BanditWare

An online recommendation system that dynamically selects the most suitable hardware for applications using a contextual multi-armed bandit algorithm.

Setup

1. Set Up a Virtual Environment:

   • With Python's built-in venv:

     python3 -m venv env

     source env/bin/activate

2. Install Dependencies:

   • Install required packages:

     pip3 install -r requirements.txt

Running BanditWare:

1. Choose a dataset you would like to use. Choose a path to a folder or csv file that holds various runs of an application.
   • For a folder, label subfolders as follows: {n}cores_{m}gb, where n is the number of cpu cores and m is the number of gigabytes of RAM that the application was run on. Make sure all csv files in subfolders have a runtime column and any feature columns of the application.
   • For a csv file, make sure the columns contain runtime and any feature columns of the application.Encode the number of cpu cores and gigabytes of RAM in a hardware column with values like "2_16" to represent 2 cpu cores and 16 GB of RAM.
2. In the terminal, if you have a data folder, run python3 pre_process.py --base_path={PATH_TO_DATA} where PATH_TO_DATA is the path from step 1. If you have a single csv, run `python3 pre_process.py --data_file={PATH_TO_DATA_FILE}, where PATH_TO_DATA_FILE is the csv defined in part 1.
   • For example, python3 pre_process.py --base_path=data/bp3d_data or python3 pre_process.py --base_path=data/matmul.csv
3. Finally, to see some of the functionality of BanditWare, run examples.py after selecting the dataset you would like to run on and specifying the feature columns of that dataset.