This is the code for Triangle Counting submitted at Graph Challenge 2019.
make
For generating the binary of datasets, we use the converter from:
https://github.com/huyang1988/TC/blob/master/README.md
Some example datasets are provided in the data folder of this repository. For using different repository, provide the path to the dataset.
i.e. for p2p08 dataset, "data/p2p08/"
1. Go to TC/gConv/ directory.
'make gConvu'
2. Go to TC/graph_converter/undirected_csr/ directory.
'make'
Copy gConvu and tuple_to_undirected_csr.bin to a folder.
3. Download the Adjacency MMIO file for datasets from graphchallenge website [https://snap.stanford.edu/data/] to the same folder.
4. ./converter.sh <MMIO_file>
We use jsrun to run multi-gpu version of the code on Summit.
https://www.olcf.ornl.gov/for-users/system-user-guides/summit/summit-user-guide/#running-jobs
Change arguments in Makefile.
- Folder name containing the binary file of dataset.
- Total number of process
- Number of Threads per Block (Minimum 32)
- Number of Blocks
- Number of Buckets for hashing (Limit: 256)
- Block-based (1) or Warp-based(0)
- Degree based workload partition (1) or Index based partition(0)
- Name of the dataset
- Vertex count
- Edge count
- Triangles count
- Max Time
- Min Time
- TEPS rate based upon max time
- Total number of process
Please send me a email if you have any questions: [email protected]