), "r" (ptr + wrap*height)
	);
    }

    for(i=0;i<w;i+=4) {
        /* top and bottom (and hopefully also the corners) */
	ptr= buf - (i + 1) * wrap - w;
	asm volatile(
		"1:				\n\t"

void MPV_common_init_mmx(MpegEncContext *s)
{
    if (mm_flags & MM_MMX) {
        const int dct_algo = s->avctx->dct_algo;

        s->dct_unquantize_h263_intra = dct_unquantize_h263_intra_mmx;
        s->dct_unquantize_h263_inter = dct_unquantize_h263_inter_mmx;
        s->dct_unquantize_mpeg1_intra = dct_unquantize_mpeg1_intra_mmx;
        s->dct_unquantize_mpeg1_inter = dct_unquantize_mpeg1_inter_mmx;
        s->dct_unquantize_mpeg2_intra = dct_unquantize_mpeg2_intra_mmx;
        s->dct_unquantize_mpeg2_inter = dct_unquantize_mpeg2_inter_mmx;

        draw_edges = draw_edges_mmx;

        if (mm_flags & MM_SSE2) {
	    s->denoise_dct= denoise_dct_sse2;
	} else {
    	    s->denoise_dct= denoise_dct_mmx;
	}