Edge Generation Scheduling

Edge Generation Scheduling (EGS) is an algorithm for DAG scheduling using deep reinforcement learning (DRL).

How to Run EGS

First, install the required python packages listed in requirements.txt using pip or conda. Then, run the EGS with:

python egs.py

Optional arguments:

• --in_dot: the path to the input .dot file (default: data/in_dag.dot).
• --out_dot: the path to the output .dot file (default: data/out_dag.dot).
• --model: the directory of the pretrained neural network model (default: models/pretrained). If no pretrained model is provided, a random policy will be used instead.
• --workers: the number of workers (processors) used to schedule the input DAG task (default: None). If the number of workers is not specified, the EGS will return a schedule with the minimum number of workers.
• --gpu_id: the ID of the GPU that is going to be used for neural network inference (default: 0).