comparison i386/dsputil_h264_template_mmx.c @ 8031:eebc7209c47f libavcodec

Convert asm keyword into __asm__. Neither the asm() nor the __asm__() keyword is part of the C99 standard, but while GCC accepts the former in C89 syntax, it is not accepted in C99 unless GNU extensions are turned on (with -fasm). The latter form is accepted in any syntax as an extension (without requiring further command-line options). Sun Studio C99 compiler also does not accept asm() while accepting __asm__(), albeit reporting warnings that it's not valid C99 syntax.

8030:a512ac8fa540

        /* 1 dimensional filter only */
        const int dxy = x ? 1 : stride;

        rnd_reg = rnd ? &ff_pw_4 : &ff_pw_3;

        asm volatile(
            "movd %0, %%mm5\n\t"
            "movq %1, %%mm4\n\t"
            "movq %2, %%mm6\n\t"         /* mm6 = rnd */
            "punpcklwd %%mm5, %%mm5\n\t"
            "punpckldq %%mm5, %%mm5\n\t" /* mm5 = B = x */
            "pxor %%mm7, %%mm7\n\t"
            "psubw %%mm5, %%mm4\n\t"     /* mm4 = A = 8-x */
            :: "rm"(x+y), "m"(ff_pw_8), "m"(*rnd_reg));

        for(i=0; i<h; i++) {
            asm volatile(
                /* mm0 = src[0..7], mm1 = src[1..8] */
                "movq %0, %%mm0\n\t"
                "movq %1, %%mm2\n\t"
                :: "m"(src[0]), "m"(src[dxy]));

            asm volatile(
                /* [mm0,mm1] = A * src[0..7] */
                /* [mm2,mm3] = B * src[1..8] */
                "movq %%mm0, %%mm1\n\t"
                "movq %%mm2, %%mm3\n\t"
                "punpcklbw %%mm7, %%mm0\n\t"

        return;
    }

    /* general case, bilinear */
    rnd_reg = rnd ? &ff_pw_32.a : &ff_pw_28.a;
    asm volatile("movd %2, %%mm4\n\t"
                 "movd %3, %%mm6\n\t"
                 "punpcklwd %%mm4, %%mm4\n\t"
                 "punpcklwd %%mm6, %%mm6\n\t"
                 "punpckldq %%mm4, %%mm4\n\t" /* mm4 = x words */
                 "punpckldq %%mm6, %%mm6\n\t" /* mm6 = y words */

                 "psubw %%mm7, %%mm4\n\t"     /* mm4 = A = xy - (8x+8y) + 64 */
                 "pxor %%mm7, %%mm7\n\t"
                 "movq %%mm4, %0\n\t"
                 : "=m" (AA), "=m" (DD) : "rm" (x), "rm" (y), "m" (ff_pw_64));

    asm volatile(
        /* mm0 = src[0..7], mm1 = src[1..8] */
        "movq %0, %%mm0\n\t"
        "movq %1, %%mm1\n\t"
        : : "m" (src[0]), "m" (src[1]));

    for(i=0; i<h; i++) {
        src += stride;

        asm volatile(
            /* mm2 = A * src[0..3] + B * src[1..4] */
            /* mm3 = A * src[4..7] + B * src[5..8] */
            "movq %%mm0, %%mm2\n\t"
            "movq %%mm1, %%mm3\n\t"
            "punpckhbw %%mm7, %%mm0\n\t"

            "pmullw %%mm5, %%mm3\n\t"
            "paddw %%mm1, %%mm2\n\t"
            "paddw %%mm0, %%mm3\n\t"
            : : "m" (AA));

        asm volatile(
            /* [mm2,mm3] += C * src[0..7] */
            "movq %0, %%mm0\n\t"
            "movq %%mm0, %%mm1\n\t"
            "punpcklbw %%mm7, %%mm0\n\t"
            "punpckhbw %%mm7, %%mm1\n\t"

            "pmullw %%mm6, %%mm1\n\t"
            "paddw %%mm0, %%mm2\n\t"
            "paddw %%mm1, %%mm3\n\t"
            : : "m" (src[0]));

        asm volatile(
            /* [mm2,mm3] += D * src[1..8] */
            "movq %1, %%mm1\n\t"
            "movq %%mm1, %%mm0\n\t"
            "movq %%mm1, %%mm4\n\t"
            "punpcklbw %%mm7, %%mm0\n\t"

            "paddw %%mm0, %%mm2\n\t"
            "paddw %%mm4, %%mm3\n\t"
            "movq %0, %%mm0\n\t"
            : : "m" (src[0]), "m" (src[1]), "m" (DD));

        asm volatile(
            /* dst[0..7] = ([mm2,mm3] + 32) >> 6 */
            "paddw %1, %%mm2\n\t"
            "paddw %1, %%mm3\n\t"
            "psrlw $6, %%mm2\n\t"
            "psrlw $6, %%mm3\n\t"

    }
}

static void H264_CHROMA_MC4_TMPL(uint8_t *dst/*align 4*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y)
{
    asm volatile(
        "pxor   %%mm7, %%mm7        \n\t"
        "movd %5, %%mm2             \n\t"
        "movd %6, %%mm3             \n\t"
        "movq "MANGLE(ff_pw_8)", %%mm4\n\t"
        "movq "MANGLE(ff_pw_8)", %%mm5\n\t"

static void H264_CHROMA_MC2_TMPL(uint8_t *dst/*align 2*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y)
{
    int tmp = ((1<<16)-1)*x + 8;
    int CD= tmp*y;
    int AB= (tmp<<3) - CD;
    asm volatile(
        /* mm5 = {A,B,A,B} */
        /* mm6 = {C,D,C,D} */
        "movd %0, %%mm5\n\t"
        "movd %1, %%mm6\n\t"
        "punpckldq %%mm5, %%mm5\n\t"

        "punpcklbw %%mm7, %%mm2\n\t"
        "pshufw $0x94, %%mm2, %%mm2\n\t"
        :: "r"(AB), "r"(CD), "m"(src[0]));


    asm volatile(
        "1:\n\t"
        "add %4, %1\n\t"
        /* mm1 = A * src[0,1] + B * src[1,2] */
        "movq    %%mm2, %%mm1\n\t"
        "pmaddwd %%mm5, %%mm1\n\t"

8031:eebc7209c47f

        /* 1 dimensional filter only */
        const int dxy = x ? 1 : stride;

        rnd_reg = rnd ? &ff_pw_4 : &ff_pw_3;

        __asm__ volatile(
            "movd %0, %%mm5\n\t"
            "movq %1, %%mm4\n\t"
            "movq %2, %%mm6\n\t"         /* mm6 = rnd */
            "punpcklwd %%mm5, %%mm5\n\t"
            "punpckldq %%mm5, %%mm5\n\t" /* mm5 = B = x */
            "pxor %%mm7, %%mm7\n\t"
            "psubw %%mm5, %%mm4\n\t"     /* mm4 = A = 8-x */
            :: "rm"(x+y), "m"(ff_pw_8), "m"(*rnd_reg));

        for(i=0; i<h; i++) {
            __asm__ volatile(
                /* mm0 = src[0..7], mm1 = src[1..8] */
                "movq %0, %%mm0\n\t"
                "movq %1, %%mm2\n\t"
                :: "m"(src[0]), "m"(src[dxy]));

            __asm__ volatile(
                /* [mm0,mm1] = A * src[0..7] */
                /* [mm2,mm3] = B * src[1..8] */
                "movq %%mm0, %%mm1\n\t"
                "movq %%mm2, %%mm3\n\t"
                "punpcklbw %%mm7, %%mm0\n\t"

        return;
    }

    /* general case, bilinear */
    rnd_reg = rnd ? &ff_pw_32.a : &ff_pw_28.a;
    __asm__ volatile("movd %2, %%mm4\n\t"
                 "movd %3, %%mm6\n\t"
                 "punpcklwd %%mm4, %%mm4\n\t"
                 "punpcklwd %%mm6, %%mm6\n\t"
                 "punpckldq %%mm4, %%mm4\n\t" /* mm4 = x words */
                 "punpckldq %%mm6, %%mm6\n\t" /* mm6 = y words */

                 "psubw %%mm7, %%mm4\n\t"     /* mm4 = A = xy - (8x+8y) + 64 */
                 "pxor %%mm7, %%mm7\n\t"
                 "movq %%mm4, %0\n\t"
                 : "=m" (AA), "=m" (DD) : "rm" (x), "rm" (y), "m" (ff_pw_64));

    __asm__ volatile(
        /* mm0 = src[0..7], mm1 = src[1..8] */
        "movq %0, %%mm0\n\t"
        "movq %1, %%mm1\n\t"
        : : "m" (src[0]), "m" (src[1]));

    for(i=0; i<h; i++) {
        src += stride;

        __asm__ volatile(
            /* mm2 = A * src[0..3] + B * src[1..4] */
            /* mm3 = A * src[4..7] + B * src[5..8] */
            "movq %%mm0, %%mm2\n\t"
            "movq %%mm1, %%mm3\n\t"
            "punpckhbw %%mm7, %%mm0\n\t"

            "pmullw %%mm5, %%mm3\n\t"
            "paddw %%mm1, %%mm2\n\t"
            "paddw %%mm0, %%mm3\n\t"
            : : "m" (AA));

        __asm__ volatile(
            /* [mm2,mm3] += C * src[0..7] */
            "movq %0, %%mm0\n\t"
            "movq %%mm0, %%mm1\n\t"
            "punpcklbw %%mm7, %%mm0\n\t"
            "punpckhbw %%mm7, %%mm1\n\t"

            "pmullw %%mm6, %%mm1\n\t"
            "paddw %%mm0, %%mm2\n\t"
            "paddw %%mm1, %%mm3\n\t"
            : : "m" (src[0]));

        __asm__ volatile(
            /* [mm2,mm3] += D * src[1..8] */
            "movq %1, %%mm1\n\t"
            "movq %%mm1, %%mm0\n\t"
            "movq %%mm1, %%mm4\n\t"
            "punpcklbw %%mm7, %%mm0\n\t"

            "paddw %%mm0, %%mm2\n\t"
            "paddw %%mm4, %%mm3\n\t"
            "movq %0, %%mm0\n\t"
            : : "m" (src[0]), "m" (src[1]), "m" (DD));

        __asm__ volatile(
            /* dst[0..7] = ([mm2,mm3] + 32) >> 6 */
            "paddw %1, %%mm2\n\t"
            "paddw %1, %%mm3\n\t"
            "psrlw $6, %%mm2\n\t"
            "psrlw $6, %%mm3\n\t"

    }
}

static void H264_CHROMA_MC4_TMPL(uint8_t *dst/*align 4*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y)
{
    __asm__ volatile(
        "pxor   %%mm7, %%mm7        \n\t"
        "movd %5, %%mm2             \n\t"
        "movd %6, %%mm3             \n\t"
        "movq "MANGLE(ff_pw_8)", %%mm4\n\t"
        "movq "MANGLE(ff_pw_8)", %%mm5\n\t"

static void H264_CHROMA_MC2_TMPL(uint8_t *dst/*align 2*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y)
{
    int tmp = ((1<<16)-1)*x + 8;
    int CD= tmp*y;
    int AB= (tmp<<3) - CD;
    __asm__ volatile(
        /* mm5 = {A,B,A,B} */
        /* mm6 = {C,D,C,D} */
        "movd %0, %%mm5\n\t"
        "movd %1, %%mm6\n\t"
        "punpckldq %%mm5, %%mm5\n\t"

        "punpcklbw %%mm7, %%mm2\n\t"
        "pshufw $0x94, %%mm2, %%mm2\n\t"
        :: "r"(AB), "r"(CD), "m"(src[0]));


    __asm__ volatile(
        "1:\n\t"
        "add %4, %1\n\t"
        /* mm1 = A * src[0,1] + B * src[1,2] */
        "movq    %%mm2, %%mm1\n\t"
        "pmaddwd %%mm5, %%mm1\n\t"