Week 10: Caching, Memory Hierarchy

Week 10: Caching, Memory Hierarchy

Zipf workload??

Zipf Distribution

Zipf Distribution

Exponential distribution. Popularity = Rank^-α, where 1 < α < 2

Zipf and Caching

Zipf and Caching

 

Write through: Changes sent immediately to next level of storage

Write back: Changes stored in cache until cache block is replaced

 

Working set: Set of memory locations that need to be cached for reasonable cache hit rate

Thrashing: When system has too small a cache, the cache dramatically degrades performance.

 

Phase cache behavior

Phase Change Behavior

Phase Change Behavior

 

 

Leave a Reply