comparison x86/fft_3dn2.c @ 10199:38ab367d4231 libavcodec

Merge FFTContext and MDCTContext

10198:78af613fc316

    if(n <= 8)
        for(i=0; i<n; i+=2)
            FFSWAP(FFTSample, z[i].im, z[i+1].re);
}

void ff_imdct_half_3dn2(MDCTContext *s, FFTSample *output, const FFTSample *input)
{
    x86_reg j, k;
    long n = 1 << s->nbits;
    long n2 = n >> 1;
    long n4 = n >> 2;
    long n8 = n >> 3;
    const uint16_t *revtab = s->fft.revtab;
    const FFTSample *tcos = s->tcos;
    const FFTSample *tsin = s->tsin;
    const FFTSample *in1, *in2;
    FFTComplex *z = (FFTComplex *)output;

            "movq    %%mm0, %0    \n\t"
            :"=m"(z[revtab[k]])
        );
    }

    ff_fft_dispatch_3dn2(z, s->fft.nbits);

#define CMUL(j,mm0,mm1)\
        "movq  (%2,"#j",2), %%mm6 \n"\
        "movq 8(%2,"#j",2), "#mm0"\n"\
        "movq        %%mm6, "#mm1"\n"\

        :"memory"
    );
    __asm__ volatile("femms");
}

void ff_imdct_calc_3dn2(MDCTContext *s, FFTSample *output, const FFTSample *input)
{
    x86_reg j, k;
    long n = 1 << s->nbits;
    long n4 = n >> 2;

    ff_imdct_half_3dn2(s, output+n4, input);

    j = -n;

10199:38ab367d4231

    if(n <= 8)
        for(i=0; i<n; i+=2)
            FFSWAP(FFTSample, z[i].im, z[i+1].re);
}

void ff_imdct_half_3dn2(FFTContext *s, FFTSample *output, const FFTSample *input)
{
    x86_reg j, k;
    long n = 1 << s->mdct_bits;
    long n2 = n >> 1;
    long n4 = n >> 2;
    long n8 = n >> 3;
    const uint16_t *revtab = s->revtab;
    const FFTSample *tcos = s->tcos;
    const FFTSample *tsin = s->tsin;
    const FFTSample *in1, *in2;
    FFTComplex *z = (FFTComplex *)output;

            "movq    %%mm0, %0    \n\t"
            :"=m"(z[revtab[k]])
        );
    }

    ff_fft_dispatch_3dn2(z, s->nbits);

#define CMUL(j,mm0,mm1)\
        "movq  (%2,"#j",2), %%mm6 \n"\
        "movq 8(%2,"#j",2), "#mm0"\n"\
        "movq        %%mm6, "#mm1"\n"\

        :"memory"
    );
    __asm__ volatile("femms");
}

void ff_imdct_calc_3dn2(FFTContext *s, FFTSample *output, const FFTSample *input)
{
    x86_reg j, k;
    long n = 1 << s->mdct_bits;
    long n4 = n >> 2;

    ff_imdct_half_3dn2(s, output+n4, input);

    j = -n;