Hash Join Semi

This is a (simplified) model of a hash join semi. We first scan the build table and hash ll the records, then we scan the probe table, and has the records to the buckets looking for matches. Any matches we find, we mark them as "matched." This allows any duplicate records from the probe table to be essentially ignored...once a record is marked as "matched" it does nothing if we go back and mark it again.
Finally, we scan our hash table and any tables that are marked as "matched" get delivered to the next row source above it.