Question $2$

The key advantage of the $q,$ design is that it eliminates contention for the cache between the instruction fetch$/$decode unit and the execution unit.

$$ DRAM

Unified Cache

external cache

split cache