Answer from cs61c-by (toan nguyen 16197920) for Question 1
Since the TLB sizes are typically much, much smaller
than L1 caches, the cost of fully associative mappings is cheap. Therefore, we can use fully associative mappings, which reduces the miss rate.