Trace Cache: A Low Latency Approach to High-Bandwidth Instruction Fetching

Q1. With a conventional instruction fetch, what would it take to fetch
    40 instructions in a cycle (for example, porting requirements for each
    structure)?

Q2. What are the weaknesses in the paper?