Caches The principle that states that if data is used, its neighbor will likely be used soon.

26
Caches The principle that states that if data is used, its neighbor will likely be used soon
  • date post

    21-Dec-2015
  • Category

    Documents

  • view

    226
  • download

    1

Transcript of Caches The principle that states that if data is used, its neighbor will likely be used soon.

Caches

The principle that states that if data is used, its neighbor will likely be used soon

Caches

The principle that states that if data is used, its neighbor will likely be used soon

Spatial Locality

Caches

The time it takes a cache to receive the data from the lower level of memory

Caches

The time it takes a cache to receive the data from the lower level of memory

Miss Penalty

Caches

The principle that states that if data is used, it will likely be used again soon

Caches

The principle that states that if data is used, it will likely be used again soon

Temporal Locality

Caches

The time it takes to find out if an item is in the cache and return the data if it is.

Caches

The time it takes to find out if an item is in the cache and return the data if it is.

Access Time

Caches

A cache configuration that requires multiple tag checks each access.

Caches

A cache configuration that requires multiple tag checks each access.

2+ Associativity

Caches

A cache attribute that takes advantage of spatial locality

Caches

A cache attribute that takes advantage of spatial locality

Large Block Size

Caches

Cache attributes that decrease access time

Caches

Cache attributes that decrease access time

Small, Low Associativity

Caches

Cache attributes that decrease miss penalty

Caches

Cache attributes that decrease miss penalty

Small Block Size, Multi-Level Caches

Caches

Cache attributes that decrease miss rate

Caches

Cache attributes that decrease miss rate

Large Cachesize, Large Blocksize,

High Associativity

Caches

log2 ( CacheSize / (BlockSize * Assoc)) =

Caches

log2 ( CacheSize / (BlockSize * Assoc))

# bits in the index

Caches

log2 ( BlockSize / WordSize ) =

Caches

log2 ( Blocksize / Wordsize ) =

# bits in the Block Offset

Caches

log2 ( WordSize) =

Caches

log2 ( WordSize ) =

# bits in the Byte Offset

Caches

#address bits - log2 ( CacheSize / Associativity ) =

Caches

#address bits - log2 ( CacheSize / Associativity ) =

# bits in the Tag