Please answer this question for me with explanantion specified!

Consider the problem of trying to get accurate benchmark results from a method. Suppose that you have two techniques to help: $1)$ Restarting your computer between each benchmark run. $2)$ Running the benchmark a number of times and averaging the result.

What would you prefer: a$)$ None of them, b$)$ using #$1$ only, c$)$ using #$2$ only, or d$)$ using both? Select your choice, and explain how it will lead to a more accurate benchmark result.