ECEC-412/621
Project 4: Implementing Cache Replacement Policy and evaluating its
performance using AI Workloads
1 PC-Signature-based Hit Predictor
1. Recent research (Section 3.2 of Supplement One) suggests that the reuse behavior of a cache block is
strongly correlated with the PC that inserted it into the cache. Read through Supplement One and
design a cache replacement policy based on the PC-Signature-based Hit Predictor.
2. Evaluate your cache replacement policy using all the AI workloads and compare against LRU and LFU.
2 Submission
1. Report how you design the cache replacement policy with a diagram and pseudo-code.
2. Summarize the experiment in Section 1.
3. Compile above all in a single PDF file.
4. All source codes.
5. Zip above all and submit through Bblearn.
1
