Difference between Direct-mapping, Associative Mapping & Set-Associative Mapping

Last Updated : 26 Sep, 2024

Cache memory stores frequently accessed data for easy retrieval. When Read request is received from the CPU, the contents of block of memory words containing the location which is specified are transferred into the cache. The cache memory can store a number of blocks. The mapping technique is required to bring the data of main memory blocks into the cache block. This article discusses about the three mapping techniques and its difference.

What is Cache ?

The small section of SRAM memory, added between main memory and processor(CPU) to speed up the process of execution, is known as cache memory. It includes a small amount of SRAM & more amount of DRAM. It is a high speed & expensive memory.

Cache hit ratio: It measures how effectively the cache fulfills the request for getting content.

Cache hit ratio = No of cache hits/ (No of cache hits + No. of cache Miss)

If data has been found in the cache, it is a cache hit else a cache miss.

Prerequisite - Cache mapping Types - Direct-mapping, Associative Mapping & Set-Associative Mapping

What is Cache Mapping?

The process /technique of bringing data of main memory blocks into the cache block is known as Cache Mapping.
The mapping techniques can be classified as :

Direct Mapping
Associative
Set-Associative

What is Direct Mapping?

Direct Mapping:

Each block from main memory has only one possible place in the cache organization in this technique.
For example : every block i of the main memory can be mapped to block j of the cache using the formula :

j = i modulo m Where : i = main memory block number j = cache block number m = number of blocks in the cache

The address here is divided into 3 fields :

i)Tag

ii)Block

iii)Word.

To map the memory address to cache: The BLOCK field of the address is used to access the cache's BLOCK. Then, the tag bits in the address is compared with the tag of the block. For a match, a cache hit occurs as the required word is found in the cache. Otherwise, a cache miss occurs and the required word has to be brought into the cache from the Main Memory. The word is now stored in the cache together with the new tag (old tag is replaced).

Example:

If we have a fully associative mapped cache of 8 KB size with block size = 128 bytes and say, the size of main memory is = 64 KB. (Assuming word size = 1 byte) Then :

Number of bits for the physical address = 16 bits (as memory size = 64 KB = 2⁶ × 2¹⁰ = 2¹⁶⁾
Number of bits for WORD = 7 bits (as block size = 128 bytes = 2⁷)
No of Index bits = 13 bits (as cache size = 8 KB = 2³ × 2¹⁰ = 2¹³⁾
No of BLOCK bits = Number of Index bits- Number of bits for WORD = 13 – 7 = 6bits

OR
(No of cache Blocks = Cache size/block size = 8 KB / 128 Bytes = 8×1024 Bytes/128 Bytes = 2⁶blocks → 6bits)
No of TAG bits = Number of bits for the physical address — Number of bits in Index = 16-13 = 3 bits

Advantages of Direct-mapping

It requires very less hardware complexity.
It is a very simple method of implementation.
Access time is very low.

Disadvantages of Direct-mapping

Cache performance is unpredictable in direct mapping.
Handling of spatial locality is poor.
Use of cache space is inefficient.
Conflict misses are high.

What is Associative Mapping?

In the associate mapping of the main memory block can be done with any of the cache block. This technique is called as fully associative cache mapping. The memory address has only 2 fields

i)word

ii)tag.

Example:

If we have a fully associative mapped cache of 8 KB size with block size = 128 bytes and say, the size of main memory is = 64 KB. Then:

Number of bits for the physical address = 16 bits (as memory size = 64 KB = 2⁶ × 2¹⁰ = 2¹⁶⁾
Number of bits in block offset = 7 bits (as block size = 128 bytes = 2⁷)
No of tag bits = Number of bits for the physical address - Number of bits in block offset = 16-7 = 9 bits
No of cache Blocks = Cache size/block size = 8 KB / 128 Bytes = 8×1024 Bytes/128 Bytes = 2⁶blocks.

Advantages of Associative Mapping &

It is the fastest and most flexible technique.
It solves the conflict miss problem

Disadvantages of Associative Mapping &

The comparison time for search a block is very high.
It is expensive.

What is Set-Associative Mapping?

It is the combination of advantages of both direct & associative mapping.
Here, the cache consists of a number sets, each of which consists of a number of blocks. The relationships are :

n = w * L i = j modulo w where i : cache set number j : main memory block number n : number of blocks in the cache w : number of sets L : number of lines in each set

This is referred to as L-way set-associative mapping. Block Bj can be translated into any of the blocks in set j using this mapping.

To map the memory address to cache: Using set field in the memory address, we access the particular set of the cache. Then, the tag bits in the address are compared with the tag of all L blocks within that set. For a match, a cache hit occur as the required word is found in the cache. Otherwise, a cache miss occurs and the required word has to be brought in the cache from the Main Memory. According to the replacement policy used, a replacement is done if the cache is full.

Example:

If we have a fully associative mapped cache of 8 KB size with block size = 128 bytes and say, the size of main memory is = 64 KB, and we have “2-way” set-associative mapping (Assume each word has 8 bits). Then :

Number of bits for the physical address = 16 bits (as memory size = 64 KB = 2⁶ * 2¹⁰ = 2¹⁶⁾
No of cache Blocks = Cache size/block size = 8 KB / 128 Bytes = 8×1024 Bytes/128 Bytes = 2⁶ cache blocks.
No of Main Memory Blocks = MM size/block size = 64 KB / 128 Bytes = 64×1024 Bytes/128 Bytes = 2⁹ MM blocks.
No of sets of size 2 = No of Cache Blocks/ L = 2⁶/2 = 2⁵cache sets.(L = 2 as it is 2-way set associative mapping)

Advantages of Set-Associative Mapping

It has highest hit rate.
Conflict Misses are very few.

Disadvantages of Set-Associative Mapping

It is very expensive.
It is slower than direct mapping.