Answer from cs61c-bd (Dae-hyeon Wi 17782249) for Question 1
TLB is small, so we can apply fully associative cache model in TLB. It makes miss rates of TLB lower.