ND4j concat takes considerable long time

Slow.java

public class Slow {
    public static void main(String[] args) {

        int hiddenDim = 200;
        int numChar = 100;
        int length = 500;
        int batchSize = 50;
        int[] pshape = new int[]{numChar, hiddenDim};
        INDArray c2v = xavier(pshape);

        INDArray h0 = Nd4j.zeros(batchSize, hiddenDim);
        INDArray c0 = Nd4j.zeros(batchSize, hiddenDim);

        long start = System.currentTimeMillis();

        INDArray fwdmap = Nd4j.zeros(batchSize, numChar);

        for (int i = 0; i < length; i++) {
            INDArray embed = fwdmap.mmul(c2v); // 1
            INDArray concat = Nd4j.concat(1, embed, h0); // 2
        }
        System.out.println(System.currentTimeMillis() - start);
    }


    static INDArray xavier(int[] shape) {
        int n = 1;
        for (int i = 0; i < shape.length - 1; i++)
            n *= shape[i];
        double sd = Math.sqrt(3d / n);
        return new UniformDistribution(-sd, sd).sample(shape);
    }
}