DFQ-BFS: A Decentralized Frontier Queue for Improving Scalability of Breadth-First-Search on GPUs

1. Getting started Instructions.

• Clone this project git clone [email protected]:NTUDSNLab/DFQ-BFS.git

• Hardware:

  • CPU x86_64 (Test on Intel(R) Core(TM) i9-10900K CPU @ 3.70GHz)
  • NVIDIA GPU (arch>=86) with device memory >= 12GB.(Support NVIDIA RTX3080(sm_86). Note that we mainly evaluate our experience on RTX3090. The execution time could be different with different devices.

• OS & Compler:

  • Ubuntu 18.04
  • CUDA = 11.6
  • nvcc = 11.6

• Important Files/Directories

  • data/: contains all datasets that we want to compare with.
  • bin/: contains all binaries from different designs (including baseline with name started with 1_) that we want to compare with. Note that all the source codes of each binary can be found in the design/ directory.
  • design/: contains all the source codes organized by tree structure, each sub-directory name describing the design choice.
  • plot.py: The python script that traversal all the datasets in data/ with all binaries in bin/, and also plot their runtime speedup with a histogram.

  noting that the first two lines # Nodes: #node Edges: #edge # FromNodeId ToNodeId are necessary!!! the following is the example:

    ```
    # Nodes: 239 Edges: 502
    # FromNodeId    ToNodeId
    0       1
    0       2
    0       3
    ...
    ```
  

2. Environment Setup

1) Pick up the implementation you want in design/ and compile it

cd design/implementation/you/want/
nvcc -O3 --compiler-options -Wall -Xptxas -v bfs.cu -o bfs

2) copy it into design/bin

cd $dir_with_desired_feature
cp bfs design/bin

3) unzip dataset under 'data/' or download it from SNAP

tar xvf data.tar

4) run the python script

python plot.py

5) check the result png file

How to Cite This Work

Thanks for your citation

@inproceedings{hsieh2023decentralized,
  title={A Decentralized Frontier Queue for Improving Scalability of Breadth-First-Search on GPUs},
  author={Hsieh, Chou-Ying and Cheng, Po-Hsiu and Chang, Chia-Ming and Kuo, Sy-Yen},
  booktitle={2023 Design, Automation \& Test in Europe Conference \& Exhibition (DATE)},
  pages={1--6},
  year={2023},
  organization={IEEE}
}