Simply "make -j" in this folder.
The program accepts two formats of data graphs, i.e., SNAP and BIN. For more descriptions on these formats, please refer to the main folder.
We can run the program without any parameter to output a helping message, which can help you to input the parameters.
There are some parameters you need to specify to run the program.
- -f, the file name of the data graph, whose format should be either SNAP or BIN.
- -d, the direction of the graph. 0 for undirected graph and 1 for directed graph. For subgraph enumeration, wet set it to 0. If 1 is specified, we do not guarantee the correctness of the result.
- -a, the algorithm to run. The available algorithms are listed in the helping message.
- -q, the query to run. The supported queries are listed in the helping message. You can other queries, but take some development efforts.
- -t, the number of threads used.
- -b, the memory buffer limit (in GB). When we run BFS-style of algorithms, we need to ensure the generated partial instances would not exceed the limit of the memory buffer. For DFS-style of algorithms, this parameter is not needed.
- Run the CPU_WCOJ (worst-case optimal join) algorithm, which is also the same as Ullman's approach, on the triangle query using 32 threads
./patternmatchcpu -f ~/datasets/com-dblp.ungraph.txt.bin -d 0 -a 0 -q 0 -t 32
- Run the CPU_BFS (PGX) algorithm on the chordal square query using 32 threads with a memory buffer limit as 64 GB.
./patternmatchcpu -f ~/datasets/com-dblp.ungraph.txt.bin -d 0 -a 2 -q 2 -t 32 -b 64
- Run reusable BFS that is based on CPU_BFS on the similar setting
./patternmatchcpu -f ~/datasets/com-dblp.ungraph.txt.bin -d 0 -a 3 -q 2 -t 32 -b 64
scripts/exp_reuse_cpu.sh shows some example runs.