Exploiting the Memory Hierarchy

1. Exploiting the Memory Hierarchy Chapter 7 P&H

2. Introduction • Desire is to have unlimited amounts of fast memory • Can exploit locality of reference to appear to have lots of fast memory • Two types of locality • Temporal Locality • Spatial Locality

3. Memory Hierarchy

4. Memory Hierarchy

5. Memory Hierarchy • data only copied between two levels at a time • minimum unit is called a block • hit rate is fraction of memory accesses found in upper level • miss rate = (1 – hit rate) • hit time Time to access upper level • miss penality time to replace block in upper level with corresponding block from lower level

6. Cache Basics • Caches first appear in early 1960’s • following cache has single word sized block • How do we know if the requested item is in the cache? • If it is where isit?

7. Direct Mapped Cache • cache address = (block address) mod (No blocks in cache) No blocks in cache should be a power of two

8. A d d r e s s ( s h o w i n g b i t p o s i t i o n s ) 3 1 3 0 1 3 1 2 1 1 2 1 0 B y t e o f f s e t 1 0 2 0 H i t D a t a T a g I n d e x I n d e x V a l i d T a g D a t a 0 1 2 1 0 2 1 1 0 2 2 1 0 2 3 2 0 3 2 Tagged Storage

9. Example • How many bits in total are required for a direct mapped cache with 64 KB of Data and One word blocks, assuming a 32 bit address?

10. Handling Misses • No Modifications required to our pipelined data-path to handle hits • On a miss a basic approach is to stall the entire data-path until the value is fetched • Separate controller used for the caches

11. Example – Miss on Instruction Cache