Question 3 [10 points]
You are trying to optimize the classical five-stage pipeline used in your company. The current pipeline uses forwarding to avoid stalls as much as possible. After doing some profiling, you discover that the ALU stage of the pipeline is taking the longest to execute, which slows down the clock. As a result, the execution time of the applications is higher. In order to address this issue, you propose to split the ALU into two pipeline stages. With the ALU split into two pipeline stages, you now have a six-stage pipeline. Note that the result of the ALU is only available after the second ALU stage and the data after the first ALU state is not useful. Describe how this change of adding a pipeline stage affects Read after write (RAW) hazards. How would you resolve the new complications for RAW hazards that result from the six-stage pipeline? Please list changes and/or hardware needed for resolving the RAW hazards. You may use two consecutive ALU instructions with a RAW hazard for the purposes of this question.

Question 4 (Short answer questions) [20 points]
(a) Amdahl’s law provides a very useful metric to measure speedup when we make applications parallel using more resources. However, Amdahl’s law’s basic form makes an assumption about the application that may not hold in real-world scenarios. State the assumption made by Amdahl’s law and describe conditions that may break the assumption.
(b) What type of misses are reduced by each of these cache optimization techniques. List all types of misses reduced for full credit. In addition, list the possible disadvantages of using the optimization technique.
• Data and instruction prefetching:

• Pipelined cache accesses:
• Higher associativity:
• Larger cache capacity:

DETAILED ASSIGNMENT

20201009172905eee524_cpts561_fall20_midterm