IntrinsicTestSuperFast.cs

public static int IntrinsicTest(byte[] lhs, byte[] rhs)
{
    byte ths = 16;
    Intrinsics.Vector128<byte> threshold = Intrinsics.Vector128.Create(ths, ths, ths, ths, ths, ths, ths, ths, ths, ths, ths, ths, ths, ths, ths, ths);
    byte bit = 1;
    var VECTOR_ONE = Intrinsics.Vector128.Create(bit, bit, bit, bit, bit, bit, bit, bit, bit, bit, bit, bit, bit, bit, bit, bit);

    var result = new Intrinsics.Vector128<int>();
    var i = 0;
    for (i = 0; i <= lhs.Length - 16; i += 16)
    {
        //using System.Numerics.Vector for loading, then converting to Intrinsics.Vector128
        var v128a = new Vector<byte>(lhs, i).AsVector128();
        var v128b = new Vector<byte>(rhs, i).AsVector128();

        var subtracted = Intrinsics.Arm.AdvSimd.AbsoluteDifference(v128a, v128b);

        var thresholdPassed = Intrinsics.Arm.AdvSimd.CompareGreaterThanOrEqual(subtracted, threshold);

        var normalized = Intrinsics.Vector128.AsSByte(thresholdPassed);

        var rtemp = Intrinsics.Arm.AdvSimd.AddPairwiseWidening(normalized);
        result = Intrinsics.Arm.AdvSimd.AddPairwiseWideningAndAdd(result, rtemp);

    }

    return -1 * (int)(Intrinsics.Arm.AdvSimd.Extract(result, 0)
                    + Intrinsics.Arm.AdvSimd.Extract(result, 1)
                    + Intrinsics.Arm.AdvSimd.Extract(result, 2)
                    + Intrinsics.Arm.AdvSimd.Extract(result, 3));
}