On the Configuration form for SecurityInsider, you will see this option:
Computing endpoint information approximately doubles the amount of memory (JavaMaxHeapSize) required for a SecurityInsider agent to run, plus it increases processing time - however, the results are more than worth it.
Without endpoint details, SecurityInsider "only" gives you information from the database and group perspective. In other words, you can see all the users who are in a group or who can access a database, but you can't easily see all the groups a user is in or all the databases a user has access to (unless you look at every group and database document individually).
Endpoint processing will tell you which access level each user has in which databases, and through which group paths they are granted the respective access.
Note that endpoint processing will significantly increase the number of documents in the SecurityInsider database by creating one document PER endpoint (users, servers, unknown entries) PER server on which databases are scanned - so, if you have 10 servers and 5,000 users, you will easily end up with 50,000+ additional documents in the SecurityInsider database.