questions.md

Assignment 1 Questions

IMPORTANT Do not reformat this file! Put your answers below each question. Use markdown formatting.

[25 points] How to reproduce the results

Explanation of the script

Script to run

Parameters to script (if any)

Commands used to gather data

Commands used for Step II

Commands used for Step III

[75 points] Questions

[8 points each] Step I: Write down your hypotheses and experimental setup

1. When you change the ISA from RISC-V to x86, what do you expect to happen to the performance of the system? Use the Iron Law of Performance to justify your answer.

2. When you change the CPU frequency from 1GHz to 4GHz, what do you expect to happen to the performance of the system? Use the Iron Law of Performance to justify your answer.

3. When you change the CPU frequency from 1GHz to 4GHz, will the speedup from 1GHz to 4GHz be the same for all ISAs? Why or why not?

4. Describe the experimental setup to answer the research question. What are the independent and dependent variables? What is the baseline?

[7 points each] Step II: Investigating the impact of the ISA

1. What is the performance for matrix multiplication for each ISA?

2. Can you match the difference in performance to the Iron Law of Performance? Why or why not?

[7 points each] Step III: Investigating the impact of the CPU and cache clock frequency

1. What is the speedup of the system when you change the CPU and cache frequency from 1GHz to 2GHz to 4GHz? Show the speedup for each ISA.

2. Does the Iron Law of Performance correctly predict the speedup for each ISA? Why or why not?

[25 points] Research question:

For a simple CPU model (i.e., fixed microarchitecture), does ISA or technology make a bigger impact on system performance?

[10 points] Next steps

1. If the workload had a significantly better (lower) CPI, how would that change the results of the experiments? E.g., what would happen if the workload and microarchitecture supported a CPI of 0.25 (or 4 instructions per cycle)?