When processing joins, remember that they are always executed in the SAP HANA database. There is no SQL implementation or ABAP implementation. For analysis in the Query Monitor, this means that you cannot toggle between execution in SAP HANA and execution in SQL (the No optimized Operations in SAP HANA/BWA option) if there are no joins involved.
You should also note that the data of every single InfoProvider must be aggregated before the join. You cannot prevent data aggregation before the join. If the cardinality of individual joins does not correspond to the data, this can result in key figure values being changed during navigation.