How to calculate covariance using numpy by taking only the first row of the result?

Question

I want to use numpy to calculate which vector is the closest to the input one among a large number of vectors, I want to use the covariance as the basis of the calculation. But numpy.cov calculates the covariance of all combinations of any pair in the matrix by default, this can waste a lot of unnecessary calculations when the input matrix is large.

# Desired Result

input vector -------------> alternative vector 1  ----------> result vector
[f64, f64...]      |        [f64, f64...]              |      [f64, ...n] 
                   ├------> alternative vector 2  -----┤
                   |        [f64, f64...]              |
                   ├------> alternative vector 3  -----┤
                   |        [f64, f64...]              |
                   ...                                 ...
                   n vectors

I am now adopting the following practices

>>> import numpy as np
>>> input_vec = np.random.normal(size=(1,100))
>>> vecs = np.random.normal(size=(10000,100))
>>> covmatrix = np.cov(np.vstack((input_vec, vecs)), bias=True)
>>> covmatrix.shape
(10001, 10001)
>>> result = covmatrix[0]

Is there any way to speed up the calculation? Thanks