aashere/gpu-project

Accelerating Hierarchical Clustering with GPU

Sanjar Ahmadov, Rajath Ravindranath Dani, & Ameya Shere

Directory Structure

This codebase is structured as follows.

• src folder contains the source code for all 4 implementations
• target folder contains the binaries for all 4 implementations
• compile.sh compiles the source code and generates the binaries
• run_experiments.py runs experiments for different problem sizes across the different clusters
• collect_experiments_results.py collects and generates a summary of the experiments results in a given folder

.
├── compile.sh
├── run_experiments.py
├── collect_experiments_results.py
└── src
│   ├── cpu_clustering.cpp
│   ├── naive_cpu_clustering.cpp
│   ├── single_gpu_clustering.cu
│   └── single_gpu_clustering_sort.cu
└── target
    ├── cpu1
    ├── cpu2
    ├── gpu1
    └── gpu2

Instructions

1. cd to root
2. Enable script permissions: chmod +x compile.sh
3. Compile and generate binaries in target folder: ./compile.sh
4. To run any of the 4 binaries: ./target/<binary-name> N M, where N is number of samples to generate and M is dimension of samples
5. To run the experiments: ./run_experiments.py folder_path
6. To get the outputs of finished experiments: ./collect_experiments_results.py folder_path