Observe in the following code, creating an numpy array and calling the builtin python sum function produces different results than numpy.sum

How is numpy’s sum function implemented? And why is the result different?

test = [.1]*10
test = [np.float64(x) for x in test]
test[5]= np.float64(-.9)

d = [np.asarray(test) for x in range(0,60000)]
sum(sum(d))

outputs

np.float64(-1.7473212210461497e-08)

but

np.sum(d)

outputs

np.float64(9.987344284922983e-12)