Sequential GPU

Gunrock GPU Implementation of Subgraph Matching VF2 Algorithm

SSH into the Ubuntu user on AWS EC2 Instance

ssh -i <keyName>.pem <username>@##.##.##.##

Change into the Gunrock directory

cd gunrock

Running the Algorithm

Format

./examples/sm/bin/test_sm_12.2_x86_64 --pattern-graph-type=market --pattern-graph-file=[FILE_PATH] --undirected=1 --pattern-undirected=1 --num-runs=1 --graph-type=market --graph-file=[FILE_PATH]

Replace [FILE_PATH] with the path to the specified dataset file:

1. Use your dataset (must be a Market file .mtx).

2. Use existing available datasets as instructed below:

   Gunrock provides small and large datasets. You can familiarize yourself with them here.

   • For small datasets, simply the name of the .mtx file is needed:

     • FILE_PATH = dataset/small/[FILE_NAME].mtx

   • For large datasets, follow the procedure below:

     cd gunrock/dataset/large
     make recurse
     

     • FILE_PATH = dataset/large/[FILE_NAME]/[FILE_NAME].mtx

For example, when CATEGORY = small and FILE_NAME = tri_sm, the FILE_PATH is:

dataset/small/tri_sm.mtx

Example

./examples/sm/bin/test_sm_12.2_x86_64 --pattern-graph-type=market --pattern-graph-file=dataset/small/tri_sm.mtx --undirected=1 --pattern-undirected=1 --num-runs=1 --graph-type=market --graph-file=dataset/small/tri_sm.mtx

Output

The output should appear immediately. For the example above, check out sample_output.txt.

Proof using GPU Power

This is the proof using GPU power.