comparison ppc/float_altivec.c @ 4387:05ab6e897ffd libavcodec

10l, there is a corner case afterall...

4386:09849d8689d6

    int i;
    vector float d, s0, s1, s2, t0, t1, edges;
    vector unsigned char align = vec_lvsr(0,dst),
                         mask = vec_lvsl(0, dst);

    t0 = vec_ld(0, dst);
#if 0 //FIXME: there is still something wrong
    if (step == 2) {
        int y;
        vector float d0, d1, s3, t2;
        vector unsigned int sel =

        }
    } else
    #endif
    if (step == 1 && src3 == 0)
        for (i=0; i<len-3; i+=4) {
            t1 = vec_ld(15, dst+i);
            s0 = vec_ld(0, src0+i);
            s1 = vec_ld(0, src1+i);
            s2 = vec_ld(0, src2+i);
            edges = vec_perm(t1 ,t0, mask);
            d = vec_madd(s0,s1,s2);
            t1 = vec_perm(d, edges, align);
            t0 = vec_perm(edges, d, align);
            vec_st(t1, 15, dst+i);
            vec_st(t0, 0, dst+i);
            t0 = t1;
        }
    else
        ff_vector_fmul_add_add_c(dst, src0, src1, src2, src3, len, step);
}

4387:05ab6e897ffd

    int i;
    vector float d, s0, s1, s2, t0, t1, edges;
    vector unsigned char align = vec_lvsr(0,dst),
                         mask = vec_lvsl(0, dst);

#if 0 //FIXME: there is still something wrong
    if (step == 2) {
        int y;
        vector float d0, d1, s3, t2;
        vector unsigned int sel =

        }
    } else
    #endif
    if (step == 1 && src3 == 0)
        for (i=0; i<len-3; i+=4) {
            t0 = vec_ld(0, dst+i);
            t1 = vec_ld(15, dst+i);
            s0 = vec_ld(0, src0+i);
            s1 = vec_ld(0, src1+i);
            s2 = vec_ld(0, src2+i);
            edges = vec_perm(t1 ,t0, mask);
            d = vec_madd(s0,s1,s2);
            t1 = vec_perm(d, edges, align);
            t0 = vec_perm(edges, d, align);
            vec_st(t1, 15, dst+i);
            vec_st(t0, 0, dst+i);
        }
    else
        ff_vector_fmul_add_add_c(dst, src0, src1, src2, src3, len, step);
}