Project 3: Implementing Cache Replacement Policy

ECEC-412/621
Project 3: Implementing Cache Replacement Policy and evaluating its
performance using AI Workloads

1 Introduction
This project is intended to be a comprehensive introduction to cache replacement policy. There are two
parts to this project:
• In part one, you will evaluate the performance of the Least-recently Used (LRU) replacement policy
with three AI workloads from SPEC CPU 2017.
• In part two, you will implement a Least-frequently Used (LFU) replacement policy and evaluate its
performance against part one.
2 Cache Replacement Policy Framework Overview
1. You will be working under directory Computer-Architecture-Teaching/C621/Cache Policy. To navigate
to the directory:
$ cd Computer-Architecture-Teaching/C621/Cache_Policy/
2. To compile and run the simulator:
$ make
$ ./Main sample.mem_trace
3. You should be able to see the following output:
Hit rate: 42.580000%
4. sample.mem trace is part of the leela AI workload from SPEC CPU 2017. To take a look at the format
of the trace file:
$ tail -10 sample.mem_trace
1 94009581925312 89824316667136 L
1 94009581924100 89824316666944 L
1 94009581932431 139713177052544 L
2 94009581932431 139686110733696 L
2 139693221530966 139686111208832 S
0 94009582105510 132740395877696 L
0 139693221530353 103162374522176 S
1 94009582105510 139713177052480 L
1 139693221530949 139713176905024 S
2 94009582105510 139686110733632 L
Each entry is composed of four components:
1
• Core ID: the core that loads/stores this cache block.
• Program Counter: the PC of the instruction that loads/stores this cache block.
• Cache Block Address: the physical address of the loaded/stored cache block.
• Load or Store.
3 Cache Replacement Policy Framework Configuration
The simulator models a set-associative last-level cache (LLC) with a LRU replacement policy. To configure
the simulator:
1. Open Cache.c:
$ vim Cache.c
2. You are encouraged to try different combinations of cache size and assoc, and observe the changes of
cache performance (hit rate). Please keep the cache block size to 64.
const unsigned b l o c k s i z e = 6 4; // S i z e o f a cache l i n e ( in By tes )
const unsigned c a c h e s i z e = 1 2 8; // S i z e o f a cache ( in KB)
const unsigned a s s o c = 4 ;
3. Re-compile and run the simulator:
$ make clean
$ make
$ ./Main sample.mem_trace
4 Experiment
1. AI workloads (Memory Traces): https://www.dropbox.com/sh/8fee5t192yche7s/AADr-T_7GKE185ye-CAAROOEa?
dl=0.
2. The simulator adopts LRU replacement policy by default. Simulate all the AI workloads with different
combinations of cache size and assoc. What combinations do you try and what do you find?
3. Implement a Least-frequently Used (LFU) replacement policy. Evaluate your policy against LRU.
Which policy performs the best?
5 Submission
1. Summarize your experiment in Section 4. Compile your report in PDF format.
2. All the source codes.
3. Zip above and submit through Bblearn.
2