import std;

​

void main()

{

    float[3] pos = [0f, 0f, 0f];

    float[3][] verts = [[1f,1f,1f],[2f,2f,2f],[3f,3f,3f],[4f,4f,4f],[5f,5f,5f],[6f,6f,6f],[7f,7f,7f],[8f,8f,8f],[9f,9f,9f],[10f,10f,10f],[11f,11f,11f],[12f,12f,12f]];

    writeln(vEUCLIDps(pos, verts));

}

​

float[] vEUCLIDps(float[3] pos, float[3][] verts)

{

    //"broadcast" single pos into an array.

    float[3][] broadcastedpos;

    broadcastedpos.length = verts.length;

    broadcastedpos.fill(pos);

    debug writefln!"d1 %s"(broadcastedpos);

    debug writefln!"d2 %s"(verts);

    for(size_t i; i < verts.length; ++i){//bugged if done with nested foreach and ref, contents are incorrect...

        broadcastedpos[i][] -= verts[i][];//VSUBPS

        broadcastedpos[i][] *= broadcastedpos[i][];//VMULPS

    }

    debug writefln!"d3 %s"(broadcastedpos);

    //broadcastedpos[][0..3] -= verts[][0..3];//doesn't work

    //broadcastedpos[][] *= broadcastedpos[][];//neither does this work

    float[] sums;

    sums.length = verts.length;

    foreach (i, ref vertice; broadcastedpos)

    {

        //sums[i] = sum(vertice);//sum does not support fixed size arrays

        sums[i] = vertice[0] + vertice[1] + vertice[2];

        sums[i] = sqrt(sums[i]);//should see SQRTPS in the assembly output, with ymm operands.

    }

    return sums;

    //return cast(float[])map!(a => sqrt(a))(sums);//incompatible types

    //no need for abs.

​

}