Closed johnlees closed 3 years ago
With large DBs (>10k samples) the BGMM and DBSCAN models become very slow to run the assignment step. It would make sense to use a shm object for the dist matrix, and run parallel blocks of assignment.
shm
With large DBs (>10k samples) the BGMM and DBSCAN models become very slow to run the assignment step. It would make sense to use a
shm
object for the dist matrix, and run parallel blocks of assignment.