Next: ...
Up: Memory Hierarchies
Previous: Why Might it Work?
- Block: a
-word cache provides fast storage for
up to B blocks, each consisting of N words
- Tag: Each block held in the cache must be tagged
with its main memory address
- Read Hit: If tag matches address, the
data is supplied to the processor
- Read Miss: no match: a block must be fetched from memory
- Allocate: when a new block is fetched,
it is stored in the cache
- Replacement: to make space for a new block to be allocated in the
cache, it may be necessary to displace an existing block.