This can be built up from first principles and combinations. Its official name is a hypergeometric random variable. Consider the basic probability ratio: the number of successes divided by the number of possible outcomes. We'll consider both, one at a time.
First let's consider the denominator. We are selecting 2 computers from a total of 8. This is 8C2, which is the total number of possible outcomes.
Next, let's look at the numerator, which is the number of successes. We do not have a specified number for how many are defective, so this is our variable x. How many defective computers are there to choose from? Since there are 3, there are 3Cx possible selections of x computers out of 3 defective ones.
But this does not tell the whole story. Just as there are 3 defective computers, there are 5 non-defective ones. Whatever remains of the 2 computers selected must come from the 5 non-defective computers. In combination language, this is 5C2-x.
So here's the final product:
(3Cx) (5C2-x)
P(X = x) = ---------------------
(8C2)