The result cache does keep query results readily available just as they were requested.
If the query performed aggregations, then the data is aggregated.
I am pretty sure that unloading any of the required tables/columns for the query would lead to cache invalidation.
But sure enough: when many users want to use the very same result of this query and actually send their query in a compatible way that allows for cache-reusage then they can benefit from the already computed result set.
In fact: this is what the result cache is all about.
Concerning the OLAP cache comparison: BW on HANA (and BW in general) does a lot more than just sending SQL queries to the underlying database. The OLAP layer performs major computations.
The OLAP cache actually stores these computation results, which makes the OLAP cache the fastest caching option for BW on HANA. This will even skip the database access alltogether.
- Lars