Force-Driven-ML-RCS

This project implements a Force-Driven List Scheduling (FD-ML-RCS) algorithm along with baseline List Scheduling. It includes automation scripts for running experiments across various scaling factors, verifying results, and generating visualizations.

Project Structure

File/Folder	Description
Source Code
LSMain.cpp	Main entry point for the scheduler.
LS.cpp, LS.h	Implementation of List Scheduling algorithms.
FDS.cpp	(Initial experimental) Implementation of Force-Directed Scheduling.
ReadInputs.cpp	Helper to parse DFG files and constraints.
checker.cpp, checker.h	(Changed) Verifier to validate scheduling results.
Scripts
run_code.bash	Main wrapper. Compiles and runs the scheduler or checker for a single instance.
automatic_run_code.bash	Batch Automation. Runs experiments across all scaling factors (1.0 to 0.1) in all configurations.
run_checker.py	Python script to execute the checker and parse results into related CSVs.
run_graphs.py	Python script to generate visualization graphs from the output CSVs.
Data & Configuration
DFG/	Folder containing input Data Flow Graphs (.txt).
Constraints/	Resource constraint files.
lib_4type.txt	Library file defining operation delays and resource types.
Output
CSV/	Results stored as .csv files, organized by scaling factor.
Graphs/	Generated plots visualizing latency, runtime, and improvements.
Results/	Intermediate result storage, organized by scaling factor.
Documentation
README.md	This documentation file.
Docs/	Slides and Report from Midterm, Slides from Final.

---

Setup & Prerequisites

1. Requirements

• C++ Compiler: g++.
• Python 3: Required for checker integration and graph generation.
• Bash Terminal: Required to run the .bash automation scripts.

2. Permissions

Ensure the scripts are executable:

chmod +x run_code.bash automatic_run_code.bash

3. Environment setup

Create the virtual environment

python3 -m venv venv

Activate it

source venv/bin/activate

Install dependencies

pip install -r requirements.txt

Manual Execution (run_code.bash)

Use this script to run a single scheduling experiment or verify a single CSV output file. It handles compilation automatically.

Usage

./run_code.bash [mode] [options]

Modes

• run: Runs the scheduling algorithm.
• check: Runs the verifier on a specific CSV result file.
• --help / -h: Displays help information.

Options for run Mode

Flag	Description
--base / -B	Run the Standard LS Implementation (Default is enhanced with Force).
--uniform / -U	Use Uniform benchmark (Default is InvDelay).
--invdelay / -I	Use Inverse Delay distribution.
--featS / -S	Enable Feature S (resource constraints) for priorities.
--featP / -P	Enable Feature P (predecessors) for priorities.
--scaling=X / -F=X	Set scaling factor (e.g., -F=0.5). Default: 1.0.
--debug / -D	Enable verbose debug output.

Example:

# Run Standard LS with Uniform distribution at 0.8 scaling
./run_code.bash run -U -B -F=0.8

Options for check Mode

Pass the CSV file to verify as the second argument.

Example:

# Verify a specific result file
./run_code.bash check CSV/1.00/Results_LS_uniform.csv

---

Automatic Batch Execution (automatic_run_code.bash)

Use this script to reproduce the full set of experiments. It iterates through scaling factors from 1.0 down to 0.1, for every configuration:

• Distributions: Uniform, Inverse Delay.
• Algorithms: Standard LS, FD-ML-RCS with No Features, FD-ML-RCS with Feature S & P.

Usage

./automatic_run_code.bash [options]

Options

Flag	Description
--run / -R	Execute Runs: Runs 6 configurations per scaling factor (Uniform/InvDelay x Standard/NoFeat/FeatSP).
--check / -C	Execute Checks: Scans the CSV/ directory and verifies every generated .csv file.

Example:

# Run everything and check everything
./automatic_run_code.bash -R -C

---

Generating Graphs

To visualize the results after running the experiments, use the python script:

python3 run_graphs.py

This will generate plots in the Graphs/ directory, comparing:

• Latency Deltas (LS versions) vs Baseline (FALLS)
• Runtime analysis.
• Improvement percentages.

Note: This project has been developed and tested on macOS. For running on Windows, it may be necessary to install the required tools via WSL or adapt the scripts accordingly.