Athena: Synergizing Data Prefetching and Off-Chip Prediction via Online Reinforcement Learning

This artifact contains the implementation and evaluation infrastructure for Athena, a reinforcement learning-based technique for synergizing data prefetching and off-chip prediction. The artifact is built on top of ChampSim, a trace-driven CPU simulator.

---

Table of Contents

1. Overview
2. Directory Structure
3. Building the Simulator
4. Obtaining Traces
5. Running Experiments
6. Understanding Results

---

Overview

The artifact supports:

• Multiple data prefetchers and off-chip predictors:

  • L1D Prefetchers: IPCP, Berti
  • L2C Prefetchers: Pythia, SPP+PPF, MLOP, SMS
  • Off-Chip Predictors (OCP): POPET, HMP, TTP

• Coordination mechanisms compared:

  • Naive (naive combination)
  • TLP (Two Level Perceptron)
  • HPAC (Hierarchical Prefetcher Aggressiveness Control)
  • MAB (Micro-Armed Bandit)
  • Athena (our proposed learning-based coordination)

---

Directory Structure

oogway/
├── bin/                    # Compiled simulator binary (champsim)
├── branch/                 # Branch predictor implementations
├── config/                 # Configuration files (.ini)
│   ├── oogway_dev.ini      # Athena configuration
│   ├── pythia.ini          # Pythia configuration
│   └── ...
├── experiments/            # Experiment outputs and results
│   ├── Fig5a/              # Raw simulation outputs for Fig5a
│   ├── Fig5b/              # Raw simulation outputs for Fig5b
│   ├── ...
│   └── results/            # Aggregated CSVs and plots
├── inc/                    # Header files
│   ├── oogway.h            # Athena implementation
│   ├── scooby.h            # Pythia prefetcher
│   └── ...
├── obj/                    # Object files (generated during build)
├── prefetcher/             # Prefetcher implementations
├── replacement/            # Cache replacement policies
├── scripts/                # Experiment management scripts
│   ├── athena.py           # Main entry point for experiments
│   ├── config.py           # Experiment configurations
│   ├── generate.py         # Job generation
│   ├── rollup.py           # Result aggregation
│   ├── visualize.py        # Visualization
│   └── ...
├── src/                    # ChampSim core source files
├── traces/                 # Trace files (user must download)
├── checksum.txt            # SHA256 checksums for trace verification
├── Makefile                # Build configuration
├── setvars.sh              # Environment setup script
└── wrapper.sh              # Job wrapper for Slurm

---

Building the Simulator

1. Clone the repository

git clone https://github.com/CMU-SAFARI/Athena.git
cd Athena

2. Set up the environment

source setvars.sh

This sets the ATHENA_HOME environment variable required by all scripts.

3. Build the simulator

make clean
make -j$(nproc)

4. Verify the build

ls bin/champsim
# Should show: bin/champsim

The build produces a single binary bin/champsim that supports all prefetcher and off-chip predictor configurations via command-line arguments.

---

Obtaining Traces

The artifact requires workload traces to run simulations. Traces are not included due to size (~30 GB total).

Download Instructions

1. Download the workload traces

curl -L "https://zenodo.org/api/records/17850673/files-archive" -o download.zip

2. Unzip the workload traces

mkdir traces
unzip download.zip -d $ATHENA_HOME/traces

Verify Trace Integrity

mv checksum.txt $ATHENA_HOME/traces
cd $ATHENA_HOME/traces
sha256sum -c checksum.txt

Trace Count

The full evaluation uses 100 workload traces across four benchmark suites:

• SPEC: 49 traces
• PARSEC: 13 traces
• LIGRA: 13 traces
• CVP: 25 traces

---

Running Experiments

Using the Athena Tool

Before launching the experiments, please make sure to update DEFAULT_NCORES, DEFAULT_PARTITION, DEFAULT_HOSTNAME, and other Slurm-related settings in $ATHENA_HOME/scripts/config.py

The athena.py script provides a unified interface for all experiment operations.

cd $ATHENA_HOME/scripts

# Launch experiments (requires Slurm cluster)
python athena.py -L <Figure> 
# Summarize results from simulation outputs
python athena.py -S <Figure>
# Relaunch failed experiments
python athena.py -R <Figure>
# Visualize results
python athena.py -V <Figure>

Available Experiments (Figures)

Figure	Description	Coordination	Components
Fig5a	OCP + L2C	POPET + Pythia	CD1
Fig5b	OCP + L1D	POPET + IPCP	CD2
Fig5c	OCP + 2 L2C	POPET + SMS + Pythia	CD3
Fig5d	OCP + L1D + L2C	POPET + IPCP + Pythia	CD4
Fig6b	L2C Type Sensitivity	POPET + various L2C	Pythia/SPP+PPF/MLOP/SMS
Fig6c	OCP Latency Sensitivity	POPET + Pythia	6/18/30 cycles
Fig6d	OCP Type Sensitivity	Various OCP + Pythia	POPET/HMP/TTP
Fig7a	L1D Type Sensitivity	POPET + various L1D + Pythia	IPCP/Berti
Fig7b	Bandwidth Sensitivity	POPET + IPCP + Pythia	1.6-12.8 GB/s
Fig8	L2C Only	SMS + Pythia (no OCP)	Generalizability Study

Example: Running Figure 5a

# 1. Set environment
source setvars.sh

# 2. Launch experiments (on Slurm cluster)
cd $ATHENA_HOME/scripts
python athena.py -L Fig5a 

# 3. Wait for jobs to complete (check with squeue)
# Each trace-experiment combination takes ~3 hours

# 4. Summarize results
python athena.py -S Fig5a

# 5. Relaunch experiments if needed (and summarize again)
python athena.py -R Fig5a
python athena.py -S Fig5a

# 6. Visualize
python athena.py -V Fig5a

Understanding Results

Output Format

Simulation outputs are stored in experiments/<Figure>/:

• <trace>_<experiment>.out - Simulation statistics
• <trace>_<experiment>.err - Error/debug output

CSV Format

Aggregated results in experiments/results/<Figure>.csv:

Column	Description
Trace	Workload trace name
Exp	Experiment configuration name
Core_0_cumulative_IPC	Instructions Per Cycle (main metric)
Filter	1 if all experiments for this trace completed

Key Metrics

The primary metric is IPC Speedup over baseline (no prefetching or OCP):

Speedup = IPC_experiment / IPC_baseline

Results are aggregated using geometric mean across traces, grouped by:

• Workload type (SPEC, PARSEC, LIGRA, CVP)
• Prefetcher-sensitivity (adverse vs. friendly)
• Overall