Performance Considerations

To get the best overall performance of the DBSCAN algorithm, do the following:

  • If input data is homogeneous, provide the input data and store results in homogeneous numeric tables of the same type as specified in the algorithmFPType class template parameter.

  • If input data is non-homogeneous, use AOS layout rather than SOA layout.

  • For the output assignments table, use a homogeneous numeric table of the int type.

DBSCAN may show different performance depending on memorySavingMode flag:

  • When flag is set to false, all neighborhoods can be computed and stored together, which may give better parallelism utilization and, therefore, better performance;

  • When flag is set to true, neighborhood computation is done during the algorithm execution. With rather small amount of simultaneous queries, you get worse parallelism utilization and, therefore, worse performance.

When total size of all neighborhoods is larger than the amount of memory available, usage of memorySavingMode=true flag can be the only option.

Optimization Notice: 
For more complete information about compiler optimizations, see our Optimization Notice.
Select sticky button color: 
Orange (only for download buttons)